Spaces:

Bit-Trading-Company
/

bit-backtest-lab

Running on Zero

App Files Files Community

Bit-Trading-Company commited on about 19 hours ago

Commit

27c0524

verified ·

1 Parent(s): 39c225c

Backtest Lab v1.0.0

Browse files

Files changed (17) hide show

.preview/components.html +540 -0
app.py +412 -536
scripts/seed_store.py +31 -21
src/adapters.py +124 -7
src/catalog.py +521 -0
src/charts.py +161 -0
src/config.py +32 -1
src/extension.py +8 -1
src/metrics.py +9 -2
src/runtime.py +13 -3
src/ui/compare_tab.py +211 -0
src/ui/components.py +359 -0
src/ui/format.py +76 -0
src/ui/theme.py +194 -58
tests/test_adapters.py +1 -1
tests/test_catalog.py +312 -0
tests/test_ui.py +22 -20

.preview/components.html ADDED Viewed

	@@ -0,0 +1,540 @@

+<!doctype html><html><head><meta charset='utf-8'><style>@font-face{font-family:'Styrene A';src:url('file:///Users/bit-trading-company/Desktop/dev/hugging-face/space/assets/fonts/StyreneA-Light.otf') format('opentype');font-weight:300;font-style:normal;font-display:swap}
+@font-face{font-family:'Styrene A';src:url('file:///Users/bit-trading-company/Desktop/dev/hugging-face/space/assets/fonts/StyreneA-Regular.otf') format('opentype');font-weight:400;font-style:normal;font-display:swap}
+@font-face{font-family:'Styrene A';src:url('file:///Users/bit-trading-company/Desktop/dev/hugging-face/space/assets/fonts/StyreneA-Medium.otf') format('opentype');font-weight:500;font-style:normal;font-display:swap}
+@font-face{font-family:'Mac Minecraft';src:url('file:///Users/bit-trading-company/Desktop/dev/hugging-face/space/assets/fonts/MacMinecraft.ttf') format('truetype');font-weight:400;font-style:normal;font-display:swap}
+:root{
+--pure-black:#000000;
+--pure-white:#ffffff;
+--stone-950:#161512;
+--stone-900:#1d1c18;
+--stone-850:#24221d;
+--stone-800:#2c2a24;
+--stone-700:#3d3a32;
+--stone-600:#54503f;
+--stone-500:#6f6a56;
+--stone-400:#918c76;
+--stone-300:#b6b09a;
+--stone-200:#d6d1bf;
+--stone-100:#ece8dc;
+--stone-50:#f5f3ea;
+--off-white:#f7f4ec;
+--accent-moss:#68781e;
+--accent-moss-strong:#7d901f;
+--accent-moss-dim:#4d5817;
+--accent-amber:#af9209;
+--accent-amber-strong:#cfab0a;
+--accent-amber-dim:#7d6a09;
+--mute-red:#8a5a54;
+--mute-orange:#8a6f54;
+--mute-yellow:#8a8154;
+--mute-green:#6e8a54;
+--mute-teal:#54898a;
+--mute-blue:#59656e;
+--mute-indigo:#5c5c8a;
+--mute-violet:#75588a;
+--fin-up:oklch(66% 0.22 149);
+--fin-up-strong:oklch(72% 0.24 149);
+--fin-down:oklch(62% 0.26 24);
+--fin-down-strong:oklch(68% 0.27 24);
+--fin-flat:var(--stone-400);
+--fin-up-cvd:oklch(58% 0.13 240);
+--fin-up-cvd-strong:oklch(64% 0.14 240);
+--fin-down-cvd:oklch(58% 0.13 55);
+--fin-down-cvd-strong:oklch(64% 0.14 55);
+--fin-buy:var(--fin-up);
+--fin-buy-strong:var(--fin-up-strong);
+--fin-sell:var(--fin-down);
+--fin-sell-strong:var(--fin-down-strong);
+--focus-ring:var(--accent-amber);
+--bg-canvas:var(--stone-950);
+--bg-panel:var(--stone-900);
+--bg-raised:var(--stone-850);
+--bg-sunken:var(--pure-black);
+--border-subtle:var(--stone-800);
+--border-default:var(--stone-700);
+--border-strong:var(--stone-500);
+--text-primary:var(--off-white);
+--text-secondary:var(--stone-300);
+--text-tertiary:var(--stone-500);
+--text-disabled:var(--stone-600);
+--text-inverse:var(--stone-950);
+--surface-accent-fg:var(--stone-950);
+--glass-bg:rgba(29,28,24,0.72);
+--glass-border:rgba(247,244,236,0.14);
+--glass-blur:20px;
+}
+[data-theme="dark"]{
+--bg-canvas:var(--stone-950);
+--bg-panel:var(--stone-900);
+--bg-raised:var(--stone-850);
+--bg-sunken:var(--pure-black);
+--border-subtle:var(--stone-800);
+--border-default:var(--stone-700);
+--border-strong:var(--stone-500);
+--text-primary:var(--off-white);
+--text-secondary:var(--stone-300);
+--text-tertiary:var(--stone-500);
+--text-disabled:var(--stone-600);
+--text-inverse:var(--stone-950);
+--surface-accent-fg:var(--stone-950);
+}
+[data-colorblind="true"]{
+--fin-up:var(--fin-up-cvd);
+--fin-up-strong:var(--fin-up-cvd-strong);
+--fin-down:var(--fin-down-cvd);
+--fin-down-strong:var(--fin-down-cvd-strong);
+}
+[data-theme="light"]{
+--bg-canvas:var(--off-white);
+--bg-panel:var(--stone-50);
+--bg-raised:var(--pure-white);
+--bg-sunken:var(--stone-100);
+--border-subtle:var(--stone-200);
+--border-default:var(--stone-300);
+--border-strong:var(--stone-500);
+--text-primary:var(--stone-950);
+--text-secondary:var(--stone-700);
+--text-tertiary:var(--stone-500);
+--text-disabled:var(--stone-300);
+--text-inverse:var(--off-white);
+--surface-accent-fg:var(--stone-950);
+--glass-bg:rgba(245,243,234,0.72);
+--glass-border:rgba(22,21,18,0.1);
+}
+/* Custom themes: copy this block, rename the [data-theme] value, and override only
+the tokens that should change. Every component reads these tokens, never raw colors,
+so a new theme needs no component edits — set data-theme on any ancestor (SidebarNav's
+theme selector does this at the app root) and everything downstream updates. */
+[data-theme="custom"]{
+--bg-canvas:var(--stone-950);
+--bg-panel:var(--stone-900);
+--bg-raised:var(--stone-850);
+--bg-sunken:var(--pure-black);
+--border-subtle:var(--stone-800);
+--border-default:var(--stone-700);
+--border-strong:var(--accent-amber-dim);
+--text-primary:var(--off-white);
+--text-secondary:var(--stone-300);
+--text-tertiary:var(--stone-500);
+--text-disabled:var(--stone-600);
+--text-inverse:var(--stone-950);
+--surface-accent-fg:var(--stone-950);
+}
+:root{
+--font-styrene:'Styrene A',sans-serif;
+--font-system:-apple-system,BlinkMacSystemFont,system-ui,sans-serif;
+--font-mono:'JetBrains Mono',ui-monospace,'SFMono-Regular',Menlo,monospace;
+--font-tiny:'Mac Minecraft',monospace;
+--font-body:var(--font-system);
+--font-heading:var(--font-styrene);
+--weight-heading:400;
+--weight-body:300;
+--weight-styrene-thin:100;
+--weight-styrene-light:300;
+--weight-styrene-regular:400;
+--weight-styrene-medium:500;
+--weight-mono:400;
+--weight-tiny:400;
+--weight-strong:700;
+--text-2xs:8px;
+--text-xs:10px;
+--text-sm:11px;
+--text-base:12px;
+--text-md:16px;
+--text-lg:19px;
+--text-xl:25px;
+--text-2xl:33px;
+--text-3xl:45px;
+--text-4xl:62px;
+--leading-tight:1.1;
+--leading-snug:1.3;
+--leading-normal:1.5;
+--leading-relaxed:1.7;
+--tracking-tight:-0.01em;
+--tracking-normal:0;
+--tracking-wide:0.04em;
+--tracking-wider:0.12em;
+}
+:root{
+--space-0:0px;
+--space-1:4px;
+--space-2:8px;
+--space-3:12px;
+--space-4:16px;
+--space-5:24px;
+--space-6:32px;
+--space-7:48px;
+--space-8:64px;
+--space-9:96px;
+--radius-0:0px;
+--radius-sm:4px;
+--radius-md:8px;
+--radius-pill:999px;
+--border-width:1px;
+--border-width-strong:2px;
+--ease:ease;/* @kind other */
+--duration:1s;/* @kind other */
+--shadow-none:none;
+--shadow-panel:0 0 0 var(--border-width) var(--border-default);
+--z-header:100;/* @kind other */
+--z-menu:200;/* @kind other */
+--z-modal:300;/* @kind other */
+--z-toast:400;/* @kind other */
+}
+/* ================= app shell ================= */
+.gradio-container{
+  max-width:100% !important; padding:0 !important;
+  background:var(--bg-canvas) !important;
+  font-family:var(--font-body); font-weight:var(--weight-body);
+  color:var(--text-primary); font-size:var(--text-base);
+}
+.gradio-container *{ border-radius:var(--radius-sm) !important; }
+footer{ display:none !important; }
+.gap.svelte-vt1mxs, .gradio-container .gap{ gap:var(--space-2) !important; }
+/* ================= typography ================= */
+.bit-h1,.bit-h2,.bit-h3{
+  font-family:var(--font-heading); font-weight:var(--weight-heading);
+  letter-spacing:var(--tracking-wide); text-transform:uppercase;
+  color:var(--text-primary); margin:0;
+}
+.bit-h1{ font-size:var(--text-md); }
+.bit-h2{ font-size:var(--text-md); }
+.bit-h3{ font-size:var(--text-base); }
+.bit-micro{
+  font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); text-transform:uppercase;
+  color:var(--text-tertiary); line-height:1.6;
+}
+.bit-spacer{ flex:1; }
+/* ================= top bar ================= */
+.bit-topbar{
+  display:flex; align-items:center; gap:var(--space-3); flex-wrap:wrap;
+  background:var(--bg-panel);
+  border-bottom:var(--border-width) solid var(--border-default);
+  padding:var(--space-3) var(--space-4);
+  position:sticky; top:0; z-index:var(--z-header);
+}
+.bit-mark{
+  width:20px; height:20px; background:var(--accent-amber);
+  display:inline-block; flex:0 0 20px;
+  box-shadow:inset 0 0 0 3px var(--bg-panel), inset 0 0 0 6px var(--accent-amber);
+}
+.bit-wordmark{
+  font-family:var(--font-heading); font-size:var(--text-md); font-weight:500;
+  letter-spacing:var(--tracking-wide); color:var(--text-primary);
+}
+.bit-slash{ color:var(--text-tertiary); }
+.bit-chip{
+  display:inline-flex; align-items:center; gap:6px; white-space:nowrap;
+  font-family:var(--font-mono); font-size:var(--text-sm);
+  border:var(--border-width) solid var(--border-default);
+  padding:3px var(--space-2); color:var(--text-secondary); background:transparent;
+}
+.bit-chip-ok{ color:var(--accent-moss-strong); border-color:var(--accent-moss-dim); }
+.bit-chip-run{ color:var(--accent-amber-strong); border-color:var(--accent-amber-dim); }
+.bit-chip-warn{ color:var(--fin-down); border-color:var(--fin-down); }
+.bit-chip-accent{ color:var(--stone-950); background:var(--accent-amber);
+  border-color:var(--accent-amber); }
+.bit-chip-ctx{ color:var(--text-tertiary); }
+a.bit-link{ text-decoration:none; }
+a.bit-link:hover{ color:var(--accent-amber-strong);
+  border-color:var(--accent-amber-dim); }
+/* ================= panels ================= */
+.bit-panel{
+  background:var(--bg-panel);
+  border:var(--border-width) solid var(--border-default);
+  padding:var(--space-4); margin-bottom:var(--space-3);
+}
+.bit-panel-head{
+  display:flex; align-items:baseline; gap:var(--space-3);
+  border-bottom:var(--border-width) solid var(--border-subtle);
+  padding-bottom:var(--space-2); margin-bottom:var(--space-3);
+}
+.bit-zone-title{
+  font-family:var(--font-heading); text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); font-size:var(--text-md);
+  color:var(--text-primary); padding:var(--space-3) var(--space-1) var(--space-2);
+  display:flex; align-items:baseline; gap:var(--space-2);
+}
+.bit-zone-title .bit-micro{ margin-left:auto; }
+/* ================= stat band ================= */
+.bit-statband{
+  display:flex; flex-wrap:wrap;
+  border:var(--border-width) solid var(--border-default);
+  background:var(--bg-panel); margin-bottom:var(--space-3);
+}
+.bit-stat{
+  flex:1 1 132px; padding:var(--space-3) var(--space-4);
+  border-right:var(--border-width) solid var(--border-subtle);
+  border-bottom:var(--border-width) solid var(--border-subtle);
+}
+.bit-stat-label{
+  font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); text-transform:uppercase;
+  color:var(--text-tertiary);
+}
+.bit-stat-value{
+  font-family:var(--font-mono); font-size:var(--text-xl);
+  line-height:var(--leading-tight); color:var(--text-primary);
+  margin:3px 0 2px; letter-spacing:var(--tracking-tight);
+}
+.bit-stat-sub{
+  font-family:var(--font-mono); font-size:var(--text-xs);
+  color:var(--text-tertiary);
+}
+.bit-up{ color:var(--fin-up-strong); } .bit-down{ color:var(--fin-down-strong); }
+/* ================= notes ================= */
+.bit-note{
+  font-family:var(--font-mono); font-size:var(--text-sm); line-height:1.5;
+  border-left:var(--border-width-strong) solid var(--accent-amber);
+  background:var(--bg-raised); padding:var(--space-2) var(--space-3);
+  color:var(--text-secondary); margin-bottom:var(--space-2);
+}
+.bit-note-danger{ border-left-color:var(--fin-down); }
+.bit-tag{
+  font-family:var(--font-tiny); font-size:var(--text-2xs);
+  letter-spacing:var(--tracking-wider); border:1px solid var(--border-default);
+  padding:1px 4px; color:var(--text-tertiary); margin-left:4px;
+}
+/* ================= empty state ================= */
+.bit-empty{
+  display:flex; flex-direction:column; align-items:center; justify-content:center;
+  gap:var(--space-3); padding:var(--space-8) var(--space-4); text-align:center;
+  border:var(--border-width) dashed var(--border-default); background:var(--bg-panel);
+}
+.bit-empty-glyph{ font-size:34px; color:var(--accent-amber-dim); line-height:1; }
+.bit-empty-copy{ color:var(--text-secondary); max-width:46ch; font-size:var(--text-base); }
+.bit-kbd-row{ display:flex; gap:var(--space-2); flex-wrap:wrap; justify-content:center; }
+.bit-kbd{
+  font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); color:var(--text-tertiary);
+  border:var(--border-width) solid var(--border-default); padding:3px 7px;
+}
+/* ================= run manager ================= */
+.bit-run-card{
+  border:var(--border-width) solid var(--border-subtle);
+  padding:var(--space-2); margin-bottom:var(--space-1); background:var(--bg-panel);
+}
+.bit-run-card-sel{ border-color:var(--accent-amber-dim); }
+.bit-run-top{ display:flex; align-items:center; gap:var(--space-2); }
+.bit-run-name{ font-size:var(--text-sm); color:var(--text-primary);
+  overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
+.bit-run-mark{ margin-left:auto; color:var(--accent-amber-strong); font-size:11px; }
+.bit-run-sharpe{ font-family:var(--font-mono); font-size:var(--text-base);
+  display:flex; gap:var(--space-2); }
+.bit-run-ret{ margin-left:auto; color:var(--text-tertiary); }
+.bit-gloss{ margin-bottom:var(--space-2); }
+.bit-gloss-def{ font-size:var(--text-sm); color:var(--text-secondary); line-height:1.45; }
+/* ================= KPI row ================= */
+.bit-kpi-row{ display:flex; flex-wrap:wrap; gap:var(--space-4); }
+.bit-kpi{ min-width:78px; }
+.bit-kpi-value{ font-family:var(--font-mono); font-size:var(--text-lg);
+  line-height:1.1; color:var(--text-primary); }
+/* ================= podium ================= */
+.bit-podium-row{ display:flex; gap:var(--space-3); flex-wrap:wrap;
+  margin-bottom:var(--space-3); }
+.bit-podium{
+  flex:1 1 200px; border:var(--border-width) solid var(--border-default);
+  background:var(--bg-panel); padding:var(--space-3);
+  border-top:var(--border-width-strong) solid var(--border-strong);
+}
+.bit-podium-1{ border-top-color:var(--accent-amber); }
+.bit-podium-2{ border-top-color:var(--accent-moss-strong); }
+.bit-podium-3{ border-top-color:var(--mute-teal); }
+.bit-podium-rank{ font-family:var(--font-tiny); font-size:var(--text-xs);
+  color:var(--text-tertiary); letter-spacing:var(--tracking-wider); }
+.bit-podium-name{ font-size:var(--text-md); color:var(--text-primary);
+  font-family:var(--font-heading); letter-spacing:var(--tracking-tight); }
+.bit-podium-value{ font-family:var(--font-mono); font-size:var(--text-2xl);
+  line-height:1.1; margin-top:var(--space-2); display:flex;
+  align-items:baseline; gap:var(--space-2); }
+.bit-podium-unit{ font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); color:var(--text-tertiary); }
+/* ================= signal aggregator ================= */
+.bit-sig-grid{ display:flex; gap:var(--space-4); flex-wrap:wrap; align-items:flex-start; }
+.bit-sig-list{ flex:1 1 340px; min-width:0; }
+.bit-sig-row{
+  display:flex; align-items:center; gap:var(--space-3);
+  border:var(--border-width) solid var(--border-subtle);
+  padding:5px var(--space-2); margin-bottom:3px;
+  font-family:var(--font-mono); font-size:var(--text-sm);
+}
+.bit-sig-name{ flex:1 1 auto; color:var(--text-secondary);
+  overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
+.bit-sig-dir{ width:74px; text-align:right; }
+.bit-sig-edge{ width:64px; text-align:right; }
+.bit-sig-acc,.bit-sig-w{ width:64px; text-align:right; color:var(--text-tertiary); }
+.bit-consensus{
+  flex:0 0 210px; border:var(--border-width) solid var(--accent-moss-dim);
+  background:var(--bg-raised); padding:var(--space-3);
+}
+.bit-consensus-dir{ font-family:var(--font-heading); font-size:var(--text-2xl);
+  line-height:1.1; margin:var(--space-1) 0; letter-spacing:var(--tracking-tight); }
+.bit-conf-track{ height:4px; background:var(--bg-sunken); margin:var(--space-2) 0 4px; }
+.bit-conf-fill{ height:100%; }
+/* ================= gradio overrides ================= */
+.gradio-container .tabs > .tab-nav{
+  border-bottom:var(--border-width) solid var(--border-default) !important;
+  background:transparent !important; gap:0 !important; flex-wrap:wrap;
+}
+.gradio-container .tabs > .tab-nav > button{
+  font-family:var(--font-heading) !important; text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); font-size:var(--text-base) !important;
+  color:var(--text-tertiary) !important; background:transparent !important;
+  border:none !important; border-bottom:2px solid transparent !important;
+  padding:var(--space-2) var(--space-4) !important;
+}
+.gradio-container .tabs > .tab-nav > button:hover{ color:var(--text-secondary) !important; }
+.gradio-container .tabs > .tab-nav > button.selected{
+  color:var(--text-primary) !important;
+  border-bottom-color:var(--accent-amber) !important;
+}
+.gradio-container .form, .gradio-container .block{
+  background:transparent !important; border:none !important;
+}
+/* Field captions only -- scoped so it never swallows option text. */
+.block > label > span, .block > .form > label > span{
+  font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
+  letter-spacing:var(--tracking-wider) !important; text-transform:uppercase;
+  color:var(--text-tertiary) !important;
+}
+input, select, textarea{
+  font-family:var(--font-mono) !important; font-size:var(--text-base) !important;
+  background:var(--bg-sunken) !important; color:var(--text-primary) !important;
+  border:var(--border-width) solid var(--border-default) !important;
+}
+input:focus, select:focus, textarea:focus{
+  outline:1px solid var(--focus-ring) !important; outline-offset:-1px;
+}
+/* Radio / checkbox options read as segmented chips. */
+.gradio-container fieldset label,
+.gradio-container .wrap label:has(input[type="radio"]),
+.gradio-container .wrap label:has(input[type="checkbox"]){
+  background:transparent !important;
+  border:var(--border-width) solid var(--border-default) !important;
+  color:var(--text-secondary) !important;
+  padding:4px 10px !important; margin:2px !important;
+}
+.gradio-container fieldset label span,
+.gradio-container .wrap label:has(input[type="radio"]) span,
+.gradio-container .wrap label:has(input[type="checkbox"]) span{
+  font-family:var(--font-mono) !important; font-size:var(--text-base) !important;
+  letter-spacing:var(--tracking-normal) !important; text-transform:none !important;
+  color:var(--text-secondary) !important; opacity:1 !important;
+}
+.gradio-container fieldset label.selected,
+.gradio-container fieldset label:has(input:checked),
+.gradio-container .wrap label:has(input[type="radio"]:checked){
+  background:var(--accent-amber) !important; border-color:var(--accent-amber) !important;
+}
+.gradio-container fieldset label.selected span,
+.gradio-container fieldset label:has(input:checked) span,
+.gradio-container .wrap label:has(input[type="radio"]:checked) span{
+  color:var(--stone-950) !important; font-weight:500 !important;
+}
+.gradio-container input[type="radio"], .gradio-container input[type="checkbox"]{
+  accent-color:var(--accent-amber);
+}
+/* Accordions read as the design's numbered sections. */
+.gradio-container .label-wrap, .bit-accordion > .label-wrap{
+  font-family:var(--font-heading) !important; text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); font-size:var(--text-base) !important;
+  color:var(--text-primary) !important;
+  border-bottom:var(--border-width) solid var(--border-subtle) !important;
+  padding:var(--space-2) 0 !important;
+}
+.bit-accordion{
+  border:none !important;
+  border-top:var(--border-width) solid var(--border-subtle) !important;
+  padding:0 !important; margin-bottom:var(--space-1) !important;
+}
+/* Buttons */
+.bit-run-btn{
+  background:var(--accent-amber) !important; color:var(--stone-950) !important;
+  font-family:var(--font-heading) !important; text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); border:none !important; font-weight:500 !important;
+}
+.bit-run-btn:hover{ background:var(--accent-amber-strong) !important; }
+.bit-ghost-btn{
+  background:transparent !important; color:var(--text-secondary) !important;
+  border:var(--border-width) solid var(--border-default) !important;
+  font-family:var(--font-mono) !important; font-size:var(--text-sm) !important;
+}
+.bit-ghost-btn:hover{ border-color:var(--border-strong) !important;
+  color:var(--text-primary) !important; }
+/* Tables */
+.bit-table table{ font-family:var(--font-mono) !important; font-size:var(--text-sm) !important; }
+.bit-table thead th{
+  font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
+  letter-spacing:var(--tracking-wider); text-transform:uppercase;
+  color:var(--text-tertiary) !important; background:var(--bg-raised) !important;
+  border-bottom:var(--border-width) solid var(--border-default) !important;
+}
+.bit-table tbody td{ border-color:var(--border-subtle) !important; }
+/* Footer */
+.bit-footer{
+  position:sticky; bottom:0; z-index:var(--z-header);
+  display:flex; justify-content:space-between; gap:var(--space-4);
+  background:var(--bg-panel); border-top:var(--border-width) solid var(--border-default);
+  padding:var(--space-2) var(--space-4);
+  font-family:var(--font-mono); font-size:var(--text-sm); color:var(--text-tertiary);
+}
+.bit-footer-right{ white-space:nowrap; }
+@media (max-width: 900px){
+  .bit-statband{ flex-direction:column; }
+  .bit-stat{ border-right:none; }
+  .bit-sig-grid{ flex-direction:column; }
+  .bit-consensus{ flex:1 1 auto; width:100%; }
+}
+</style></head><body class='gradio-container' style='margin:0'>
+<div class="bit-topbar">
+  <span class="bit-mark"></span>
+  <span class="bit-wordmark">BIT</span>
+  <span class="bit-slash">/</span>
+  <span class="bit-h1">Backtest Lab</span>
+  <span class="bit-chip bit-chip-ctx">BTC-USD · 1D · 2023-08-16 → 2026-08-15 · WALK-FORWARD</span>
+  <span class="bit-spacer"></span>
+  <span class="bit-micro" style="margin:0 4px">0.1s</span><span class="bit-chip bit-chip-ok">RUN 7bee8d1b COMPLETE</span>
+  <a class="bit-chip bit-link" target="_blank" rel="noopener"
+     href="https://huggingface.co/datasets/The-Bit-Trading-Company/bit-signal-store">SIGNAL STORE ↗</a>
+  <a class="bit-chip bit-link" target="_blank" rel="noopener"
+     href="https://huggingface.co/spaces/Bit-Trading-Company/bit-backtest-lab">SPACE ↗</a>
+</div><div style="padding:16px"><div class="bit-zone-title">Leaderboard podium</div><div class="bit-podium-row"><div class="bit-podium bit-podium-1"><div class="bit-podium-rank">#1</div><div class="bit-podium-name">Bollinger Breakout</div><div class="bit-micro">NVDA · 1h</div><div class="bit-podium-value" style="color:var(--accent-moss-strong)">3.33<span class="bit-podium-unit">OOS SHARPE</span></div><div class="bit-micro">ret +22.3% · dd -19.3% · 49 trades</div></div><div class="bit-podium bit-podium-2"><div class="bit-podium-rank">#2</div><div class="bit-podium-name">Chronos Forecast Follower</div><div class="bit-micro">BTC-USD · 1d · Chronos-Bolt Small</div><div class="bit-podium-value" style="color:var(--accent-moss-strong)">1.43<span class="bit-podium-unit">OOS SHARPE</span></div><div class="bit-micro">ret +122.3% · dd -23.1% · 35 trades</div></div><div class="bit-podium bit-podium-3"><div class="bit-podium-rank">#3</div><div class="bit-podium-name">MACD Momentum</div><div class="bit-micro">NVDA · 1h</div><div class="bit-podium-value" style="color:var(--accent-moss-strong)">1.26<span class="bit-podium-unit">OOS SHARPE</span></div><div class="bit-micro">ret +21.7% · dd -29.7% · 93 trades</div></div></div><div class="bit-zone-title">Signal aggregator</div><div class="bit-panel"><div class="bit-panel-head"><span class="bit-h3">Signal aggregator</span><span class="bit-micro">8 MODELS</span></div><div class="bit-sig-grid"><div class="bit-sig-list"><div class="bit-sig-row"><span class="bit-sig-name">Chronos-Bolt Small</span><span class="bit-sig-dir" style="color:var(--fin-up-strong)">▲ LONG</span><span class="bit-sig-edge" style="color:var(--fin-up-strong)">+0.27%</span><span class="bit-sig-acc">acc 53%</span><span class="bit-sig-w">w0.07</span></div><div class="bit-sig-row"><span class="bit-sig-name">Chronos-Bolt Base</span><span class="bit-sig-dir" style="color:var(--fin-down-strong)">▼ SHORT</span><span class="bit-sig-edge" style="color:var(--fin-down-strong)">-1.15%</span><span class="bit-sig-acc">acc 51%</span><span class="bit-sig-w">w0.03</span></div><div class="bit-sig-row"><span class="bit-sig-name">Chronos-Bolt Mini</span><span class="bit-sig-dir" style="color:var(--fin-down-strong)">▼ SHORT</span><span class="bit-sig-edge" style="color:var(--fin-down-strong)">-1.23%</span><span class="bit-sig-acc">acc 51%</span><span class="bit-sig-w">w0.03</span></div><div class="bit-sig-row"><span class="bit-sig-name">Baseline · Drift <span class="bit-tag">BASELINE</span></span><span class="bit-sig-dir" style="color:var(--fin-down-strong)">▼ SHORT</span><span class="bit-sig-edge" style="color:var(--fin-down-strong)">-0.06%</span><span class="bit-sig-acc">acc 51%</span><span class="bit-sig-w">w0.03</span></div><div class="bit-sig-row"><span class="bit-sig-name">Baseline · Seasonal naive <span class="bit-tag">BASELINE</span></span><span class="bit-sig-dir" style="color:var(--fin-down-strong)">▼ SHORT</span><span class="bit-sig-edge" style="color:var(--fin-down-strong)">-8.33%</span><span class="bit-sig-acc">acc 51%</span><span class="bit-sig-w">w0.03</span></div><div class="bit-sig-row"><span class="bit-sig-name">Chronos-Bolt Tiny</span><span class="bit-sig-dir" style="color:var(--fin-up-strong)">▲ LONG</span><span class="bit-sig-edge" style="color:var(--fin-up-strong)">+0.60%</span><span class="bit-sig-acc">acc 50%</span><span class="bit-sig-w">w0.01</span></div><div class="bit-sig-row"><span class="bit-sig-name">Baseline · Random walk <span class="bit-tag">BASELINE</span></span><span class="bit-sig-dir" style="color:var(--text-tertiary)">■ FLAT</span><span class="bit-sig-edge" style="color:var(--text-tertiary)">+0.00%</span><span class="bit-sig-acc">acc —</span><span class="bit-sig-w">w0.00</span></div><div class="bit-sig-row"><span class="bit-sig-name">Chronos T5 Small</span><span class="bit-sig-dir" style="color:var(--fin-up-strong)">▲ LONG</span><span class="bit-sig-edge" style="color:var(--fin-up-strong)">+0.95%</span><span class="bit-sig-acc">acc 48%</span><span class="bit-sig-w">w0.00</span></div></div><div class="bit-consensus"><div class="bit-micro">BTC-USD · 1d</div><div class="bit-consensus-dir" style="color:var(--fin-down-strong)">SHORT ▼</div><div class="bit-micro">WEIGHTED EDGE -1.44%</div><div class="bit-conf-track"><div class="bit-conf-fill" style="width:50%;background:var(--fin-down-strong)"></div></div><div class="bit-micro">4/8 MODELS AGREE</div></div></div><div class="bit-micro" style="color:var(--text-tertiary)">weight = realised directional accuracy above a coin flip, on this asset and timeframe</div></div><div class="bit-note bit-note-danger">Learned models call direction correctly 50.2% of the time; naive baselines manage 51.1%. <b>The models are not beating the baselines on direction.</b> Read the leaderboard with that in mind.</div><div class="bit-zone-title">Coverage</div><div class="bit-note bit-note-danger">The signal store has no coverage yet.</div></div>
+<div class="bit-footer">
+  <span>Backtested results are hypothetical, derived from historical data, and are not indicative of future results. Nothing here is investment advice. The Bit Trading Company is not a licensed investment adviser.</span>
+  <span class="bit-footer-right">BITTRADING BACKTEST LAB v1.1.0</span>
+</div></body></html>

app.py CHANGED Viewed

@@ -1,32 +1,38 @@
 """Bit Trading Company — Backtest Lab.
-Gradio Blocks implementation of the Backtest Lab design: Strategy Builder on
-the left, tabbed Results Canvas in the centre, Run Manager on the right, with
-the disclaimer pinned to the footer.
 The app reads exclusively from the cached signal store. The only path that can
 reach an external provider is the batch refresh in `scripts/seed_store.py`, and
-the only path that runs inference is the ZeroGPU extension flow in
-`src/extension.py`, which spends the signed-in user's own quota.
 """
 from __future__ import annotations
 import logging
 import os
-from dataclasses import asdict
 import gradio as gr
 import pandas as pd
-from src import charts, comparisons, config, runtime, strategies
 from src.runtime import RunError, RunRecord, RunRequest
 from src.ui import theme
 logging.basicConfig(level=logging.INFO, format="%(asctime)s %(name)s %(message)s")
 log = logging.getLogger("bit.app")
 MAX_COMPARE = 6
 GLOSSARY = [
     ("SHARPE", "Annualized mean excess return divided by return volatility. "
                "Above 1 is good; above 3 usually means a bug."),
@@ -37,120 +43,30 @@ GLOSSARY = [
     ("MAE / MFE", "Worst and best unrealized excursion while the trade was open."),
     ("WALK-FORWARD", "Train on a rolling window, test on the next unseen window, repeat."),
     ("OOS", "Out of sample: data the parameters never saw during fitting."),
 ]
 # --------------------------------------------------------------------------
-# Formatting helpers
 # --------------------------------------------------------------------------
-def pct(v, digits=1, signed=True) -> str:
-    if v is None or pd.isna(v):
-        return "—"
-    return f"{v * 100:+.{digits}f}%" if signed else f"{v * 100:.{digits}f}%"
-def num(v, digits=2) -> str:
-    if v is None or pd.isna(v):
-        return "—"
-    return f"{v:.{digits}f}"
-def money(v) -> str:
-    if v is None or pd.isna(v):
-        return "—"
-    return f"${v:,.0f}"
-def _cls(v) -> str:
-    if v is None or pd.isna(v) or v == 0:
-        return ""
-    return "bit-up" if v > 0 else "bit-down"
-def _arrow(v) -> str:
-    if v is None or pd.isna(v) or v == 0:
-        return ""
-    return " ▲" if v > 0 else " ▼"
-def seg(metrics, fmt, *args, **kwargs) -> str:
-    """Format a segment metric, or an em dash when that segment has no bars.
-    A segment with no data must never render as 0.00 -- "the out-of-sample
-    Sharpe is zero" and "there is no out-of-sample period" are different claims,
-    and only one of them is true here.
-    """
-    if metrics is None or metrics.bars == 0:
-        return "—"
-    return fmt(*args, **kwargs)
-def stat_band_html(rec: RunRecord | None) -> str:
-    """The stat band. Every stat carries its IS and OOS split, per the design."""
-    if rec is None:
-        return ""
-    r = rec.result
-    a, i, o = r.metrics_all, r.metrics_is, r.metrics_oos
-    def isoos(fmt, ikey, okey, *fargs):
-        iv = seg(i, fmt, getattr(i, ikey), *fargs)
-        ov = seg(o, fmt, getattr(o, okey), *fargs)
-        return f"IS {iv} · OOS {ov}"
-    bench_gap = a.total_return - (
-        float(r.benchmark_equity.iloc[-1] / r.benchmark_equity.iloc[0] - 1.0)
-        if len(r.benchmark_equity) else 0.0
-    )
-    cells = [
-        ("Total return", f"{pct(a.total_return)}{_arrow(a.total_return)}", _cls(a.total_return),
-         isoos(pct, "total_return", "total_return"),
-         "Cumulative return of the strategy equity curve, costs included."),
-        ("CAGR", pct(a.cagr), _cls(a.cagr), isoos(pct, "cagr", "cagr"),
-         "Compound annual growth rate implied by the equity curve."),
-        ("Sharpe", num(a.sharpe), _cls(a.sharpe),
-         isoos(num, "sharpe", "sharpe"), GLOSSARY[0][1]),
-        ("Sortino", num(a.sortino), _cls(a.sortino),
-         isoos(num, "sortino", "sortino"), GLOSSARY[1][1]),
-        ("Max drawdown", pct(a.max_drawdown), "bit-down",
-         isoos(pct, "max_drawdown", "max_drawdown"), GLOSSARY[2][1]),
-        ("Win rate", pct(a.win_rate, 0, signed=False), "",
-         f"IS {seg(i, pct, i.win_rate, 0, False)} · OOS {seg(o, pct, o.win_rate, 0, False)}",
-         "Share of closed trades with positive net P&L."),
-        ("Profit factor", num(a.profit_factor), _cls(a.profit_factor - 1.0),
-         isoos(num, "profit_factor", "profit_factor"), GLOSSARY[3][1]),
-        ("Trades", f"{a.trade_count}", "",
-         f"IS {seg(i, str, i.trade_count)} · OOS {seg(o, str, o.trade_count)}",
-         "Closed round-trip trades in the period."),
-        ("Exposure", pct(a.exposure, 0, signed=False), "",
-         f"IS {seg(i, pct, i.exposure, 0, False)} · OOS {seg(o, pct, o.exposure, 0, False)}",
-         "Fraction of bars holding a position."),
-        ("vs buy & hold", f"{pct(bench_gap)}", _cls(bench_gap),
-         f"costs paid {money(r.costs_paid)}",
-         "Strategy return minus buy-and-hold return over the same window."),
-    ]
-    html = ['<div class="bit-statband">']
-    for label, value, cls, sub, tip in cells:
-        html.append(
-            f'<div class="bit-stat" title="{tip}">'
-            f'<div class="bit-stat-label">{label}</div>'
-            f'<div class="bit-stat-value {cls}">{value}</div>'
-            f'<div class="bit-stat-sub">{sub}</div></div>'
-        )
-    html.append("</div>")
-    for note in getattr(r.plan, "notes", []):
-        html.append(f'<div class="bit-note bit-note-danger">{note}</div>')
-    if rec.result.metrics_holdout is not None:
-        h = rec.result.metrics_holdout
-        html.append(
-            f'<div class="bit-note">LOCKED HOLDOUT · return {pct(h.total_return)} · '
-            f'Sharpe {num(h.sharpe)} · {h.bars} bars never used for any parameter choice.</div>'
-        )
-    return "".join(html)
-def context_chip(req: RunRequest) -> str:
     mode = {"walk_forward": "WALK-FORWARD", "holdout": "HOLDOUT",
             "split": "SPLIT", "none": "NO SPLIT"}.get(req.validation_mode, "")
     try:
@@ -158,53 +74,7 @@ def context_chip(req: RunRequest) -> str:
         span = f"{s.date()} → {e.date()}"
     except Exception:
         span = req.date_range
-    return (f'<span class="bit-chip">{req.asset} · {req.timeframe.upper()} · '
-            f'{span} · {mode}</span>')
-def top_bar_html(status_text="NO RUN LOADED", status_cls="bit-chip", chip="") -> str:
-    return f"""
-<div class="bit-topbar">
-  <span class="bit-mark"></span>
-  <span class="bit-h1">BIT</span>
-  <span style="color:var(--text-tertiary)">/</span>
-  <span class="bit-h1">Backtest Lab</span>
-  {chip}
-  <span style="flex:1"></span>
-  <span class="{status_cls}">{status_text}</span>
-  <a class="bit-chip" href="https://huggingface.co/datasets/{config.STORE_REPO}"
-     target="_blank" rel="noopener">SIGNAL STORE ↗</a>
-</div>"""
-FOOTER_HTML = f"""
-<div class="bit-footer">
-  <span>{config.DISCLAIMER}</span>
-  <span style="white-space:nowrap">BITTRADING BACKTEST LAB v1.0.0</span>
-</div>"""
-EMPTY_HTML = """
-<div class="bit-empty">
-  <div class="bit-h2">No run loaded</div>
-  <div style="color:var(--text-secondary);max-width:46ch">
-    Configure a strategy on the left, or start from a worked example and edit it.
-  </div>
-  <div style="display:flex;gap:8px;margin-top:8px">
-    <span class="bit-kbd">⌘↵ RUN</span>
-    <span class="bit-kbd">⌘S SAVE CONFIG</span>
-    <span class="bit-kbd">COSTS DEFAULT ON</span>
-  </div>
-</div>"""
-def glossary_html() -> str:
-    rows = "".join(
-        f'<div style="margin-bottom:8px"><div class="bit-micro">{t}</div>'
-        f'<div style="font-size:11px;color:var(--text-secondary)">{d}</div></div>'
-        for t, d in GLOSSARY
-    )
-    return f'<div class="bit-panel">{rows}</div>'
 def trades_frame(rec: RunRecord | None) -> pd.DataFrame:
@@ -218,67 +88,81 @@ def trades_frame(rec: RunRecord | None) -> pd.DataFrame:
         "Entry": t["entry_ts"].dt.strftime("%Y-%m-%d %H:%M"),
         "Exit": t["exit_ts"].dt.strftime("%Y-%m-%d %H:%M"),
         "Side": t["side"].str.upper(),
-        "Entry px": t["entry_px"].round(2),
-        "Exit px": t["exit_px"].round(2),
         "Size": t["size"].round(4),
-        "Gross": t["gross_pnl"].round(2),
-        "Costs": t["costs"].round(2),
-        "Net": t["net_pnl"].round(2),
-        "R": t["r_multiple"].round(2),
-        "Bars": t["duration_bars"],
-        "MAE": (t["mae"] * 100).round(1),
-        "Segment": t["segment"],
-        "Trigger": t["trigger"],
     })
 def report_markdown(rec: RunRecord | None) -> str:
     if rec is None:
         return "_Run a backtest to generate the report._"
     r, req = rec.result, rec.request
     a, o, h = r.metrics_all, r.metrics_oos, r.metrics_holdout
-    bench = float(r.benchmark_equity.iloc[-1] / r.benchmark_equity.iloc[0] - 1.0) \
-        if len(r.benchmark_equity) else 0.0
     ratio = (o.sharpe / r.metrics_is.sharpe) if r.metrics_is.sharpe else float("nan")
     grade, checks = runtime.overfit_verdict(rec)
     lines = [
         f"### {rec.label}",
-        f"`RUN {rec.run_id} · {rec.created_at} · "
-        f"{req.validation_mode.upper()} · COSTS {'ON' if req.costs_on else 'OFF'}`",
-        "",
         f"Over {a.bars} bars and {a.trade_count} trades the strategy returns "
         f"**{pct(a.total_return)}** (CAGR {pct(a.cagr)}, Sharpe {num(a.sharpe)}) "
         f"against **{pct(bench)}** for buy and hold, with a maximum drawdown of "
         f"{pct(a.max_drawdown)}. Modelled costs of {money(r.costs_paid)} are already "
-        f"deducted — the costed number is the real one.",
-        "",
         (f"Out-of-sample Sharpe is {num(o.sharpe)}, which is {num(ratio)} of the "
          f"in-sample figure." if o.bars else
          "**No out-of-sample period was produced for this configuration**, so every "
          "number above is in-sample. Widen the date range or shorten the training "
-         "window before reading anything into it.")
-        + (f" On the locked holdout — {h.bars} bars that no parameter choice ever "
-           f"touched — it returns {pct(h.total_return)} at Sharpe {num(h.sharpe)}."
-           if h else ""),
-        "",
-        f"**Verdict: {grade}**",
         "",
     ]
     lines += [f"- {mark} {text}" for mark, text in checks]
     lines += ["", "#### Config snapshot", "```json",
-              _pretty_config(req), "```"]
     return "\n".join(lines)
-def _pretty_config(req: RunRequest) -> str:
-    import json
-    return json.dumps(asdict(req), indent=2, sort_keys=True)
-# --------------------------------------------------------------------------
-# Handlers
-# --------------------------------------------------------------------------
 def collect_request(strategy, asset, timeframe, date_range, model_slug,
@@ -289,7 +173,7 @@ def collect_request(strategy, asset, timeframe, date_range, model_slug,
     preset = strategies.PRESETS.get(strategy)
     params = {}
     if preset:
-        for (key, _label, _d, _lo, _hi), value in zip(preset.params, (p1, p2, p3)):
             if value is not None:
                 params[key] = value
     return RunRequest(
@@ -298,8 +182,8 @@ def collect_request(strategy, asset, timeframe, date_range, model_slug,
         costs_on=bool(costs_on), commission_bps=float(commission_bps),
         slippage_bps=float(slippage_bps),
         slippage_model="volume_scaled" if slippage_model == "Volume-scaled" else "fixed",
-        sizing_mode={"Fixed %": "fixed_pct", "Vol-target 15% ann.": "vol_target"}.get(
-            sizing_mode, "fixed_pct"),
         size_pct=float(size_pct), leverage=float(leverage),
         sl_pct=(float(sl_pct) / 100.0 if sl_pct else None),
         tp_pct=(float(tp_pct) / 100.0 if tp_pct else None),
@@ -312,71 +196,78 @@ def collect_request(strategy, asset, timeframe, date_range, model_slug,
     )
 def build_app() -> gr.Blocks:
-    css = theme.full_css()
-    with gr.Blocks(theme=theme.bit_theme(), css=css, title="Bit · Backtest Lab",
-                   analytics_enabled=False, fill_height=True) as demo:
-        history = gr.State([])       # list[RunRecord]
-        selected = gr.State([])      # run_ids chosen for comparison
-        current = gr.State(None)     # RunRecord
-        top_bar = gr.HTML(top_bar_html())
         with gr.Row(equal_height=False):
-            # ============================ LEFT ============================
-            with gr.Column(scale=2, min_width=280):
-                gr.HTML('<div class="bit-h2" style="padding:12px 4px 4px">Strategy Builder</div>')
                 with gr.Accordion("1 · STRATEGY", open=True, elem_classes="bit-accordion"):
-                    strategy = gr.Dropdown(
-                        choices=[p.name for p in strategies.PRESETS.values()],
-                        value="SMA Crossover", label="Preset", interactive=True,
-                    )
                     preset_note = gr.HTML("")
                     p1 = gr.Number(label="Fast MA", value=20, precision=4)
                     p2 = gr.Number(label="Slow MA", value=50, precision=4)
                     p3 = gr.Number(label="—", value=None, visible=False, precision=4)
-                    model_slug = gr.Dropdown(
-                        choices=runtime.available_models(), value=None,
-                        label="Forecast model", visible=False, interactive=True,
-                    )
-                with gr.Accordion("2 · UNIVERSE & DATA", open=True, elem_classes="bit-accordion"):
-                    asset = gr.Dropdown(choices=runtime.available_assets(),
-                                        value="BTC-USD", label="Asset")
-                    timeframe = gr.Radio(choices=list(config.TIMEFRAMES),
-                                         value="1d", label="Timeframe")
-                    date_range = gr.Radio(choices=["1Y", "3Y", "5Y", "Max"],
-                                          value="3Y", label="Date range")
                     coverage_note = gr.HTML("")
-                with gr.Accordion("3 · COSTS & EXECUTION", open=False, elem_classes="bit-accordion"):
                     costs_on = gr.Checkbox(value=True, label="Costs on")
-                    gr.HTML('<div class="bit-note">Costs on. Turning these off is how '
-                            'strategies lie to you.</div>')
                     commission_bps = gr.Number(value=10.0, label="Commission bps / side")
                     slippage_bps = gr.Number(value=5.0, label="Slippage bps")
-                    slippage_model = gr.Radio(choices=["Fixed bps", "Volume-scaled"],
                                               value="Fixed bps", label="Slippage model")
-                    gr.Radio(choices=["Next bar open"], value="Next bar open",
-                             label="Fill", interactive=False,
                              info="Next-bar-open execution is enforced by the engine.")
-                with gr.Accordion("4 · SIZING & RISK", open=False, elem_classes="bit-accordion"):
-                    sizing_mode = gr.Radio(choices=["Fixed %", "Vol-target 15% ann."],
                                            value="Fixed %", label="Sizing")
                     size_pct = gr.Slider(0.05, 1.0, value=1.0, step=0.05,
-                                         label="Position size (fraction of equity)")
                     leverage = gr.Slider(1.0, 3.0, value=1.0, step=0.5, label="Leverage")
                     sl_pct = gr.Number(value=None, label="Stop loss %")
                     tp_pct = gr.Number(value=None, label="Take profit %")
                     trail_pct = gr.Number(value=None, label="Trailing stop %")
-                with gr.Accordion("5 · VALIDATION", open=False, elem_classes="bit-accordion"):
                     validation_mode = gr.Radio(
-                        choices=["Walk-forward", "Simple split", "Holdout only", "None"],
                         value="Walk-forward", label="Mode")
                     train_m = gr.Number(value=12, label="Train months", precision=0)
                     test_m = gr.Number(value=3, label="Test months", precision=0)
@@ -393,94 +284,167 @@ def build_app() -> gr.Blocks:
                 share_out = gr.Textbox(label="Share token", visible=False,
                                        show_copy_button=True, lines=2)
-            # =========================== CENTER ===========================
-            with gr.Column(scale=7, min_width=520):
-                run_status = gr.HTML("")
                 stat_band = gr.HTML("")
-                empty_state = gr.HTML(EMPTY_HTML)
                 with gr.Tabs():
                     with gr.Tab("Overview"):
-                        equity_plot = gr.Plot(label=None)
                         with gr.Row():
                             log_scale = gr.Checkbox(value=False, label="Log scale")
                             cvd = gr.Checkbox(value=False, label="Colorblind-safe prices")
-                        regime_plot = gr.Plot(label=None)
                         with gr.Row():
-                            underwater_plot = gr.Plot(label=None)
-                            rolling_plot = gr.Plot(label=None)
-                        price_plot = gr.Plot(label=None)
                         with gr.Row():
-                            pnl_plot = gr.Plot(label=None)
-                            hold_plot = gr.Plot(label=None)
-                            mae_plot = gr.Plot(label=None)
                         costs_note = gr.HTML("")
                     with gr.Tab("Trades"):
-                        trades_head = gr.HTML("")
-                        trades_table = gr.Dataframe(
-                            value=pd.DataFrame(), interactive=False, wrap=False,
-                            elem_classes="bit-table", max_height=520,
-                        )
                         export_btn = gr.Button("Export CSV →", size="sm",
                                                elem_classes="bit-ghost-btn")
                         export_file = gr.File(label="trades.csv", visible=False)
-                    with gr.Tab("Comparison"):
-                        gr.HTML('<div class="bit-micro">CELL = OOS SHARPE · '
-                                'PRECOMPUTED FROM THE SIGNAL STORE</div>')
-                        heatmap_plot = gr.Plot(label=None)
-                        gr.HTML('<div class="bit-micro">SELECTED RUNS · '
-                                f'MAX {MAX_COMPARE}</div>')
-                        compare_picker = gr.CheckboxGroup(choices=[], value=[],
-                                                          label="Runs to compare")
-                        overlay_plot = gr.Plot(label=None)
-                        small_mult_plot = gr.Plot(label=None)
-                        with gr.Row():
-                            corr_plot = gr.Plot(label=None)
-                            regime_bars_plot = gr.Plot(label=None)
-                        metrics_table = gr.Dataframe(value=pd.DataFrame(),
-                                                     interactive=False,
-                                                     elem_classes="bit-table")
                     with gr.Tab("Robustness"):
                         verdict_html = gr.HTML("")
                         with gr.Row():
-                            wf_plot = gr.Plot(label=None)
-                            mc_plot = gr.Plot(label=None)
                         with gr.Row():
-                            sens_plot = gr.Plot(label=None)
-                            slip_plot = gr.Plot(label=None)
-                        robust_btn = gr.Button(
-                            "Run sensitivity + slippage stress (slower)",
-                            size="sm", elem_classes="bit-ghost-btn")
                     with gr.Tab("Report"):
                         report_md = gr.Markdown("_Run a backtest to generate the report._")
-                        report_equity = gr.Plot(label=None)
                         with gr.Row():
-                            save_run_btn = gr.Button("Save run summary", size="sm",
                                                      elem_classes="bit-ghost-btn")
                             copy_cfg_btn = gr.Button("Copy share link", size="sm",
                                                      elem_classes="bit-ghost-btn")
                         save_note = gr.HTML("")
                     with gr.Tab("Coverage"):
-                        gr.HTML('<div class="bit-micro">SIGNAL STORE COVERAGE MAP · '
-                                'MODEL × ASSET × TIMEFRAME</div>')
-                        coverage_table = gr.Dataframe(
-                            value=runtime.coverage_frame(), interactive=False,
-                            elem_classes="bit-table", max_height=420)
-                        gr.HTML('<div class="bit-micro" style="margin-top:8px">'
-                                'EXTEND COVERAGE</div>')
                         extend_panel = gr.HTML("")
                         with gr.Row():
-                            ext_model = gr.Dropdown(choices=list(config.SEED_MODELS),
-                                                    label="Model", scale=2)
-                            ext_asset = gr.Dropdown(choices=list(config.ASSETS),
-                                                    label="Asset", scale=2)
-                            ext_tf = gr.Dropdown(choices=list(config.TIMEFRAMES),
-                                                 value="1d", label="Timeframe", scale=1)
                         with gr.Row():
                             ext_start = gr.Textbox(label="Start (YYYY-MM-DD)", scale=2)
                             ext_end = gr.Textbox(label="End (YYYY-MM-DD)", scale=2)
@@ -490,169 +454,136 @@ def build_app() -> gr.Blocks:
                             extend_btn = gr.Button("Extend coverage", size="sm",
                                                    elem_classes="bit-run-btn")
                         extend_out = gr.HTML("")
-                        gr.HTML('<div class="bit-micro" style="margin-top:16px">'
-                                'ADD MODEL</div>')
                         with gr.Row():
                             add_family = gr.Dropdown(
-                                choices=list(config.ALLOWED_ADAPTER_FAMILIES),
-                                value="chronos", label="Adapter family", scale=1)
                             add_model_id = gr.Textbox(label="HF model id (owner/name)",
                                                       scale=2)
                             add_btn = gr.Button("Smoke test & add", size="sm",
                                                 elem_classes="bit-ghost-btn", scale=1)
                         add_out = gr.HTML("")
-            # ============================ RIGHT ===========================
-            with gr.Column(scale=2, min_width=250):
-                gr.HTML('<div class="bit-h2" style="padding:12px 4px 4px">Run Manager</div>')
-                login_slot = gr.HTML("")
                 try:
                     gr.LoginButton(value="Sign in with Hugging Face", size="sm")
                 except Exception:
-                    login_slot.value = (
-                        '<div class="bit-note">Sign-in appears when the Space runs '
-                        'on Hugging Face with OAuth enabled.</div>')
-                history_html = gr.HTML(
-                    '<div class="bit-micro">NO RUNS YET IN THIS SESSION</div>')
                 with gr.Accordion("METRICS GLOSSARY", open=False,
                                   elem_classes="bit-accordion"):
-                    gr.HTML(glossary_html())
                 with gr.Accordion("HOW RESULTS ARE COMPUTED", open=False,
                                   elem_classes="bit-accordion"):
-                    gr.Markdown(
-                        "- Fills execute at the **next bar's open**. A decision at bar "
-                        "`t` can never trade at bar `t`.\n"
-                        "- Strategies are checked for lookahead by perturbing future "
-                        "prices and asserting past outputs do not move.\n"
-                        "- Costs are **on by default**: commission per side plus a "
-                        "slippage model, both booked onto every trade row.\n"
-                        "- The locked holdout is excluded from every parameter-selection "
-                        "path, not merely reported separately.\n"
-                        "- The store holds raw model outputs only. Trading rules are "
-                        "applied live, per run."
-                    )
-        gr.HTML(FOOTER_HTML)
-        # ------------------------------------------------------------------
-        # Wiring
-        # ------------------------------------------------------------------
-        builder_inputs = [strategy, asset, timeframe, date_range, model_slug,
-                          p1, p2, p3, costs_on, commission_bps, slippage_bps,
-                          slippage_model, sizing_mode, size_pct, leverage,
-                          sl_pct, tp_pct, trail_pct,
-                          validation_mode, train_m, test_m, roll_m, holdout_m]
-        overview_outputs = [equity_plot, regime_plot, underwater_plot, rolling_plot,
-                            price_plot, pnl_plot, hold_plot, mae_plot, costs_note]
         def on_strategy_change(name):
             preset = strategies.PRESETS.get(name)
             if preset is None:
-                return (gr.update(), gr.update(), gr.update(), gr.update(), "")
-            if not preset.available:
-                note = (f'<div class="bit-note bit-note-danger">{name} is unavailable. '
-                        f'{preset.unavailable_reason}</div>')
-            else:
-                note = ""
             ups = []
             for i in range(3):
                 if i < len(preset.params):
-                    key, label, default, lo, hi = preset.params[i]
                     ups.append(gr.update(label=label, value=default, visible=True))
                 else:
                     ups.append(gr.update(visible=False, value=None))
-            model_up = gr.update(visible=preset.needs_signals,
-                                 choices=runtime.available_models(),
-                                 value=(runtime.available_models() or [None])[0]
-                                 if preset.needs_signals else None)
-            return (*ups, model_up, note)
-        strategy.change(on_strategy_change, [strategy], [p1, p2, p3, model_slug, preset_note])
         def on_universe_change(a, tf):
             cov = runtime.price_coverage_for(a, tf)
             models = runtime.available_models(a, tf)
             if cov is None:
-                html = ('<div class="bit-note bit-note-danger">No cached price coverage '
-                        f'for {a} {tf}. Pick another pair or extend coverage.</div>')
             else:
-                html = (f'<div class="bit-micro">CACHED {cov[0]} → {cov[1]}'
-                        + (f' · MODELS: {", ".join(models)}' if models else
-                           ' · NO MODEL SIGNALS') + '</div>')
-            return html, gr.update(choices=models,
-                                   value=(models[0] if models else None))
         asset.change(on_universe_change, [asset, timeframe], [coverage_note, model_slug])
-        timeframe.change(on_universe_change, [asset, timeframe], [coverage_note, model_slug])
         def do_run(hist, *vals, progress=gr.Progress()):
             progress(0.05, desc="Reading cached slices")
             req = collect_request(*vals)
             try:
-                progress(0.35, desc="Simulating trades")
                 rec = runtime.execute(req)
             except (RunError, ValueError) as e:
-                err = f'<div class="bit-note bit-note-danger">{e}</div>'
-                return (hist, None, top_bar_html("RUN FAILED", "bit-chip bit-chip-warn"),
-                        err, "", gr.update(visible=True), *(gr.update(),) * 9,
-                        "", pd.DataFrame(), "", gr.update(choices=[], value=[]),
                         "_Run failed._", None)
-            progress(0.75, desc="Building charts")
             hist = ([rec] + list(hist))[:40]
-            figs = build_overview(rec, log_scale=False, cvd=False)
-            progress(0.95, desc="Computing robustness")
-            choices = [f"{r.run_id} · {r.label}" for r in hist]
-            return (
-                hist, rec,
-                top_bar_html(f"RUN {rec.run_id} COMPLETE · {rec.elapsed_s:.1f}S",
-                             "bit-chip bit-chip-ok", context_chip(req)),
-                "", stat_band_html(rec), gr.update(visible=False),
-                *figs,
-                trades_head_html(rec), trades_frame(rec),
-                history_list_html(hist),
-                gr.update(choices=choices, value=choices[:1]),
-                report_markdown(rec),
-                charts.equity_curve(rec.result.equity, rec.result.benchmark_equity,
-                                    plan=rec.result.plan),
-            )
-        run_outputs = [history, current, top_bar, run_status, stat_band, empty_state,
-                       *overview_outputs, trades_head, trades_table, history_html,
-                       compare_picker, report_md, report_equity]
-        run_btn.click(do_run, [history, *builder_inputs], run_outputs)
         def load_example():
-            """The design's worked example: Sentiment-Gated Momentum on BTC 1h."""
-            return ("Sentiment-Gated Momentum", "BTC-USD", "1h", "1Y",
-                    gr.update(value=20, label="Fast MA", visible=True),
-                    gr.update(value=50, label="Slow MA", visible=True),
-                    gr.update(value=0.40, label="Sentiment gate", visible=True),
                     True, "Walk-forward")
         example_btn.click(
             load_example, None,
-            [strategy, asset, timeframe, date_range, p1, p2, p3, costs_on, validation_mode],
-        ).then(do_run, [history, *builder_inputs], run_outputs)
         def replot(rec, log_s, colorblind):
             if rec is None:
                 return (gr.update(),) * 9
             return build_overview(rec, log_scale=log_s, cvd=colorblind)
-        log_scale.change(replot, [current, log_scale, cvd], overview_outputs)
-        cvd.change(replot, [current, log_scale, cvd], overview_outputs)
-        def do_share(*vals):
-            req = collect_request(*vals)
-            return gr.update(value=req.encode(), visible=True)
-        share_btn.click(do_share, builder_inputs, [share_out])
         copy_cfg_btn.click(lambda r: gr.update(value=r.request.encode(), visible=True)
                            if r else gr.update(), [current], [share_out])
@@ -667,76 +598,29 @@ def build_app() -> gr.Blocks:
         def do_save(rec):
             if rec is None:
-                return '<div class="bit-note">Nothing to save yet.</div>'
             try:
-                rid = runtime.save_run_summary(rec)
-                return (f'<div class="bit-note">Run <b>{rid}</b> staged for the store. '
-                        f'It is committed with the next batch.</div>')
             except Exception as e:
-                return f'<div class="bit-note bit-note-danger">Could not save: {e}</div>'
         save_run_btn.click(do_save, [current], [save_note])
-        def do_compare(hist, picks):
-            hist = list(hist or [])
-            by_id = {f"{r.run_id} · {r.label}": r for r in hist}
-            chosen = [by_id[p] for p in (picks or [])[:MAX_COMPARE] if p in by_id]
-            if not chosen:
-                empty = charts.empty_figure("select runs to compare")
-                return empty, empty, empty, empty, pd.DataFrame()
-            curves = {r.label[:28]: r.result.equity for r in chosen}
-            rets = {r.label[:28]: r.result.equity.pct_change().dropna() for r in chosen}
-            regimes = [runtime.regime_breakdown(r) for r in chosen]
-            regime_df = regimes[0] if regimes else pd.DataFrame()
-            for extra in regimes[1:]:
-                if not extra.empty and not regime_df.empty:
-                    regime_df = regime_df.merge(extra, on="regime", how="outer")
-            rows = []
-            for label, key in (("Total return", "total_return"), ("CAGR", "cagr"),
-                               ("OOS Sharpe", None), ("Max drawdown", "max_drawdown"),
-                               ("Win rate", "win_rate"), ("Trades", "trade_count")):
-                row = {"Metric": label}
-                for r in chosen:
-                    if key is None:
-                        row[r.label[:22]] = num(r.result.metrics_oos.sharpe)
-                    elif key == "trade_count":
-                        row[r.label[:22]] = r.result.metrics_all.trade_count
-                    elif key == "win_rate":
-                        row[r.label[:22]] = pct(r.result.metrics_all.win_rate, 0, False)
-                    else:
-                        row[r.label[:22]] = pct(getattr(r.result.metrics_all, key))
-                rows.append(row)
-            return (charts.overlaid_returns(curves),
-                    charts.small_multiples(curves),
-                    charts.correlation_matrix(rets),
-                    charts.regime_bars(regime_df),
-                    pd.DataFrame(rows))
-        compare_picker.change(do_compare, [history, compare_picker],
-                              [overlay_plot, small_mult_plot, corr_plot,
-                               regime_bars_plot, metrics_table])
         def do_robustness(rec):
             if rec is None:
                 e = charts.empty_figure("run a backtest first")
-                return "", e, e, e, e
             grade, checks = runtime.overfit_verdict(rec)
-            items = "".join(
-                f'<div style="font-size:11px;color:var(--text-secondary)">{m} {t}</div>'
-                for m, t in checks)
-            html = (f'<div class="bit-panel"><div class="bit-h2">Overfit verdict: '
-                    f'<span style="color:var(--accent-amber-strong)">{grade}</span></div>'
                     f'{items}</div>')
-            wf = charts.walk_forward_bars(rec.result.windows)
-            mc = charts.monte_carlo_cone(charts.monte_carlo_paths(rec.result.trades))
-            e = charts.empty_figure("press the button below to run the sweep")
-            return html, wf, mc, e, e
-        current.change(do_robustness, [current],
-                       [verdict_html, wf_plot, mc_plot, sens_plot, slip_plot])
         def do_sweep(rec, progress=gr.Progress()):
             if rec is None:
@@ -749,37 +633,83 @@ def build_app() -> gr.Blocks:
                 sens = charts.empty_figure("this preset has fewer than two parameters")
             else:
                 progress(0.1, desc="Parameter sweep")
-                base_x = req.params.get(keys[0], 20)
-                base_y = req.params.get(keys[1], 50)
-                xs = sorted({max(2, int(base_x * m)) for m in (0.5, 0.75, 1.0, 1.5, 2.0)})
-                ys = sorted({max(3, int(base_y * m)) for m in (0.5, 0.75, 1.0, 1.5, 2.0)})
-                grid = runtime.parameter_sweep(req, keys[0], xs, keys[1], ys)
-                sens = charts.parameter_sensitivity(grid, x=keys[0], y=keys[1])
             progress(0.7, desc="Slippage stress")
-            slip = charts.slippage_stress(runtime.slippage_stress(req))
-            return sens, slip
         robust_btn.click(do_sweep, [current], [sens_plot, slip_plot])
-        # ---- Coverage / extension (Phase 4) ----
-        from src import extension
-        estimate_btn.click(extension.estimate_ui, [ext_model, ext_asset, ext_tf,
-                                                   ext_start, ext_end], [extend_out])
         extend_btn.click(extension.extend_ui,
                          [ext_model, ext_asset, ext_tf, ext_start, ext_end],
                          [extend_out, coverage_table])
         add_btn.click(extension.add_model_ui, [add_family, add_model_id],
                       [add_out, coverage_table])
-        # ---- Load: heatmap, coverage, share-link restore ----
         def on_load(request: gr.Request):
-            store = runtime.get_store()
-            heat = comparisons.load_table(store, comparisons.HEATMAP)
-            fig = charts.strategy_timeframe_heatmap(heat)
-            cov_html = ""
             restored = [gr.update()] * 4
-            token = None
             try:
                 token = dict(request.query_params).get("cfg") if request else None
             except Exception:
@@ -790,82 +720,28 @@ def build_app() -> gr.Blocks:
                     restored = [gr.update(value=req.strategy), gr.update(value=req.asset),
                                 gr.update(value=req.timeframe),
                                 gr.update(value=req.date_range)]
-                    cov_html = '<div class="bit-note">Config restored from share link.</div>'
                 except Exception as e:
-                    cov_html = (f'<div class="bit-note bit-note-danger">'
-                                f'Share link rejected: {e}</div>')
-            return (fig, runtime.coverage_frame(), extension.status_html(),
-                    cov_html, *restored)
         demo.load(on_load, None,
-                  [heatmap_plot, coverage_table, extend_panel, run_status,
                    strategy, asset, timeframe, date_range])
         demo.load(lambda: on_universe_change("BTC-USD", "1d"), None,
                   [coverage_note, model_slug])
     return demo
-def trades_head_html(rec: RunRecord | None) -> str:
-    if rec is None or rec.result.trades.empty:
-        return '<div class="bit-micro">NO TRADES</div>'
-    n = len(rec.result.trades)
-    costs = rec.result.costs_paid
-    return (f'<div class="bit-micro">{n} TOTAL · COSTS PAID {money(costs)} · '
-            f'FILLS AT NEXT BAR OPEN</div>')
-def history_list_html(hist) -> str:
-    if not hist:
-        return '<div class="bit-micro">NO RUNS YET IN THIS SESSION</div>'
-    rows = []
-    for r in hist[:12]:
-        s = r.sharpe
-        color = ("var(--accent-moss-strong)" if s >= 1
-                 else "var(--fin-down)" if s < 0 else "var(--text-secondary)")
-        rows.append(
-            f'<div style="border:1px solid var(--border-subtle);padding:6px;margin-bottom:4px">'
-            f'<div style="font-size:11px;color:var(--text-primary)">{r.label[:34]}</div>'
-            f'<div class="bit-micro">{r.meta}</div>'
-            f'<div style="font-family:var(--font-mono);font-size:12px;color:{color}">'
-            f'SHARPE {s:.2f}</div></div>'
-        )
-    return "".join(rows)
-def build_overview(rec: RunRecord, *, log_scale: bool, cvd: bool):
-    r = rec.result
-    bpy = config.bars_per_year(rec.request.asset, rec.request.timeframe)
-    window = {"1d": 90, "1h": 24 * 30, "15m": 4 * 24 * 14}.get(rec.request.timeframe, 90)
-    costs_html = (
-        f'<div class="bit-note">COSTS PAID TOTAL: {money(r.costs_paid)}. '
-        f'The costed number is the real one.</div>'
-        if rec.request.costs_on else
-        '<div class="bit-note bit-note-danger">COSTS ARE OFF. '
-        'These numbers are not achievable.</div>'
-    )
-    return (
-        charts.equity_curve(r.equity, r.benchmark_equity, plan=r.plan,
-                            log_scale=log_scale, cvd=cvd),
-        charts.regime_strip(r.prices),
-        charts.underwater_chart(r.equity),
-        charts.rolling_sharpe_chart(r.equity, window, bpy),
-        charts.price_with_trades(r.prices, r.trades, cvd=cvd),
-        charts.pnl_histogram(r.trades, cvd=cvd),
-        charts.holding_period_histogram(r.trades),
-        charts.mae_mfe_scatter(r.trades, cvd=cvd),
-        costs_html,
-    )
 demo = build_app()
 if __name__ == "__main__":
     gr.set_static_paths(paths=theme.static_paths())
     demo.queue(max_size=32).launch(
-        server_name="0.0.0.0",
-        server_port=int(os.environ.get("PORT", 7860)),
-        show_api=False,
-    )

 """Bit Trading Company — Backtest Lab.
+Layout follows the design: Strategy Builder on the left, tabbed Results Canvas
+in the centre, Run Manager on the right, disclaimer pinned to the footer.
+This module is layout and wiring only. Formatting lives in `src/ui/format.py`,
+markup in `src/ui/components.py`, the global comparison in
+`src/ui/compare_tab.py`, and every number in `src/metrics.py`.
 The app reads exclusively from the cached signal store. The only path that can
 reach an external provider is the batch refresh in `scripts/seed_store.py`, and
+the only path that runs inference is the ZeroGPU flow in `src/extension.py`,
+which spends the signed-in user's own quota.
 """
 from __future__ import annotations
 import logging
 import os
 import gradio as gr
 import pandas as pd
+from src import catalog, charts, config, extension, runtime, strategies
 from src.runtime import RunError, RunRecord, RunRequest
+from src.ui import components as C
+from src.ui import compare_tab as CT
 from src.ui import theme
+from src.ui.format import EM, count, money, num, pct
 logging.basicConfig(level=logging.INFO, format="%(asctime)s %(name)s %(message)s")
 log = logging.getLogger("bit.app")
 MAX_COMPARE = 6
 GLOSSARY = [
     ("SHARPE", "Annualized mean excess return divided by return volatility. "
                "Above 1 is good; above 3 usually means a bug."),
     ("MAE / MFE", "Worst and best unrealized excursion while the trade was open."),
     ("WALK-FORWARD", "Train on a rolling window, test on the next unseen window, repeat."),
     ("OOS", "Out of sample: data the parameters never saw during fitting."),
+    ("BASELINE", "A naive forecast (random walk, drift, seasonal). If a learned "
+                 "model cannot beat these, it has not earned its inference cost."),
 ]
+METHOD_NOTES = """
+- Fills execute at the **next bar's open**. A decision at bar `t` can never trade at bar `t`.
+- Strategies are checked for lookahead by perturbing future prices up *and* down
+  and asserting past outputs do not move.
+- Costs are **on by default**: commission per side plus a slippage model, both
+  booked onto every trade row. `gross − costs = net` holds exactly.
+- The locked holdout is excluded from every parameter-selection path, not merely
+  reported separately.
+- The store holds raw model outputs only. Trading rules are applied live, per run.
+- Catalog rows all use one canonical config (costs on, walk-forward 12/3/3,
+  6-month holdout) so they are comparable with each other.
+"""
 # --------------------------------------------------------------------------
+# Presentation helpers
 # --------------------------------------------------------------------------
+def context_chip_text(req: RunRequest) -> str:
     mode = {"walk_forward": "WALK-FORWARD", "holdout": "HOLDOUT",
             "split": "SPLIT", "none": "NO SPLIT"}.get(req.validation_mode, "")
     try:
         span = f"{s.date()} → {e.date()}"
     except Exception:
         span = req.date_range
+    return f"{req.asset} · {req.timeframe.upper()} · {span} · {mode}"
 def trades_frame(rec: RunRecord | None) -> pd.DataFrame:
         "Entry": t["entry_ts"].dt.strftime("%Y-%m-%d %H:%M"),
         "Exit": t["exit_ts"].dt.strftime("%Y-%m-%d %H:%M"),
         "Side": t["side"].str.upper(),
+        "Entry px": t["entry_px"].round(2), "Exit px": t["exit_px"].round(2),
         "Size": t["size"].round(4),
+        "Gross": t["gross_pnl"].round(2), "Costs": t["costs"].round(2),
+        "Net": t["net_pnl"].round(2), "R": t["r_multiple"].round(2),
+        "Bars": t["duration_bars"], "MAE": (t["mae"] * 100).round(1),
+        "Segment": t["segment"], "Trigger": t["trigger"],
     })
+def trades_head(rec: RunRecord | None) -> str:
+    if rec is None or rec.result.trades.empty:
+        return C.micro("no trades")
+    return C.micro(f"{len(rec.result.trades)} total · costs paid "
+                   f"{money(rec.result.costs_paid)} · fills at next bar open")
 def report_markdown(rec: RunRecord | None) -> str:
     if rec is None:
         return "_Run a backtest to generate the report._"
+    import json
+    from dataclasses import asdict
     r, req = rec.result, rec.request
     a, o, h = r.metrics_all, r.metrics_oos, r.metrics_holdout
+    bench = (float(r.benchmark_equity.iloc[-1] / r.benchmark_equity.iloc[0] - 1.0)
+             if len(r.benchmark_equity) else float("nan"))
     ratio = (o.sharpe / r.metrics_is.sharpe) if r.metrics_is.sharpe else float("nan")
     grade, checks = runtime.overfit_verdict(rec)
     lines = [
         f"### {rec.label}",
+        f"`RUN {rec.run_id} · {rec.created_at} · {req.validation_mode.upper()} · "
+        f"COSTS {'ON' if req.costs_on else 'OFF'}`", "",
         f"Over {a.bars} bars and {a.trade_count} trades the strategy returns "
         f"**{pct(a.total_return)}** (CAGR {pct(a.cagr)}, Sharpe {num(a.sharpe)}) "
         f"against **{pct(bench)}** for buy and hold, with a maximum drawdown of "
         f"{pct(a.max_drawdown)}. Modelled costs of {money(r.costs_paid)} are already "
+        f"deducted — the costed number is the real one.", "",
         (f"Out-of-sample Sharpe is {num(o.sharpe)}, which is {num(ratio)} of the "
          f"in-sample figure." if o.bars else
          "**No out-of-sample period was produced for this configuration**, so every "
          "number above is in-sample. Widen the date range or shorten the training "
+         "window before reading anything into it."),
         "",
+        (f"On the locked holdout — {h.bars} bars no parameter choice ever touched — "
+         f"it returns {pct(h.total_return)} at Sharpe {num(h.sharpe)}." if h else ""),
+        "", f"**Verdict: {grade}**", "",
     ]
     lines += [f"- {mark} {text}" for mark, text in checks]
     lines += ["", "#### Config snapshot", "```json",
+              json.dumps(asdict(req), indent=2, sort_keys=True), "```"]
     return "\n".join(lines)
+def build_overview(rec: RunRecord, *, log_scale: bool, cvd: bool):
+    r = rec.result
+    bpy = config.bars_per_year(rec.request.asset, rec.request.timeframe)
+    window = {"1d": 90, "1h": 24 * 30, "15m": 4 * 24 * 14}.get(rec.request.timeframe, 90)
+    costs_html = (
+        C.note(f"<b>COSTS PAID TOTAL: {money(r.costs_paid)}</b>. "
+               "The costed number is the real one.")
+        if rec.request.costs_on else
+        C.note("<b>COSTS ARE OFF.</b> These numbers are not achievable.", danger=True))
+    return (
+        charts.equity_curve(r.equity, r.benchmark_equity, plan=r.plan,
+                            log_scale=log_scale, cvd=cvd),
+        charts.regime_strip(r.prices),
+        charts.underwater_chart(r.equity),
+        charts.rolling_sharpe_chart(r.equity, window, bpy),
+        charts.price_with_trades(r.prices, r.trades, cvd=cvd),
+        charts.pnl_histogram(r.trades, cvd=cvd),
+        charts.holding_period_histogram(r.trades),
+        charts.mae_mfe_scatter(r.trades, cvd=cvd),
+        costs_html,
+    )
 def collect_request(strategy, asset, timeframe, date_range, model_slug,
     preset = strategies.PRESETS.get(strategy)
     params = {}
     if preset:
+        for (key, _l, _d, _lo, _hi), value in zip(preset.params, (p1, p2, p3)):
             if value is not None:
                 params[key] = value
     return RunRequest(
         costs_on=bool(costs_on), commission_bps=float(commission_bps),
         slippage_bps=float(slippage_bps),
         slippage_model="volume_scaled" if slippage_model == "Volume-scaled" else "fixed",
+        sizing_mode={"Fixed %": "fixed_pct",
+                     "Vol-target 15% ann.": "vol_target"}.get(sizing_mode, "fixed_pct"),
         size_pct=float(size_pct), leverage=float(leverage),
         sl_pct=(float(sl_pct) / 100.0 if sl_pct else None),
         tp_pct=(float(tp_pct) / 100.0 if tp_pct else None),
     )
+# --------------------------------------------------------------------------
+# App
+# --------------------------------------------------------------------------
 def build_app() -> gr.Blocks:
+    store = runtime.get_store()
+    all_assets = runtime.available_assets()
+    all_tfs = list(config.TIMEFRAMES)
+    all_strategies = [p.name for p in strategies.PRESETS.values()]
+    all_models = sorted(config.SEED_MODELS)
+    with gr.Blocks(theme=theme.bit_theme(), css=theme.full_css(),
+                   title="Bit · Backtest Lab", analytics_enabled=False,
+                   fill_height=True) as demo:
+        history = gr.State([])
+        current = gr.State(None)
+        top_bar = gr.HTML(C.top_bar())
         with gr.Row(equal_height=False):
+            # ------------------------- LEFT -------------------------
+            with gr.Column(scale=2, min_width=270):
+                gr.HTML('<div class="bit-zone-title">Strategy Builder</div>')
                 with gr.Accordion("1 · STRATEGY", open=True, elem_classes="bit-accordion"):
+                    strategy = gr.Dropdown(all_strategies, value="SMA Crossover",
+                                           label="Preset", interactive=True)
                     preset_note = gr.HTML("")
                     p1 = gr.Number(label="Fast MA", value=20, precision=4)
                     p2 = gr.Number(label="Slow MA", value=50, precision=4)
                     p3 = gr.Number(label="—", value=None, visible=False, precision=4)
+                    model_slug = gr.Dropdown(runtime.available_models(), value=None,
+                                             label="Forecast model", visible=False)
+                with gr.Accordion("2 · UNIVERSE & DATA", open=True,
+                                  elem_classes="bit-accordion"):
+                    asset = gr.Dropdown(all_assets, value="BTC-USD", label="Asset")
+                    timeframe = gr.Radio(all_tfs, value="1d", label="Timeframe")
+                    date_range = gr.Radio(["1Y", "3Y", "5Y", "Max"], value="3Y",
+                                          label="Date range")
                     coverage_note = gr.HTML("")
+                with gr.Accordion("3 · COSTS & EXECUTION", open=False,
+                                  elem_classes="bit-accordion"):
                     costs_on = gr.Checkbox(value=True, label="Costs on")
+                    gr.HTML(C.note("Costs on. Turning these off is how strategies "
+                                   "lie to you."))
                     commission_bps = gr.Number(value=10.0, label="Commission bps / side")
                     slippage_bps = gr.Number(value=5.0, label="Slippage bps")
+                    slippage_model = gr.Radio(["Fixed bps", "Volume-scaled"],
                                               value="Fixed bps", label="Slippage model")
+                    gr.Radio(["Next bar open"], value="Next bar open", label="Fill",
+                             interactive=False,
                              info="Next-bar-open execution is enforced by the engine.")
+                with gr.Accordion("4 · SIZING & RISK", open=False,
+                                  elem_classes="bit-accordion"):
+                    sizing_mode = gr.Radio(["Fixed %", "Vol-target 15% ann."],
                                            value="Fixed %", label="Sizing")
                     size_pct = gr.Slider(0.05, 1.0, value=1.0, step=0.05,
+                                         label="Position size")
                     leverage = gr.Slider(1.0, 3.0, value=1.0, step=0.5, label="Leverage")
                     sl_pct = gr.Number(value=None, label="Stop loss %")
                     tp_pct = gr.Number(value=None, label="Take profit %")
                     trail_pct = gr.Number(value=None, label="Trailing stop %")
+                with gr.Accordion("5 · VALIDATION", open=False,
+                                  elem_classes="bit-accordion"):
                     validation_mode = gr.Radio(
+                        ["Walk-forward", "Simple split", "Holdout only", "None"],
                         value="Walk-forward", label="Mode")
                     train_m = gr.Number(value=12, label="Train months", precision=0)
                     test_m = gr.Number(value=3, label="Test months", precision=0)
                 share_out = gr.Textbox(label="Share token", visible=False,
                                        show_copy_button=True, lines=2)
+            # ------------------------ CENTER ------------------------
+            with gr.Column(scale=8, min_width=560):
                 stat_band = gr.HTML("")
                 with gr.Tabs():
+                    # ---------------- COMPARE (landing) ----------------
+                    with gr.Tab("Compare"):
+                        gr.HTML('<div class="bit-zone-title">Everything this Space '
+                                'has computed<span class="bit-micro">CATALOG · '
+                                'PRECOMPUTED FROM THE SIGNAL STORE</span></div>')
+                        catalog_meta = gr.HTML("")
+                        with gr.Tabs():
+                            with gr.Tab("Leaderboard"):
+                                podium = gr.HTML("")
+                                with gr.Row():
+                                    f_metric = gr.Dropdown(
+                                        list(CT.RANK_METRICS), value="OOS Sharpe",
+                                        label="Rank by", scale=2)
+                                    f_top = gr.Slider(5, 50, value=15, step=5,
+                                                      label="Show top", scale=2)
+                                    f_min_trades = gr.Slider(
+                                        0, 100, value=0, step=5,
+                                        label="Min trades", scale=2)
+                                # Collapsed by default: an unfiltered board is the
+                                # useful default view, and an open wall of chips
+                                # pushes the table and charts below the fold.
+                                with gr.Accordion("FILTERS · ALL COMBINATIONS SHOWN",
+                                                  open=False,
+                                                  elem_classes="bit-accordion"):
+                                    with gr.Row():
+                                        f_assets = gr.CheckboxGroup(all_assets, value=[],
+                                                                    label="Assets")
+                                        f_tfs = gr.CheckboxGroup(all_tfs, value=[],
+                                                                 label="Timeframes")
+                                    f_strats = gr.CheckboxGroup(
+                                        list(catalog.CATALOG_STRATEGIES), value=[],
+                                        label="Strategies")
+                                    f_models = gr.CheckboxGroup(all_models, value=[],
+                                                                label="Models")
+                                    with gr.Row():
+                                        f_oos = gr.Checkbox(
+                                            value=True, label="Require out-of-sample")
+                                        f_hide_base = gr.Checkbox(
+                                            value=False, label="Hide baseline models")
+                                        reset_btn = gr.Button(
+                                            "Reset filters", size="sm",
+                                            elem_classes="bit-ghost-btn")
+                                lb_meta = gr.HTML("")
+                                lb_table = gr.Dataframe(pd.DataFrame(), interactive=False,
+                                                        wrap=False, max_height=430,
+                                                        elem_classes="bit-table")
+                                gr.HTML('<div class="bit-zone-title">Returns over time'
+                                        '<span class="bit-micro">TOP RANKED · '
+                                        'CUMULATIVE, COSTS INCLUDED</span></div>')
+                                lb_overlay = gr.Plot()
+                                gr.HTML('<div class="bit-zone-title">Risk vs return'
+                                        '<span class="bit-micro">MARKER AREA = TRADE '
+                                        'COUNT</span></div>')
+                                lb_scatter = gr.Plot()
+                            with gr.Tab("Models"):
+                                models_tf = gr.Radio(["all"] + all_tfs, value="1d",
+                                                     label="Timeframe")
+                                models_note = gr.HTML("")
+                                with gr.Row():
+                                    acc_plot = gr.Plot()
+                                    cal_plot = gr.Plot()
+                                gr.HTML('<div class="bit-zone-title">Best result per '
+                                        'model<span class="bit-micro">ACROSS EVERY '
+                                        'STRATEGY AND ASSET</span></div>')
+                                model_bars = gr.Plot()
+                                score_table = gr.Dataframe(pd.DataFrame(),
+                                                           interactive=False,
+                                                           max_height=380,
+                                                           elem_classes="bit-table")
+                            with gr.Tab("Signals"):
+                                with gr.Row():
+                                    sig_asset = gr.Dropdown(all_assets, value="BTC-USD",
+                                                            label="Asset")
+                                    sig_tf = gr.Radio(all_tfs, value="1d",
+                                                      label="Timeframe")
+                                sig_panel = gr.HTML("")
+                                gr.HTML(C.micro(
+                                    "every model's latest stored forecast for this "
+                                    "slice · weighted by realised accuracy"))
+                            with gr.Tab("Run history"):
+                                gr.HTML(C.micro("session runs and runs saved to the "
+                                                "signal store"))
+                                runs_refresh = gr.Button("Refresh from store", size="sm",
+                                                         elem_classes="bit-ghost-btn")
+                                runs_table = gr.Dataframe(pd.DataFrame(),
+                                                          interactive=False,
+                                                          max_height=430,
+                                                          elem_classes="bit-table")
+                    # ---------------- OVERVIEW ----------------
                     with gr.Tab("Overview"):
+                        empty_state = gr.HTML(C.empty_state())
+                        equity_plot = gr.Plot()
                         with gr.Row():
                             log_scale = gr.Checkbox(value=False, label="Log scale")
                             cvd = gr.Checkbox(value=False, label="Colorblind-safe prices")
+                        regime_plot = gr.Plot()
                         with gr.Row():
+                            underwater_plot = gr.Plot()
+                            rolling_plot = gr.Plot()
+                        price_plot = gr.Plot()
                         with gr.Row():
+                            pnl_plot = gr.Plot()
+                            hold_plot = gr.Plot()
+                            mae_plot = gr.Plot()
                         costs_note = gr.HTML("")
                     with gr.Tab("Trades"):
+                        trades_head_html = gr.HTML("")
+                        trades_table = gr.Dataframe(pd.DataFrame(), interactive=False,
+                                                    wrap=False, max_height=520,
+                                                    elem_classes="bit-table")
                         export_btn = gr.Button("Export CSV →", size="sm",
                                                elem_classes="bit-ghost-btn")
                         export_file = gr.File(label="trades.csv", visible=False)
                     with gr.Tab("Robustness"):
                         verdict_html = gr.HTML("")
                         with gr.Row():
+                            wf_plot = gr.Plot()
+                            mc_plot = gr.Plot()
+                        robust_btn = gr.Button("Run sensitivity + slippage stress",
+                                               size="sm", elem_classes="bit-ghost-btn")
                         with gr.Row():
+                            sens_plot = gr.Plot()
+                            slip_plot = gr.Plot()
                     with gr.Tab("Report"):
                         report_md = gr.Markdown("_Run a backtest to generate the report._")
+                        report_equity = gr.Plot()
                         with gr.Row():
+                            save_run_btn = gr.Button("Save run to store", size="sm",
                                                      elem_classes="bit-ghost-btn")
                             copy_cfg_btn = gr.Button("Copy share link", size="sm",
                                                      elem_classes="bit-ghost-btn")
                         save_note = gr.HTML("")
                     with gr.Tab("Coverage"):
+                        coverage_kpis = gr.HTML("")
+                        coverage_table = gr.Dataframe(pd.DataFrame(), interactive=False,
+                                                      max_height=380,
+                                                      elem_classes="bit-table")
+                        gr.HTML('<div class="bit-zone-title">Extend coverage'
+                                '<span class="bit-micro">RUNS ON YOUR OWN GPU QUOTA'
+                                '</span></div>')
                         extend_panel = gr.HTML("")
                         with gr.Row():
+                            ext_model = gr.Dropdown(all_models, label="Model", scale=2)
+                            ext_asset = gr.Dropdown(list(config.ASSETS), label="Asset",
+                                                    scale=2)
+                            ext_tf = gr.Dropdown(all_tfs, value="1d",
+                                                 label="Timeframe", scale=1)
                         with gr.Row():
                             ext_start = gr.Textbox(label="Start (YYYY-MM-DD)", scale=2)
                             ext_end = gr.Textbox(label="End (YYYY-MM-DD)", scale=2)
                             extend_btn = gr.Button("Extend coverage", size="sm",
                                                    elem_classes="bit-run-btn")
                         extend_out = gr.HTML("")
+                        gr.HTML('<div class="bit-zone-title">Add model</div>')
                         with gr.Row():
                             add_family = gr.Dropdown(
+                                list(config.ALLOWED_ADAPTER_FAMILIES), value="chronos",
+                                label="Adapter family", scale=1)
                             add_model_id = gr.Textbox(label="HF model id (owner/name)",
                                                       scale=2)
                             add_btn = gr.Button("Smoke test & add", size="sm",
                                                 elem_classes="bit-ghost-btn", scale=1)
                         add_out = gr.HTML("")
+            # ------------------------- RIGHT -------------------------
+            with gr.Column(scale=2, min_width=240):
+                gr.HTML('<div class="bit-zone-title">Run Manager</div>')
                 try:
                     gr.LoginButton(value="Sign in with Hugging Face", size="sm")
                 except Exception:
+                    gr.HTML(C.note("Sign-in appears when this runs on a Space."))
+                history_html = gr.HTML(C.micro("no runs yet in this session"))
                 with gr.Accordion("METRICS GLOSSARY", open=False,
                                   elem_classes="bit-accordion"):
+                    gr.HTML(C.glossary(GLOSSARY))
                 with gr.Accordion("HOW RESULTS ARE COMPUTED", open=False,
                                   elem_classes="bit-accordion"):
+                    gr.Markdown(METHOD_NOTES)
+        gr.HTML(C.footer())
+        # ==================== wiring ====================
+        builder = [strategy, asset, timeframe, date_range, model_slug,
+                   p1, p2, p3, costs_on, commission_bps, slippage_bps,
+                   slippage_model, sizing_mode, size_pct, leverage,
+                   sl_pct, tp_pct, trail_pct,
+                   validation_mode, train_m, test_m, roll_m, holdout_m]
+        overview_out = [equity_plot, regime_plot, underwater_plot, rolling_plot,
+                        price_plot, pnl_plot, hold_plot, mae_plot, costs_note]
         def on_strategy_change(name):
             preset = strategies.PRESETS.get(name)
             if preset is None:
+                return gr.update(), gr.update(), gr.update(), gr.update(), ""
+            note = ("" if preset.available else
+                    C.note(f"<b>{name}</b> is unavailable. {preset.unavailable_reason}",
+                           danger=True))
             ups = []
             for i in range(3):
                 if i < len(preset.params):
+                    key, label, default, _lo, _hi = preset.params[i]
                     ups.append(gr.update(label=label, value=default, visible=True))
                 else:
                     ups.append(gr.update(visible=False, value=None))
+            models = runtime.available_models()
+            return (*ups, gr.update(visible=preset.needs_signals, choices=models,
+                                    value=(models[0] if models and preset.needs_signals
+                                           else None)), note)
+        strategy.change(on_strategy_change, [strategy],
+                        [p1, p2, p3, model_slug, preset_note])
         def on_universe_change(a, tf):
             cov = runtime.price_coverage_for(a, tf)
             models = runtime.available_models(a, tf)
             if cov is None:
+                html = C.note(f"No cached price coverage for {a} {tf}.", danger=True)
             else:
+                html = C.micro(f"cached {cov[0]} → {cov[1]} · "
+                               + (f"{len(models)} models" if models else "no signals"))
+            return html, gr.update(choices=models, value=(models[0] if models else None))
         asset.change(on_universe_change, [asset, timeframe], [coverage_note, model_slug])
+        timeframe.change(on_universe_change, [asset, timeframe],
+                         [coverage_note, model_slug])
         def do_run(hist, *vals, progress=gr.Progress()):
             progress(0.05, desc="Reading cached slices")
             req = collect_request(*vals)
             try:
+                progress(0.4, desc="Simulating trades")
                 rec = runtime.execute(req)
             except (RunError, ValueError) as e:
+                return (hist, None, C.top_bar("RUN FAILED", "warn"),
+                        C.note(str(e), danger=True), gr.update(visible=True),
+                        *(gr.update(),) * 9, "", pd.DataFrame(),
+                        C.micro("no runs yet in this session"),
                         "_Run failed._", None)
+            progress(0.8, desc="Building charts")
             hist = ([rec] + list(hist))[:40]
+            return (hist, rec,
+                    C.top_bar(f"RUN {rec.run_id} COMPLETE", "ok",
+                              context=context_chip_text(req),
+                              elapsed=f"{rec.elapsed_s:.1f}s"),
+                    C.stat_band(rec), gr.update(visible=False),
+                    *build_overview(rec, log_scale=False, cvd=False),
+                    trades_head(rec), trades_frame(rec),
+                    C.run_history(hist), report_markdown(rec),
+                    charts.equity_curve(rec.result.equity, rec.result.benchmark_equity,
+                                        plan=rec.result.plan))
+        run_out = [history, current, top_bar, stat_band, empty_state, *overview_out,
+                   trades_head_html, trades_table, history_html, report_md,
+                   report_equity]
+        run_btn.click(do_run, [history, *builder], run_out)
         def load_example():
+            return ("Chronos Forecast Follower", "BTC-USD", "1d", "3Y",
+                    gr.update(value=0.005, label="Entry edge", visible=True),
+                    gr.update(value=0.0, label="Exit edge", visible=True),
+                    gr.update(visible=False, value=None),
+                    gr.update(visible=True, choices=runtime.available_models(),
+                              value="chronos-bolt-small"),
                     True, "Walk-forward")
         example_btn.click(
             load_example, None,
+            [strategy, asset, timeframe, date_range, p1, p2, p3, model_slug,
+             costs_on, validation_mode],
+        ).then(do_run, [history, *builder], run_out)
         def replot(rec, log_s, colorblind):
             if rec is None:
                 return (gr.update(),) * 9
             return build_overview(rec, log_scale=log_s, cvd=colorblind)
+        log_scale.change(replot, [current, log_scale, cvd], overview_out)
+        cvd.change(replot, [current, log_scale, cvd], overview_out)
+        share_btn.click(lambda *v: gr.update(value=collect_request(*v).encode(),
+                                             visible=True), builder, [share_out])
         copy_cfg_btn.click(lambda r: gr.update(value=r.request.encode(), visible=True)
                            if r else gr.update(), [current], [share_out])
         def do_save(rec):
             if rec is None:
+                return C.note("Nothing to save yet.")
             try:
+                rid = runtime.save_run_summary(rec, push=True)
+                return C.note(f"Run <b>{rid}</b> saved to the signal store. "
+                              "It now appears in Run history for everyone.")
             except Exception as e:
+                return C.note(f"Could not save: {e}", danger=True)
         save_run_btn.click(do_save, [current], [save_note])
         def do_robustness(rec):
             if rec is None:
                 e = charts.empty_figure("run a backtest first")
+                return "", e, e
             grade, checks = runtime.overfit_verdict(rec)
+            items = "".join(f'<div class="bit-gloss-def">{m} {t}</div>'
+                            for m, t in checks)
+            html = (f'<div class="bit-panel">{C.panel_head("Overfit verdict", grade)}'
                     f'{items}</div>')
+            return (html, charts.walk_forward_bars(rec.result.windows),
+                    charts.monte_carlo_cone(charts.monte_carlo_paths(rec.result.trades)))
+        current.change(do_robustness, [current], [verdict_html, wf_plot, mc_plot])
         def do_sweep(rec, progress=gr.Progress()):
             if rec is None:
                 sens = charts.empty_figure("this preset has fewer than two parameters")
             else:
                 progress(0.1, desc="Parameter sweep")
+                bx = req.params.get(keys[0], 20)
+                by = req.params.get(keys[1], 50)
+                xs = sorted({max(2, int(bx * m)) for m in (0.5, 0.75, 1.0, 1.5, 2.0)})
+                ys = sorted({max(3, int(by * m)) for m in (0.5, 0.75, 1.0, 1.5, 2.0)})
+                sens = charts.parameter_sensitivity(
+                    runtime.parameter_sweep(req, keys[0], xs, keys[1], ys),
+                    x=keys[0], y=keys[1])
             progress(0.7, desc="Slippage stress")
+            return sens, charts.slippage_stress(runtime.slippage_stress(req))
         robust_btn.click(do_sweep, [current], [sens_plot, slip_plot])
+        # ---- Compare tab ----
+        lb_filters = [f_assets, f_tfs, f_strats, f_models, f_metric,
+                      f_min_trades, f_hide_base, f_oos, f_top]
+        lb_out = [podium, lb_table, lb_overlay, lb_scatter, lb_meta]
+        def refresh_leaderboard(assets_, tfs, strats, models, metric,
+                                min_trades, hide_base, req_oos, top_n):
+            return CT.build_leaderboard_view(
+                runtime.get_store(), assets=assets_, timeframes=tfs,
+                strategies_=strats, models=models, metric_label=metric,
+                min_trades=min_trades, hide_baselines=hide_base,
+                require_oos=req_oos, top_n=top_n)
+        for ctrl in lb_filters:
+            ctrl.change(refresh_leaderboard, lb_filters, lb_out)
+        reset_btn.click(lambda: ([], [], [], [], "OOS Sharpe", 0, False, True, 15),
+                        None, lb_filters).then(refresh_leaderboard, lb_filters, lb_out)
+        models_out = [models_note, acc_plot, cal_plot, model_bars, score_table]
+        models_tf.change(lambda tf: CT.build_models_view(runtime.get_store(), tf),
+                         [models_tf], models_out)
+        def refresh_signals(a, tf):
+            return CT.build_signals_view(runtime.get_store(), a, tf)
+        sig_asset.change(refresh_signals, [sig_asset, sig_tf], [sig_panel])
+        sig_tf.change(refresh_signals, [sig_asset, sig_tf], [sig_panel])
+        def refresh_runs(hist):
+            saved = catalog.load_saved_runs(runtime.get_store())
+            return CT.runs_table(hist, saved)
+        runs_refresh.click(refresh_runs, [history], [runs_table])
+        history.change(refresh_runs, [history], [runs_table])
+        # ---- Coverage / extension ----
+        estimate_btn.click(extension.estimate_ui,
+                           [ext_model, ext_asset, ext_tf, ext_start, ext_end],
+                           [extend_out])
         extend_btn.click(extension.extend_ui,
                          [ext_model, ext_asset, ext_tf, ext_start, ext_end],
                          [extend_out, coverage_table])
         add_btn.click(extension.add_model_ui, [add_family, add_model_id],
                       [add_out, coverage_table])
+        # ---- Load ----
         def on_load(request: gr.Request):
+            st = runtime.get_store()
+            meta = catalog.catalog_meta(st)
+            meta_html = C.micro(
+                f"{meta.get('leaderboard_rows', 0)} combinations · "
+                f"{meta.get('scorecard_rows', 0)} model slices · generated "
+                f"{str(meta.get('generated_at', ''))[:16].replace('T', ' ')} · "
+                f"{meta.get('canonical_config', '')}") if meta else C.micro("catalog not built")
+            lb_view = CT.build_leaderboard_view(
+                st, assets=None, timeframes=None, strategies_=None, models=None,
+                metric_label="OOS Sharpe", min_trades=0, hide_baselines=False,
+                require_oos=True, top_n=15)
+            models_view = CT.build_models_view(st, "1d")
+            sig = CT.build_signals_view(st, "BTC-USD", "1d")
+            cov_cells = runtime.coverage_map()
             restored = [gr.update()] * 4
             try:
                 token = dict(request.query_params).get("cfg") if request else None
             except Exception:
                     restored = [gr.update(value=req.strategy), gr.update(value=req.asset),
                                 gr.update(value=req.timeframe),
                                 gr.update(value=req.date_range)]
                 except Exception as e:
+                    log.warning("share link rejected: %s", e)
+            return (meta_html, *lb_view, *models_view, sig,
+                    C.coverage_summary(cov_cells), runtime.coverage_frame(),
+                    extension.status_html(),
+                    CT.runs_table([], catalog.load_saved_runs(st)), *restored)
         demo.load(on_load, None,
+                  [catalog_meta, *lb_out, *models_out, sig_panel,
+                   coverage_kpis, coverage_table, extend_panel, runs_table,
                    strategy, asset, timeframe, date_range])
         demo.load(lambda: on_universe_change("BTC-USD", "1d"), None,
                   [coverage_note, model_slug])
     return demo
 demo = build_app()
 if __name__ == "__main__":
     gr.set_static_paths(paths=theme.static_paths())
     demo.queue(max_size=32).launch(
+        server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 7860)),
+        show_api=False)

scripts/seed_store.py CHANGED Viewed

@@ -28,7 +28,7 @@ import pandas as pd
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
-from src import comparisons, config  # noqa: E402
 from src.adapters import build_windows, get_adapter  # noqa: E402
 from src.data import refresh  # noqa: E402
 from src.store import SignalStore  # noqa: E402
@@ -61,32 +61,37 @@ EQUITIES = ["SPY", "QQQ", "NVDA"]
 def plan_v1() -> list[SeedTarget]:
-    """The v1 seed: crypto-first, honest about provider depth limits.
-    Daily coverage is real for every asset on both Chronos models. Hourly is
-    real for crypto on the small model. 15-minute coverage is placeholder-only
-    in v1 -- the inference cost is large and the provider depth for equities is
-    60 days, so it is labelled rather than faked as real.
     """
     targets: list[SeedTarget] = []
-    for asset in CRYPTO + EQUITIES:
-        targets.append(SeedTarget("chronos-bolt-small", asset, "1d", 3.0))
-    for asset in CRYPTO:
-        targets.append(SeedTarget("chronos-bolt-base", asset, "1d", 3.0))
-    # Hourly: crypto has full depth from the exchange; equities are capped at
-    # the provider's ~730 days, recorded as a boundary in the manifest.
-    for asset in CRYPTO:
-        targets.append(SeedTarget("chronos-bolt-small", asset, "1h", 1.0))
     for asset in EQUITIES:
         targets.append(SeedTarget("chronos-bolt-small", asset, "1h", 1.5))
-    # 15-minute crypto. Batched Chronos-Bolt inference turned out to cost about
-    # a millisecond per step on this hardware, so these are real rather than
-    # placeholder -- the v1 seed ships with no synthetic slices at all.
-    for asset in CRYPTO:
-        targets.append(SeedTarget("chronos-bolt-small", asset, "15m", 0.25))
     return targets
@@ -301,6 +306,11 @@ def main(argv=None) -> int:
         )
         log.info("%s", results[-1])
     if args.push and not args.offline:
         log.info("== pushing to %s ==", args.repo)
         oid = store.flush(f"Seed store ({args.plan})")

 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+from src import catalog, comparisons, config  # noqa: E402
 from src.adapters import build_windows, get_adapter  # noqa: E402
 from src.data import refresh  # noqa: E402
 from src.store import SignalStore  # noqa: E402
 def plan_v1() -> list[SeedTarget]:
+    """The v1 seed: every seedable model on daily bars for the whole universe,
+    plus intraday coverage for the fast Chronos-Bolt family on crypto.
+    Daily is the comparison backbone -- every model sees exactly the same bars
+    on every asset, so leaderboard differences are the model, not the coverage.
+    Intraday is added where inference is cheap enough to be honest about.
     """
     targets: list[SeedTarget] = []
+    all_assets = CRYPTO + EQUITIES
+    # Backbone: every model x every asset, daily.
+    for model in config.SEEDABLE_MODELS:
+        for asset in all_assets:
+            targets.append(SeedTarget(model, asset, "1d", 3.0))
+    # Intraday: the bolt family plus baselines on crypto.
+    intraday_models = [m for m in config.SEEDABLE_MODELS
+                       if m.startswith("chronos-bolt") or m.startswith("baseline")]
+    for model in intraday_models:
+        for asset in CRYPTO:
+            targets.append(SeedTarget(model, asset, "1h", 1.0))
+    # 15-minute: the small model and the naive baseline, crypto only.
+    for model in ("chronos-bolt-small", "baseline-naive"):
+        for asset in CRYPTO:
+            targets.append(SeedTarget(model, asset, "15m", 0.25))
+    # Hourly equities for the reference model, capped by provider depth.
     for asset in EQUITIES:
         targets.append(SeedTarget("chronos-bolt-small", asset, "1h", 1.5))
     return targets
         )
         log.info("%s", results[-1])
+        log.info("== building catalog ==")
+        cat = catalog.build(store)
+        results.append(cat.summary())
+        log.info("%s", cat.summary())
     if args.push and not args.offline:
         log.info("== pushing to %s ==", args.repo)
         oid = store.flush(f"Seed store ({args.plan})")

src/adapters.py CHANGED Viewed

@@ -132,6 +132,19 @@ def _default_device() -> str:
     return "cpu"
 # --------------------------------------------------------------------------
 # Chronos / Chronos-Bolt
 # --------------------------------------------------------------------------
@@ -170,6 +183,24 @@ class ChronosAdapter(ForecastAdapter):
         )
         return self
     def predict(self, context_windows: np.ndarray) -> Forecast:
         if self._model is None:
             self.load()
@@ -178,15 +209,24 @@ class ChronosAdapter(ForecastAdapter):
         ctx = np.asarray(context_windows, dtype="float32")
         if ctx.ndim == 1:
             ctx = ctx[None, :]
-        tensors = [torch.tensor(row) for row in ctx]
         q_levels = list(DEFAULT_QUANTILES)
-        with torch.inference_mode():
-            quantiles, _mean = self._model.predict_quantiles(
-                tensors, prediction_length=1, quantile_levels=q_levels,
-            )
-        # (n_series, prediction_length, n_quantiles) -> horizon 1
-        arr = quantiles.float().cpu().numpy()[:, 0, :]
         return Forecast(q10=arr[:, 0], q50=arr[:, 1], q90=arr[:, 2],
                         context_len=ctx.shape[1])
@@ -254,6 +294,82 @@ class TimesFMAdapter(ForecastAdapter):
                         context_len=ctx.shape[1])
 # --------------------------------------------------------------------------
 # Placeholder (no GPU / no model available)
 # --------------------------------------------------------------------------
@@ -308,6 +424,7 @@ class PlaceholderAdapter(ForecastAdapter):
 _FAMILIES: dict[str, type[ForecastAdapter]] = {
     "chronos": ChronosAdapter,
     "timesfm": TimesFMAdapter,
     "placeholder": PlaceholderAdapter,
 }

     return "cpu"
+def _release(device: str) -> None:
+    """Free accelerator memory between inference chunks."""
+    try:
+        import torch
+        if device == "cuda":
+            torch.cuda.empty_cache()
+        elif device == "mps" and hasattr(torch, "mps"):
+            torch.mps.empty_cache()
+    except Exception:
+        pass
 # --------------------------------------------------------------------------
 # Chronos / Chronos-Bolt
 # --------------------------------------------------------------------------
         )
         return self
+    # Bolt emits quantiles directly and is happy with wide batches. The
+    # original T5 Chronos *samples* forecast paths instead, so its peak memory
+    # is batch x num_samples x context and a wide batch OOMs a GPU outright.
+    # Chunking here rather than at the call site means every caller -- seed
+    # script, ZeroGPU function, tests -- is protected without knowing which
+    # variant it holds.
+    BOLT_CHUNK = 256
+    T5_CHUNK = 16
+    T5_NUM_SAMPLES = 20
+    @property
+    def _is_bolt(self) -> bool:
+        return "bolt" in (self.model_id or "").lower()
+    @property
+    def chunk_size(self) -> int:
+        return self.BOLT_CHUNK if self._is_bolt else self.T5_CHUNK
     def predict(self, context_windows: np.ndarray) -> Forecast:
         if self._model is None:
             self.load()
         ctx = np.asarray(context_windows, dtype="float32")
         if ctx.ndim == 1:
             ctx = ctx[None, :]
         q_levels = list(DEFAULT_QUANTILES)
+        extra = {} if self._is_bolt else {"num_samples": self.T5_NUM_SAMPLES}
+        step = max(1, self.chunk_size)
+        parts = []
+        for start in range(0, len(ctx), step):
+            block = ctx[start:start + step]
+            tensors = [torch.tensor(row) for row in block]
+            with torch.inference_mode():
+                quantiles, _mean = self._model.predict_quantiles(
+                    tensors, prediction_length=1, quantile_levels=q_levels, **extra,
+                )
+            parts.append(quantiles.float().cpu().numpy()[:, 0, :])
+            del quantiles, tensors
+            _release(self.device)
+        arr = np.vstack(parts)
         return Forecast(q10=arr[:, 0], q50=arr[:, 1], q90=arr[:, 2],
                         context_len=ctx.shape[1])
                         context_len=ctx.shape[1])
+# --------------------------------------------------------------------------
+# Naive baselines
+# --------------------------------------------------------------------------
+class BaselineAdapter(ForecastAdapter):
+    """Classical forecasting baselines, exposed as first-class models.
+    These exist so the leaderboard always carries a floor. A large pretrained
+    forecaster that cannot beat "tomorrow looks like today" has not earned its
+    inference cost, and burying that comparison would be the single easiest way
+    to make this whole Space misleading.
+    The method is selected by the model id: `baseline/naive`, `baseline/drift`,
+    or `baseline/seasonal`. All three are deterministic and causal -- each uses
+    only the trailing window, which ends at the bar being forecast.
+    The interval is a Gaussian band around the point forecast, scaled by the
+    window's own realised volatility, so coverage is comparable with a
+    learned model's q10-q90 rather than arbitrarily wide.
+    """
+    family = "baseline"
+    adapter_version = "1"
+    Z90 = 1.2815515655446004  # normal 90th percentile
+    def __init__(self, model_id: str, revision: str | None = None,
+                 context_len: int = 128, device: str | None = None,
+                 season: int = 7):
+        super().__init__(model_id, revision, context_len, device)
+        self.season = season
+        self.method = (model_id.split("/")[-1] or "naive").lower()
+        if self.method not in ("naive", "drift", "seasonal"):
+            raise AdapterError(
+                f"unknown baseline method {self.method!r}; "
+                "use baseline/naive, baseline/drift or baseline/seasonal"
+            )
+    def load(self, model_id: str | None = None, revision: str | None = None):
+        if model_id:
+            self.model_id = model_id
+            self.method = model_id.split("/")[-1].lower()
+        # Baselines have no weights, so the "revision" is the adapter's own
+        # version -- still pinned, still reproducible.
+        self._resolved_revision = f"baseline-{self.adapter_version}"
+        return self
+    def resolve_revision(self) -> str:
+        self._resolved_revision = f"baseline-{self.adapter_version}"
+        return self._resolved_revision
+    def predict(self, context_windows: np.ndarray) -> Forecast:
+        ctx = np.asarray(context_windows, dtype="float64")
+        if ctx.ndim == 1:
+            ctx = ctx[None, :]
+        last = ctx[:, -1]
+        if self.method == "naive":
+            point = last
+        elif self.method == "drift":
+            # Average per-step change across the window, added once.
+            steps = ctx.shape[1] - 1
+            point = last + (ctx[:, -1] - ctx[:, 0]) / max(steps, 1)
+        else:  # seasonal
+            k = min(self.season, ctx.shape[1])
+            point = ctx[:, -k]
+        # One-step-ahead uncertainty from the window's own step volatility.
+        sigma = np.std(np.diff(ctx, axis=1), axis=1)
+        sigma = np.where(np.isfinite(sigma) & (sigma > 0), sigma,
+                         np.abs(last) * 1e-4)
+        band = self.Z90 * sigma
+        return Forecast(q10=point - band, q50=point, q90=point + band,
+                        context_len=ctx.shape[1])
 # --------------------------------------------------------------------------
 # Placeholder (no GPU / no model available)
 # --------------------------------------------------------------------------
 _FAMILIES: dict[str, type[ForecastAdapter]] = {
     "chronos": ChronosAdapter,
     "timesfm": TimesFMAdapter,
+    "baseline": BaselineAdapter,
     "placeholder": PlaceholderAdapter,
 }

src/catalog.py ADDED Viewed

	@@ -0,0 +1,521 @@

+"""The catalog: everything this Space has ever computed, in one place.
+The Comparison tab used to see only the runs made in the current browser
+session. That cannot answer the question people actually have -- *which
+strategy, on which model, on which asset, has ever worked best?* -- because the
+answer lives across sessions, users and machines.
+So the catalog is **precomputed into the store** rather than recomputed live:
+* `comparisons/leaderboard.parquet` -- one row per
+  (strategy x model x asset x timeframe) with the full IS / OOS / holdout metric
+  set, all produced under one fixed canonical config so rows are comparable.
+* `comparisons/equity_curves.parquet` -- long-format cumulative returns,
+  resampled to daily, so "returns over time" for dozens of algorithms renders
+  instantly instead of re-running dozens of backtests per page load.
+* `comparisons/signal_scorecard.parquet` -- per-model forecast quality:
+  calibration, directional accuracy against naive baselines, pinball loss.
+All three are regenerated after any coverage extension, so the catalog never
+describes a store that no longer exists. Session runs and saved `runs/*.json`
+are merged on top at read time.
+"""
+from __future__ import annotations
+import json
+import logging
+import time
+from dataclasses import dataclass, field
+import numpy as np
+import pandas as pd
+from . import config, strategies
+from .engine import BacktestConfig, Costs, Validation, run_backtest
+from .metrics import (
+    calibration_coverage,
+    calibration_error,
+    directional_accuracy,
+    pinball_loss,
+)
+from .store import SignalStore
+log = logging.getLogger("bit.catalog")
+LEADERBOARD = "comparisons/leaderboard.parquet"
+EQUITY_CURVES = "comparisons/equity_curves.parquet"
+SCORECARD = "comparisons/signal_scorecard.parquet"
+CATALOG_INDEX = "comparisons/catalog.json"
+# Strategies that take part in the catalog sweep. `Buy & Hold` is included
+# deliberately: a strategy that cannot beat holding the asset has not earned
+# its complexity, and the leaderboard should show that side by side.
+CATALOG_STRATEGIES = (
+    "Buy & Hold (benchmark)",
+    "SMA Crossover",
+    "RSI Mean Reversion",
+    "Bollinger Breakout",
+    "MACD Momentum",
+    "Sentiment-Gated Momentum",
+    "Chronos Forecast Follower",
+)
+# Every catalog row is produced under exactly this config. Changing it
+# invalidates every published comparison, which is why it lives in one place.
+CANONICAL = dict(
+    costs=Costs(enabled=True),
+    validation=Validation(mode="walk_forward", train_months=12, test_months=3,
+                          roll_months=3, holdout_months=6),
+)
+CURVE_POINTS_MAX = 800
+# Below this many closed trades, a Sharpe ratio is noise dressed as a result.
+# The leaderboard still shows these rows -- hiding them would be its own kind of
+# lie -- but flags them and filters them out of the default ranked view.
+MIN_MEANINGFUL_TRADES = 20
+def canonical_config(asset: str, timeframe: str, strategy: str) -> BacktestConfig:
+    return BacktestConfig(
+        asset=asset, timeframe=timeframe, strategy=strategy,
+        params=strategies.defaults_for(strategy), **CANONICAL,
+    )
+def entry_key(strategy: str, model_slug: str, asset: str, timeframe: str) -> str:
+    return f"{strategy}|{model_slug or '-'}|{asset}|{timeframe}"
+# --------------------------------------------------------------------------
+# Building
+# --------------------------------------------------------------------------
+@dataclass
+class BuildReport:
+    rows: int = 0
+    curves: int = 0
+    scorecard_rows: int = 0
+    skipped: list[str] = field(default_factory=list)
+    failed: list[str] = field(default_factory=list)
+    elapsed_s: float = 0.0
+    def summary(self) -> str:
+        return (f"catalog: {self.rows} leaderboard rows, {self.curves} curves, "
+                f"{self.scorecard_rows} scorecard rows, {len(self.skipped)} skipped, "
+                f"{len(self.failed)} failed in {self.elapsed_s:.1f}s")
+def _resample_curve(equity: pd.Series) -> pd.DataFrame:
+    """Normalised cumulative return, thinned to a comparable daily series."""
+    if equity is None or len(equity) < 2:
+        return pd.DataFrame(columns=["ts", "cum_return"])
+    cum = equity / float(equity.iloc[0]) - 1.0
+    # Daily is the common denominator across 15m / 1h / 1d runs, so every
+    # series in the overlay shares an x-axis granularity.
+    daily = cum.resample("1D").last().dropna()
+    if len(daily) > CURVE_POINTS_MAX:
+        step = int(np.ceil(len(daily) / CURVE_POINTS_MAX))
+        daily = daily.iloc[::step]
+    return pd.DataFrame({"ts": daily.index, "cum_return": daily.to_numpy()})
+def _scorecard_row(prices: pd.DataFrame, signals: pd.DataFrame,
+                   model_slug: str, asset: str, timeframe: str) -> dict | None:
+    """Forecast quality for one model slice, independent of any trading rule."""
+    if signals.empty or prices.empty:
+        return None
+    close = prices["close"]
+    aligned = signals.reindex(close.index).dropna(subset=["q50"])
+    if aligned.empty:
+        return None
+    actual_next = close.shift(-1).reindex(aligned.index)
+    valid = actual_next.notna()
+    if valid.sum() < 30:
+        return None
+    idx = valid.index[valid]
+    actual_next = actual_next[valid]
+    q10, q50, q90 = aligned.loc[idx, "q10"], aligned.loc[idx, "q50"], aligned.loc[idx, "q90"]
+    ref = close.reindex(idx)
+    prev_move = close.diff().reindex(idx).fillna(0.0)
+    rng = np.random.default_rng(0)
+    coin = pd.Series(rng.choice([-1.0, 1.0], size=len(idx)), index=idx)
+    cov = calibration_coverage(actual_next, q10, q90)
+    acc = directional_accuracy(actual_next, q50, ref)
+    mom = directional_accuracy(actual_next, ref + prev_move, ref)
+    rnd = directional_accuracy(actual_next, ref + coin, ref)
+    spec = config.SEED_MODELS.get(model_slug)
+    return {
+        "model_slug": model_slug,
+        "model_display": spec.display if spec else model_slug,
+        "family": spec.family if spec else "unknown",
+        "is_baseline": bool(spec and spec.family == "baseline"),
+        "asset": asset, "timeframe": timeframe, "n": int(valid.sum()),
+        "coverage_q10_q90": cov,
+        "calibration_error": calibration_error(cov, 0.80),
+        "directional_accuracy": acc,
+        "beats_momentum": (acc - mom) if pd.notna(acc) and pd.notna(mom) else float("nan"),
+        "beats_random": (acc - rnd) if pd.notna(acc) and pd.notna(rnd) else float("nan"),
+        "baseline_momentum": mom, "baseline_random": rnd,
+        "pinball_q50": pinball_loss(actual_next, q50, 0.50),
+        "band_width_pct": float(((q90 - q10) / ref.abs().clip(lower=1e-9)).mean()),
+    }
+def build(store: SignalStore, *, write: bool = True,
+          strategies_subset: tuple[str, ...] = CATALOG_STRATEGIES,
+          progress=None) -> BuildReport:
+    """Sweep every covered combination and write the catalog artifacts."""
+    t0 = time.perf_counter()
+    rep = BuildReport()
+    manifest = store.load_manifest()
+    price_keys = sorted(manifest.prices)
+    rows: list[dict] = []
+    curves: list[pd.DataFrame] = []
+    scorecard: list[dict] = []
+    total = len(price_keys)
+    for n, pk in enumerate(price_keys):
+        asset, timeframe = pk.split("|")
+        if progress:
+            progress((n + 1) / max(total, 1), desc=f"catalog {asset} {timeframe}")
+        prices = store.get_prices(asset, timeframe)
+        if len(prices) < 120:
+            rep.skipped.append(f"{pk}: only {len(prices)} bars")
+            continue
+        bpy = config.bars_per_year(asset, timeframe)
+        models = sorted({e.model_slug for e in
+                         manifest.find_signals(asset=asset, timeframe=timeframe)})
+        signal_cache: dict[str, pd.DataFrame] = {}
+        for m in models:
+            sig = store.get_signals(m, asset, timeframe)
+            signal_cache[m] = sig
+            sc = _scorecard_row(prices, sig, m, asset, timeframe)
+            if sc:
+                scorecard.append(sc)
+        for strategy in strategies_subset:
+            preset = strategies.PRESETS.get(strategy)
+            if preset is None or not preset.available:
+                continue
+            # Signal strategies fan out over models; the rest run once.
+            targets = models if preset.needs_signals else [""]
+            if preset.needs_signals and not models:
+                rep.skipped.append(f"{strategy}|{pk}: no model coverage")
+                continue
+            for model_slug in targets:
+                key = entry_key(strategy, model_slug, asset, timeframe)
+                try:
+                    cfg = canonical_config(asset, timeframe, strategy)
+                    out = strategies.build(strategy, prices, cfg.params,
+                                           signal_cache.get(model_slug, pd.DataFrame()))
+                    res = run_backtest(prices, out, cfg, bars_per_year=bpy)
+                except Exception as e:
+                    log.warning("catalog cell failed %s: %s", key, e)
+                    rep.failed.append(f"{key}: {type(e).__name__}")
+                    continue
+                a, i, o = res.metrics_all, res.metrics_is, res.metrics_oos
+                h = res.metrics_holdout
+                spec = config.SEED_MODELS.get(model_slug)
+                bench = (float(res.benchmark_equity.iloc[-1] /
+                               res.benchmark_equity.iloc[0] - 1.0)
+                         if len(res.benchmark_equity) else float("nan"))
+                rows.append({
+                    "key": key,
+                    "strategy": strategy,
+                    "model_slug": model_slug,
+                    "model_display": spec.display if spec else ("—" if not model_slug else model_slug),
+                    "is_baseline_model": bool(spec and spec.family == "baseline"),
+                    "asset": asset, "timeframe": timeframe,
+                    "bars": a.bars, "trades": a.trade_count,
+                    "total_return": a.total_return, "cagr": a.cagr,
+                    "sharpe": a.sharpe, "sortino": a.sortino,
+                    "max_drawdown": a.max_drawdown, "volatility": a.volatility,
+                    "win_rate": a.win_rate, "profit_factor": a.profit_factor,
+                    "exposure": a.exposure, "costs_paid": res.costs_paid,
+                    "is_sharpe": i.sharpe if i.bars else float("nan"),
+                    "oos_sharpe": o.sharpe if o.bars else float("nan"),
+                    "oos_return": o.total_return if o.bars else float("nan"),
+                    "oos_max_drawdown": o.max_drawdown if o.bars else float("nan"),
+                    "oos_trades": o.trade_count if o.bars else 0,
+                    "has_oos": bool(o.bars),
+                    "holdout_sharpe": h.sharpe if h else float("nan"),
+                    "holdout_return": h.total_return if h else float("nan"),
+                    "has_holdout": h is not None,
+                    "oos_is_ratio": (o.sharpe / i.sharpe) if (i.bars and i.sharpe) else float("nan"),
+                    "benchmark_return": bench,
+                    "excess_vs_hold": a.total_return - bench if pd.notna(bench) else float("nan"),
+                    "significant": bool(a.trade_count >= MIN_MEANINGFUL_TRADES),
+                    "wf_windows": len(res.windows),
+                    "wf_positive": sum(1 for w in res.windows if w.metrics.total_return > 0),
+                    "generated_at": pd.Timestamp.now(tz="UTC").isoformat(),
+                })
+                curve = _resample_curve(res.equity)
+                if not curve.empty:
+                    curve["key"] = key
+                    curve["strategy"] = strategy
+                    curve["asset"] = asset
+                    curve["timeframe"] = timeframe
+                    curve["model_slug"] = model_slug
+                    curves.append(curve)
+    lb = pd.DataFrame(rows)
+    ec = pd.concat(curves, ignore_index=True) if curves else pd.DataFrame(
+        columns=["ts", "cum_return", "key", "strategy", "asset", "timeframe", "model_slug"])
+    sc = pd.DataFrame(scorecard)
+    rep.rows, rep.curves, rep.scorecard_rows = len(lb), len(curves), len(sc)
+    rep.elapsed_s = time.perf_counter() - t0
+    if write:
+        if not lb.empty:
+            store.write_table(LEADERBOARD, lb)
+        if not ec.empty:
+            store.write_table(EQUITY_CURVES, ec)
+        if not sc.empty:
+            store.write_table(SCORECARD, sc)
+        store.write_json(CATALOG_INDEX, {
+            "generated_at": pd.Timestamp.now(tz="UTC").isoformat(),
+            "leaderboard_rows": len(lb),
+            "equity_curve_series": len(curves),
+            "scorecard_rows": len(sc),
+            "strategies": list(strategies_subset),
+            "canonical_config": "costs on; walk-forward 12/3/3; 6mo locked holdout",
+            "skipped": rep.skipped[:50],
+            "failed": rep.failed[:50],
+        })
+    return rep
+# --------------------------------------------------------------------------
+# Reading
+# --------------------------------------------------------------------------
+def _read(store: SignalStore, path: str) -> pd.DataFrame:
+    df = store.read_parquet(path)
+    return df if df is not None else pd.DataFrame()
+def load_leaderboard(store: SignalStore) -> pd.DataFrame:
+    return _read(store, LEADERBOARD)
+def load_equity_curves(store: SignalStore) -> pd.DataFrame:
+    df = _read(store, EQUITY_CURVES)
+    if not df.empty:
+        df["ts"] = pd.to_datetime(df["ts"], utc=True)
+    return df
+def load_scorecard(store: SignalStore) -> pd.DataFrame:
+    return _read(store, SCORECARD)
+def catalog_meta(store: SignalStore) -> dict:
+    p = store._fetch(CATALOG_INDEX)
+    if p is None:
+        return {}
+    try:
+        return json.loads(p.read_text())
+    except Exception:
+        return {}
+def load_saved_runs(store: SignalStore) -> pd.DataFrame:
+    """Runs persisted to `runs/` -- across every session and every user."""
+    rows = []
+    try:
+        from huggingface_hub import HfApi
+        if store.offline:
+            import pathlib
+            files = [f"runs/{p.name}" for p in
+                     (store.local_root / "runs").glob("*.json")] \
+                if (store.local_root / "runs").exists() else []
+        else:
+            api = HfApi(token=store.token)
+            files = [f for f in api.list_repo_files(
+                store.repo_id, repo_type=config.STORE_REPO_TYPE)
+                if f.startswith("runs/") and f.endswith(".json")]
+    except Exception as e:
+        log.warning("could not list saved runs: %s", e)
+        return pd.DataFrame()
+    for f in files[:500]:
+        p = store._fetch(f)
+        if p is None:
+            continue
+        try:
+            d = json.loads(p.read_text())
+        except Exception:
+            continue
+        cfg = d.get("config", {})
+        m = (d.get("metrics") or {})
+        allm, oos = m.get("all") or {}, m.get("oos") or {}
+        rows.append({
+            "run_id": d.get("run_id", ""), "label": d.get("label", ""),
+            "created_at": d.get("created_at", ""),
+            "strategy": cfg.get("strategy", ""), "asset": cfg.get("asset", ""),
+            "timeframe": cfg.get("timeframe", ""),
+            "model_slug": cfg.get("model_slug", ""),
+            "total_return": allm.get("total_return", float("nan")),
+            "sharpe": allm.get("sharpe", float("nan")),
+            "oos_sharpe": oos.get("sharpe", float("nan")),
+            "max_drawdown": allm.get("max_drawdown", float("nan")),
+            "trades": allm.get("trade_count", 0),
+            "costs_paid": allm.get("costs_paid", float("nan")),
+            "share_token": d.get("share_token", ""),
+            "source": "saved run",
+        })
+    return pd.DataFrame(rows)
+# --------------------------------------------------------------------------
+# Views the UI asks for
+# --------------------------------------------------------------------------
+def rank(df: pd.DataFrame, by: str = "oos_sharpe", top: int | None = None,
+         ascending: bool = False, significant_only: bool = False) -> pd.DataFrame:
+    """Sort the leaderboard.
+    `significant_only` drops rows with too few trades to mean anything. A
+    Sharpe of 4 on 13 trades outranks everything real if you let it, so the
+    default ranked view uses this and says so.
+    """
+    if df.empty or by not in df.columns:
+        return df
+    out = df
+    if significant_only and "significant" in out.columns:
+        out = out[out["significant"]]
+    out = out.sort_values(by, ascending=ascending, na_position="last")
+    return out.head(top) if top else out
+def filter_leaderboard(df: pd.DataFrame, *, assets=None, timeframes=None,
+                       strategies_=None, models=None, min_trades: int = 0,
+                       require_oos: bool = False,
+                       hide_baselines: bool = False) -> pd.DataFrame:
+    if df.empty:
+        return df
+    out = df
+    if assets:
+        out = out[out["asset"].isin(assets)]
+    if timeframes:
+        out = out[out["timeframe"].isin(timeframes)]
+    if strategies_:
+        out = out[out["strategy"].isin(strategies_)]
+    if models:
+        out = out[out["model_slug"].isin(models) | (out["model_slug"] == "")]
+    if min_trades:
+        out = out[out["trades"] >= min_trades]
+    if require_oos and "has_oos" in out.columns:
+        out = out[out["has_oos"]]
+    if hide_baselines and "is_baseline_model" in out.columns:
+        out = out[~out["is_baseline_model"]]
+    return out
+def curves_for(curves: pd.DataFrame, keys: list[str]) -> dict[str, pd.Series]:
+    """Pull named cumulative-return series out of the stored long frame."""
+    if curves.empty or not keys:
+        return {}
+    sub = curves[curves["key"].isin(keys)]
+    out: dict[str, pd.Series] = {}
+    for key, g in sub.groupby("key"):
+        g = g.sort_values("ts")
+        out[str(key)] = pd.Series(g["cum_return"].to_numpy(),
+                                  index=pd.DatetimeIndex(g["ts"]))
+    return out
+def model_consensus(scorecard: pd.DataFrame, store: SignalStore,
+                    asset: str, timeframe: str) -> pd.DataFrame:
+    """Every model's latest forecast for one asset, plus how much to trust it.
+    This is the Signal Aggregator view: each model's most recent direction and
+    implied move, weighted by how well that model has actually been calibrated
+    on this slice, so a confident-but-miscalibrated model does not dominate.
+    """
+    manifest = store.load_manifest()
+    models = sorted({e.model_slug for e in
+                     manifest.find_signals(asset=asset, timeframe=timeframe)})
+    prices = store.get_prices(asset, timeframe)
+    if prices.empty or not models:
+        return pd.DataFrame()
+    last_close = float(prices["close"].iloc[-1])
+    rows = []
+    for m in models:
+        sig = store.get_signals(m, asset, timeframe)
+        if sig.empty:
+            continue
+        last = sig.iloc[-1]
+        ref = float(prices["close"].reindex([sig.index[-1]]).iloc[0]) \
+            if sig.index[-1] in prices.index else last_close
+        edge = float(last["q50"]) / ref - 1.0
+        sc = scorecard[(scorecard.get("model_slug") == m)
+                       & (scorecard.get("asset") == asset)
+                       & (scorecard.get("timeframe") == timeframe)] \
+            if not scorecard.empty else pd.DataFrame()
+        acc = float(sc["directional_accuracy"].iloc[0]) if len(sc) else float("nan")
+        cal_err = float(sc["calibration_error"].iloc[0]) if len(sc) else float("nan")
+        # Weight = how much better than a coin flip this model has been here.
+        weight = max(0.0, (acc - 0.5) * 2) if pd.notna(acc) else 0.0
+        spec = config.SEED_MODELS.get(m)
+        rows.append({
+            "model_slug": m,
+            "model": spec.display if spec else m,
+            "is_baseline": bool(spec and spec.family == "baseline"),
+            "as_of": sig.index[-1],
+            "reference_px": ref,
+            "q10": float(last["q10"]), "q50": float(last["q50"]),
+            "q90": float(last["q90"]),
+            "edge": edge,
+            "direction": "LONG" if edge > 0.0005 else ("SHORT" if edge < -0.0005 else "FLAT"),
+            "band_width": (float(last["q90"]) - float(last["q10"])) / max(abs(ref), 1e-9),
+            "directional_accuracy": acc,
+            "calibration_error": cal_err,
+            "weight": weight,
+        })
+    df = pd.DataFrame(rows)
+    if df.empty:
+        return df
+    return df.sort_values("weight", ascending=False).reset_index(drop=True)
+def consensus_verdict(consensus: pd.DataFrame) -> dict:
+    """Weighted aggregate of the per-model directions."""
+    if consensus.empty:
+        return {"direction": "NO DATA", "confidence": 0.0, "edge": 0.0,
+                "n_models": 0, "agree": 0}
+    w = consensus["weight"].fillna(0.0)
+    if w.sum() <= 0:
+        w = pd.Series(1.0, index=consensus.index)
+    edge = float((consensus["edge"] * w).sum() / w.sum())
+    direction = "LONG" if edge > 0.0005 else ("SHORT" if edge < -0.0005 else "FLAT")
+    agree = int((consensus["direction"] == direction).sum())
+    return {
+        "direction": direction,
+        "edge": edge,
+        "confidence": float(agree / len(consensus)),
+        "n_models": int(len(consensus)),
+        "agree": agree,
+    }

src/charts.py CHANGED Viewed

@@ -613,3 +613,164 @@ def monte_carlo_paths(trades: pd.DataFrame, n_paths: int = 1000, seed: int = 0)
     for i in range(n_paths):
         out[i] = np.cumprod(1.0 + rng.permutation(rets)) - 1.0
     return out

     for i in range(n_paths):
         out[i] = np.cumprod(1.0 + rng.permutation(rets)) - 1.0
     return out
+# --------------------------------------------------------------------------
+# Catalog / global comparison
+# --------------------------------------------------------------------------
+def multi_return_overlay(curves: dict[str, pd.Series], *, height: int = 420,
+                         highlight: str | None = None,
+                         max_series: int = 24) -> go.Figure:
+    """Cumulative return of many algorithms on one shared axis.
+    Series arrive already normalised to cumulative return by the catalog, so
+    nothing is re-based here and every line is directly comparable. Beyond
+    `max_series` the chart stops being readable, so extras are dropped and the
+    caller is expected to say so rather than silently truncating.
+    """
+    if not curves:
+        return empty_figure("select rows to plot", height)
+    fig = go.Figure()
+    items = list(curves.items())[:max_series]
+    for i, (name, series) in enumerate(items):
+        if series is None or len(series) == 0:
+            continue
+        is_hl = highlight is not None and name == highlight
+        color = SERIES_COLORS[i % len(SERIES_COLORS)]
+        fig.add_trace(go.Scatter(
+            x=series.index, y=(series * 100.0).to_numpy(), name=name[:40],
+            line=dict(color=PALETTE["amber_strong"] if is_hl else color,
+                      width=2.4 if is_hl else 1.3),
+            opacity=1.0 if (is_hl or highlight is None) else 0.45,
+            hovertemplate=f"{name}<br>%{{y:.1f}}%<extra></extra>",
+        ))
+    fig.add_hline(y=0, line=dict(color=PALETTE["border"], width=1))
+    fig.update_yaxes(ticksuffix="%")
+    return _base_layout(fig, height, showlegend=True, margin=(8, 8, 30, 8))
+def risk_return_scatter(df: pd.DataFrame, *, height: int = 380,
+                        x: str = "max_drawdown", y: str = "total_return",
+                        size: str = "trades", color_by: str = "strategy") -> go.Figure:
+    """Where every catalog row sits in risk/return space.
+    Marker area encodes trade count, so a spectacular result built on four
+    trades looks as small as it deserves to.
+    """
+    if df is None or df.empty:
+        return empty_figure("no catalog rows", height)
+    d = df.dropna(subset=[x, y]).copy()
+    if d.empty:
+        return empty_figure("no plottable rows", height)
+    d["_size"] = d[size].fillna(0).clip(lower=1) ** 0.5 if size in d.columns else 4
+    groups = list(dict.fromkeys(d[color_by])) if color_by in d.columns else ["all"]
+    fig = go.Figure()
+    for i, g in enumerate(groups):
+        sub = d[d[color_by] == g] if color_by in d.columns else d
+        if sub.empty:
+            continue
+        label = sub.get("model_display", pd.Series([""] * len(sub), index=sub.index))
+        fig.add_trace(go.Scatter(
+            x=(sub[x] * 100).to_numpy(), y=(sub[y] * 100).to_numpy(),
+            mode="markers", name=str(g)[:26],
+            marker=dict(color=SERIES_COLORS[i % len(SERIES_COLORS)],
+                        size=sub["_size"].to_numpy(), sizemode="area",
+                        sizeref=max(d["_size"].max() ** 2 / 900, 1e-9), sizemin=4,
+                        line=dict(width=0.5, color=PALETTE["border"])),
+            customdata=np.stack([sub["asset"], sub["timeframe"], label,
+                                 sub.get("trades", pd.Series(0, index=sub.index))], axis=-1),
+            hovertemplate=("%{customdata[0]} · %{customdata[1]} · %{customdata[2]}<br>"
+                           "drawdown %{x:.1f}% · return %{y:.1f}%<br>"
+                           "%{customdata[3]} trades<extra></extra>"),
+        ))
+    fig.add_hline(y=0, line=dict(color=PALETTE["border"], width=1))
+    fig.update_xaxes(title=dict(text="MAX DRAWDOWN", font=dict(size=9)), ticksuffix="%")
+    fig.update_yaxes(title=dict(text="TOTAL RETURN", font=dict(size=9)), ticksuffix="%")
+    return _base_layout(fig, height, showlegend=True, margin=(8, 8, 30, 34))
+def model_accuracy_bars(scorecard: pd.DataFrame, *, height: int = 320,
+                        timeframe: str | None = None) -> go.Figure:
+    """Directional accuracy per model, with the coin-flip line drawn in.
+    Baselines are coloured differently on purpose: the interesting question is
+    not which model scores highest, it is whether any learned model clears the
+    naive ones at all.
+    """
+    if scorecard is None or scorecard.empty:
+        return empty_figure("no scorecard rows", height)
+    d = scorecard
+    if timeframe:
+        d = d[d["timeframe"] == timeframe]
+    d = d.dropna(subset=["directional_accuracy"])
+    if d.empty:
+        return empty_figure("no directional calls recorded", height)
+    agg = (d.groupby(["model_display", "is_baseline"])["directional_accuracy"]
+           .mean().reset_index().sort_values("directional_accuracy"))
+    colors = [PALETTE["mute_blue"] if b else PALETTE["amber_strong"]
+              for b in agg["is_baseline"]]
+    fig = go.Figure(go.Bar(
+        x=(agg["directional_accuracy"] * 100).to_numpy(),
+        y=agg["model_display"].to_numpy(), orientation="h",
+        marker_color=colors,
+        hovertemplate="%{y}: %{x:.1f}% of directional calls correct<extra></extra>",
+    ))
+    fig.add_vline(x=50, line=dict(color=PALETTE["down"], width=1.5, dash="dot"),
+                  annotation_text="COIN FLIP", annotation_position="top",
+                  annotation_font=dict(family=FONT, size=9, color=PALETTE["down"]))
+    fig.update_xaxes(ticksuffix="%", range=[
+        max(0, float((agg["directional_accuracy"] * 100).min()) - 4),
+        float((agg["directional_accuracy"] * 100).max()) + 4])
+    return _base_layout(fig, height, margin=(8, 8, 20, 8))
+def calibration_scatter(scorecard: pd.DataFrame, *, height: int = 320) -> go.Figure:
+    """Band coverage against the 80% nominal line -- who is actually calibrated."""
+    if scorecard is None or scorecard.empty:
+        return empty_figure("no scorecard rows", height)
+    d = scorecard.dropna(subset=["coverage_q10_q90"])
+    if d.empty:
+        return empty_figure("no calibration data", height)
+    fig = go.Figure()
+    for i, (name, sub) in enumerate(d.groupby("model_display")):
+        fig.add_trace(go.Scatter(
+            x=sub["timeframe"], y=(sub["coverage_q10_q90"] * 100).to_numpy(),
+            mode="markers", name=str(name)[:24],
+            marker=dict(size=9, color=SERIES_COLORS[i % len(SERIES_COLORS)],
+                        symbol="diamond" if sub["is_baseline"].iloc[0] else "circle"),
+            customdata=sub[["asset"]].to_numpy(),
+            hovertemplate="%{customdata[0]}<br>coverage %{y:.1f}%<extra></extra>",
+        ))
+    fig.add_hline(y=80, line=dict(color=PALETTE["moss_strong"], width=1.5, dash="dot"),
+                  annotation_text="NOMINAL 80%", annotation_position="top left",
+                  annotation_font=dict(family=FONT, size=9, color=PALETTE["moss_strong"]))
+    fig.update_yaxes(ticksuffix="%", title=dict(text="q10–q90 COVERAGE", font=dict(size=9)))
+    return _base_layout(fig, height, showlegend=True, margin=(8, 8, 30, 34))
+def model_leaderboard_bars(df: pd.DataFrame, *, height: int = 320,
+                           metric: str = "oos_sharpe") -> go.Figure:
+    """Best result each model achieved, side by side."""
+    if df is None or df.empty or metric not in df.columns:
+        return empty_figure("no catalog rows", height)
+    d = df[df.get("model_slug", "") != ""].dropna(subset=[metric])
+    if d.empty:
+        return empty_figure("no model-driven rows", height)
+    agg = (d.groupby(["model_display", "is_baseline_model"])[metric]
+           .max().reset_index().sort_values(metric))
+    colors = [PALETTE["mute_blue"] if b else PALETTE["amber_strong"]
+              for b in agg["is_baseline_model"]]
+    fig = go.Figure(go.Bar(
+        x=agg[metric].to_numpy(), y=agg["model_display"].to_numpy(),
+        orientation="h", marker_color=colors,
+        hovertemplate="%{y}: best " + metric.replace("_", " ") + " %{x:.2f}<extra></extra>",
+    ))
+    fig.add_vline(x=0, line=dict(color=PALETTE["border"], width=1))
+    return _base_layout(fig, height, margin=(8, 8, 8, 8))

src/config.py CHANGED Viewed

@@ -158,18 +158,49 @@ class ModelSpec:
 SEED_MODELS: dict[str, ModelSpec] = {
     m.slug: m
     for m in [
         ModelSpec("chronos-bolt-small", "amazon/chronos-bolt-small", "chronos",
                   "Chronos-Bolt Small", context_len=512),
         ModelSpec("chronos-bolt-base", "amazon/chronos-bolt-base", "chronos",
                   "Chronos-Bolt Base", context_len=512),
         ModelSpec("timesfm-2-500m", "google/timesfm-2.0-500m-pytorch", "timesfm",
                   "TimesFM 2.0 500M", context_len=512),
     ]
 }
 # Adapter families a user may pick from in the "Add model" flow. Restricting to
 # a fixed set is what keeps arbitrary model code from ever being executed.
-ALLOWED_ADAPTER_FAMILIES = ("chronos", "timesfm")
 # --------------------------------------------------------------------------

 SEED_MODELS: dict[str, ModelSpec] = {
     m.slug: m
     for m in [
+        # Chronos-Bolt: the fast encoder-decoder family. All four sizes share one
+        # adapter, so comparing them isolates model capacity from everything else.
+        ModelSpec("chronos-bolt-tiny", "amazon/chronos-bolt-tiny", "chronos",
+                  "Chronos-Bolt Tiny", context_len=512),
+        ModelSpec("chronos-bolt-mini", "amazon/chronos-bolt-mini", "chronos",
+                  "Chronos-Bolt Mini", context_len=512),
         ModelSpec("chronos-bolt-small", "amazon/chronos-bolt-small", "chronos",
                   "Chronos-Bolt Small", context_len=512),
         ModelSpec("chronos-bolt-base", "amazon/chronos-bolt-base", "chronos",
                   "Chronos-Bolt Base", context_len=512),
+        # Original Chronos (T5-based, sampling rather than direct quantiles).
+        ModelSpec("chronos-t5-small", "amazon/chronos-t5-small", "chronos",
+                  "Chronos T5 Small", context_len=512),
+        # Naive baselines, deliberately first-class. A forecasting model that
+        # cannot beat "tomorrow looks like today" is not worth deploying, and
+        # the leaderboard should make that impossible to miss.
+        ModelSpec("baseline-naive", "baseline/naive", "baseline",
+                  "Baseline · Random walk", context_len=128),
+        ModelSpec("baseline-drift", "baseline/drift", "baseline",
+                  "Baseline · Drift", context_len=128),
+        ModelSpec("baseline-seasonal", "baseline/seasonal", "baseline",
+                  "Baseline · Seasonal naive", context_len=128),
+        # Registered but unseeded: the timesfm package is heavy and optional.
         ModelSpec("timesfm-2-500m", "google/timesfm-2.0-500m-pytorch", "timesfm",
                   "TimesFM 2.0 500M", context_len=512),
     ]
 }
+# Models the seed plan actually runs. TimesFM is excluded until its dependency
+# is pinned in requirements.txt.
+SEEDABLE_MODELS = tuple(k for k in SEED_MODELS if not k.startswith("timesfm"))
+BASELINE_MODELS = tuple(k for k, v in SEED_MODELS.items() if v.family == "baseline")
+def is_baseline(model_slug: str) -> bool:
+    return model_slug in BASELINE_MODELS
 # Adapter families a user may pick from in the "Add model" flow. Restricting to
 # a fixed set is what keeps arbitrary model code from ever being executed.
+ALLOWED_ADAPTER_FAMILIES = ("chronos", "timesfm", "baseline")
 # --------------------------------------------------------------------------

src/extension.py CHANGED Viewed

@@ -24,7 +24,7 @@ from dataclasses import dataclass
 import gradio as gr
 import pandas as pd
-from . import comparisons, config, runtime
 from .adapters import AdapterError, ModelNotAllowed, build_windows, get_adapter, validate_model_id
 from .store import _utc
@@ -244,10 +244,17 @@ def extend_coverage(model_slug: str, asset: str, timeframe: str, start, end,
         )
         if progress:
             progress(0.9, desc="Regenerating comparison tables")
         try:
             comparisons.regenerate(store, assets=[asset])
         except Exception:
             log.exception("comparison regeneration failed (coverage still written)")
         oid = store.flush(f"Extend {model_slug}/{asset}/{timeframe} by @{username}")
     runtime.cache_clear()

 import gradio as gr
 import pandas as pd
+from . import catalog, comparisons, config, runtime
 from .adapters import AdapterError, ModelNotAllowed, build_windows, get_adapter, validate_model_id
 from .store import _utc
         )
         if progress:
             progress(0.9, desc="Regenerating comparison tables")
+        # Coverage just changed, so every derived view is now stale. Both are
+        # rebuilt inside the same write lock, before the commit, so readers
+        # never see new signals alongside an old leaderboard.
         try:
             comparisons.regenerate(store, assets=[asset])
         except Exception:
             log.exception("comparison regeneration failed (coverage still written)")
+        try:
+            catalog.build(store)
+        except Exception:
+            log.exception("catalog rebuild failed (coverage still written)")
         oid = store.flush(f"Extend {model_slug}/{asset}/{timeframe} by @{username}")
     runtime.cache_clear()

src/metrics.py CHANGED Viewed

@@ -240,14 +240,21 @@ def calibration_error(coverage: float, nominal: float = 0.80) -> float:
 def directional_accuracy(actual_next: pd.Series, predicted_next: pd.Series,
                          reference: pd.Series) -> float:
-    """Share of bars where the predicted direction matched the realised one."""
     df = pd.concat([actual_next, predicted_next, reference], axis=1).dropna()
     if df.empty:
         return float("nan")
     a, p, ref = df.iloc[:, 0], df.iloc[:, 1], df.iloc[:, 2]
     actual_dir = np.sign(a - ref)
     pred_dir = np.sign(p - ref)
-    mask = actual_dir != 0
     if not mask.any():
         return float("nan")
     return float((actual_dir[mask] == pred_dir[mask]).mean())

 def directional_accuracy(actual_next: pd.Series, predicted_next: pd.Series,
                          reference: pd.Series) -> float:
+    """Share of *directional calls* that matched the realised direction.
+    Bars where the forecast is exactly flat are excluded, not counted as
+    misses. A random-walk forecast predicts "no change" every bar; it is never
+    wrong about direction because it never claims one. Scoring it 0% would say
+    it is always wrong, which is a different and false statement. When a model
+    never takes a side, its directional accuracy is undefined and returns NaN.
+    """
     df = pd.concat([actual_next, predicted_next, reference], axis=1).dropna()
     if df.empty:
         return float("nan")
     a, p, ref = df.iloc[:, 0], df.iloc[:, 1], df.iloc[:, 2]
     actual_dir = np.sign(a - ref)
     pred_dir = np.sign(p - ref)
+    mask = (actual_dir != 0) & (pred_dir != 0)
     if not mask.any():
         return float("nan")
     return float((actual_dir[mask] == pred_dir[mask]).mean())

src/runtime.py CHANGED Viewed

@@ -462,8 +462,12 @@ def overfit_verdict(rec: RunRecord) -> tuple[str, list[tuple[str, str]]]:
     return grade, checks
-def save_run_summary(rec: RunRecord) -> str:
-    """Write a shareable run summary into the store's runs/ folder."""
     payload = {
         "run_id": rec.run_id, "label": rec.label, "created_at": rec.created_at,
         "config": asdict(rec.request), "share_token": rec.request.encode(),
@@ -476,5 +480,11 @@ def save_run_summary(rec: RunRecord) -> str:
             if rec.result.metrics_holdout else None,
         },
     }
-    get_store().write_json(f"runs/{rec.run_id}.json", payload)
     return rec.run_id

     return grade, checks
+def save_run_summary(rec: RunRecord, *, push: bool = False) -> str:
+    """Write a shareable run summary into the store's runs/ folder.
+    With `push`, the summary is committed immediately so it shows up in the
+    global Run history for everyone rather than waiting for the next batch.
+    """
     payload = {
         "run_id": rec.run_id, "label": rec.label, "created_at": rec.created_at,
         "config": asdict(rec.request), "share_token": rec.request.encode(),
             if rec.result.metrics_holdout else None,
         },
     }
+    store = get_store()
+    store.write_json(f"runs/{rec.run_id}.json", payload)
+    if push:
+        try:
+            store.flush(f"Save run {rec.run_id}: {rec.label}")
+        except Exception as e:
+            log.warning("run summary staged but not pushed: %s", e)
     return rec.run_id

src/ui/compare_tab.py ADDED Viewed

	@@ -0,0 +1,211 @@

+"""The global Compare interface.
+Answers the question the per-session Comparison tab could not: *of everything
+this Space has ever computed, what actually worked?*
+It reads the precomputed catalog rather than re-running backtests, so a view
+over 168 strategy/model/asset/timeframe combinations opens instantly. Three
+data sources are merged:
+* the catalog sweep in the signal store (every combination, one canonical config)
+* `runs/*.json` -- runs people explicitly saved, across sessions and users
+* the current browser session's runs
+Two editorial decisions are deliberate. Low-trade rows are flagged and excluded
+from the default ranking, because a Sharpe of 4 on 13 trades will otherwise sit
+on top of the board forever. And naive baselines are shown next to learned
+models everywhere, because "did this beat doing nothing clever?" is the only
+question that matters first.
+"""
+from __future__ import annotations
+import pandas as pd
+from .. import catalog, charts
+from .format import EM, count, money, num, pct
+from . import components as C
+RANK_METRICS = {
+    "OOS Sharpe": "oos_sharpe",
+    "Total return": "total_return",
+    "CAGR": "cagr",
+    "Sharpe (all)": "sharpe",
+    "Sortino": "sortino",
+    "Profit factor": "profit_factor",
+    "Excess vs buy & hold": "excess_vs_hold",
+    "Holdout Sharpe": "holdout_sharpe",
+    "Max drawdown (least bad)": "max_drawdown",
+}
+TABLE_COLUMNS = [
+    ("strategy", "Strategy"), ("model_display", "Model"), ("asset", "Asset"),
+    ("timeframe", "TF"), ("oos_sharpe", "OOS Sharpe"), ("sharpe", "Sharpe"),
+    ("total_return", "Return"), ("cagr", "CAGR"), ("max_drawdown", "Max DD"),
+    ("win_rate", "Win%"), ("profit_factor", "PF"), ("trades", "Trades"),
+    ("excess_vs_hold", "vs Hold"), ("holdout_sharpe", "Holdout"),
+    ("costs_paid", "Costs"),
+]
+def leaderboard_table(df: pd.DataFrame) -> pd.DataFrame:
+    """Format catalog rows for display, flagging thin evidence."""
+    if df is None or df.empty:
+        return pd.DataFrame(columns=[label for _, label in TABLE_COLUMNS] + ["Note"])
+    out = pd.DataFrame()
+    for col, label in TABLE_COLUMNS:
+        if col not in df.columns:
+            continue
+        s = df[col]
+        if col in ("total_return", "cagr", "max_drawdown", "excess_vs_hold"):
+            out[label] = s.map(lambda v: pct(v))
+        elif col == "win_rate":
+            out[label] = s.map(lambda v: pct(v, 0, signed=False))
+        elif col in ("oos_sharpe", "sharpe", "profit_factor", "holdout_sharpe"):
+            out[label] = s.map(lambda v: num(v))
+        elif col == "costs_paid":
+            out[label] = s.map(money)
+        elif col == "trades":
+            out[label] = s.map(count)
+        else:
+            out[label] = s.fillna(EM)
+    if "significant" in df.columns:
+        out["Note"] = df["significant"].map(
+            lambda ok: "" if ok else f"< {catalog.MIN_MEANINGFUL_TRADES} trades")
+    return out.reset_index(drop=True)
+def scorecard_table(sc: pd.DataFrame) -> pd.DataFrame:
+    cols = ["Model", "Asset", "TF", "n", "Coverage", "Cal. error",
+            "Dir. accuracy", "vs momentum", "vs random", "Band width"]
+    if sc is None or sc.empty:
+        return pd.DataFrame(columns=cols)
+    out = pd.DataFrame({
+        "Model": sc["model_display"],
+        "Asset": sc["asset"], "TF": sc["timeframe"],
+        "n": sc["n"].map(count),
+        "Coverage": sc["coverage_q10_q90"].map(lambda v: pct(v, 1, signed=False)),
+        "Cal. error": sc["calibration_error"].map(lambda v: pct(v, 1)),
+        "Dir. accuracy": sc["directional_accuracy"].map(
+            lambda v: pct(v, 1, signed=False)),
+        "vs momentum": sc["beats_momentum"].map(lambda v: pct(v, 1)),
+        "vs random": sc["beats_random"].map(lambda v: pct(v, 1)),
+        "Band width": sc["band_width_pct"].map(lambda v: pct(v, 2, signed=False)),
+    })
+    return out.reset_index(drop=True)
+def runs_table(session_runs, saved: pd.DataFrame) -> pd.DataFrame:
+    """Session runs and store-persisted runs in one list."""
+    cols = ["When", "Source", "Label", "Strategy", "Asset", "TF",
+            "Return", "Sharpe", "OOS Sharpe", "Max DD", "Trades"]
+    rows = []
+    for r in (session_runs or []):
+        m = r.result.metrics_all
+        rows.append({
+            "When": r.created_at.replace("T", " ")[:16], "Source": "this session",
+            "Label": r.label, "Strategy": r.request.strategy,
+            "Asset": r.request.asset, "TF": r.request.timeframe,
+            "Return": pct(m.total_return), "Sharpe": num(m.sharpe),
+            "OOS Sharpe": num(r.result.metrics_oos.sharpe)
+            if r.result.metrics_oos.bars else EM,
+            "Max DD": pct(m.max_drawdown), "Trades": count(m.trade_count),
+        })
+    if saved is not None and not saved.empty:
+        for _, r in saved.iterrows():
+            rows.append({
+                "When": str(r.get("created_at", "")).replace("T", " ")[:16],
+                "Source": "signal store", "Label": r.get("label", ""),
+                "Strategy": r.get("strategy", ""), "Asset": r.get("asset", ""),
+                "TF": r.get("timeframe", ""),
+                "Return": pct(r.get("total_return")), "Sharpe": num(r.get("sharpe")),
+                "OOS Sharpe": num(r.get("oos_sharpe")),
+                "Max DD": pct(r.get("max_drawdown")),
+                "Trades": count(r.get("trades")),
+            })
+    if not rows:
+        return pd.DataFrame(columns=cols)
+    return pd.DataFrame(rows, columns=cols).sort_values("When", ascending=False)
+# --------------------------------------------------------------------------
+# View builders
+# --------------------------------------------------------------------------
+def build_leaderboard_view(store, *, assets, timeframes, strategies_, models,
+                           metric_label, min_trades, hide_baselines,
+                           require_oos, top_n):
+    """Everything the Leaderboard sub-tab renders, in one pass."""
+    lb = catalog.load_leaderboard(store)
+    if lb.empty:
+        empty = charts.empty_figure("catalog not generated yet")
+        return (C.note("The catalog has not been generated yet. Run "
+                       "<code>scripts/seed_store.py</code> or extend coverage.",
+                       danger=True),
+                pd.DataFrame(), empty, empty, C.micro("no rows"))
+    metric = RANK_METRICS.get(metric_label, "oos_sharpe")
+    filtered = catalog.filter_leaderboard(
+        lb, assets=assets or None, timeframes=timeframes or None,
+        strategies_=strategies_ or None, models=models or None,
+        min_trades=int(min_trades or 0), require_oos=bool(require_oos),
+        hide_baselines=bool(hide_baselines))
+    # Every metric here ranks high-to-low, drawdown included: drawdowns are
+    # stored as negative numbers, so -0.10 sorts above -0.50 already and
+    # "least bad" falls out of the default ordering.
+    ascending = False
+    top = int(top_n or 25)
+    # Ranking prefers evidence. Rows with too few trades are excluded, and only
+    # if that leaves nothing at all do we fall back to showing thin results --
+    # an empty board would be less honest than a flagged one.
+    ranked = catalog.rank(filtered, metric, top=top, ascending=ascending,
+                          significant_only=True)
+    if ranked.empty:
+        ranked = catalog.rank(filtered, metric, top=top, ascending=ascending)
+    curves = catalog.curves_for(catalog.load_equity_curves(store),
+                                list(ranked["key"])[:12])
+    labels = {}
+    for _, r in ranked.iterrows():
+        k = r["key"]
+        if k in curves:
+            name = f"{r['strategy']} · {r['asset']} {r['timeframe']}"
+            if r.get("model_slug"):
+                name += f" · {r['model_display']}"
+            labels[name] = curves[k]
+    thin = int((~filtered["significant"]).sum()) if "significant" in filtered else 0
+    meta = C.micro(
+        f"{len(filtered)} of {len(lb)} rows match · ranked by {metric_label} · "
+        f"showing top {len(ranked)}"
+        + (f" · {thin} rows below {catalog.MIN_MEANINGFUL_TRADES} trades excluded "
+           f"from ranking" if thin else ""))
+    return (C.podium(ranked, metric),
+            leaderboard_table(ranked),
+            charts.multi_return_overlay(labels),
+            charts.risk_return_scatter(filtered),
+            meta)
+def build_models_view(store, timeframe: str | None):
+    sc = catalog.load_scorecard(store)
+    lb = catalog.load_leaderboard(store)
+    tf = None if timeframe in (None, "", "all") else timeframe
+    sub = sc if tf is None else sc[sc["timeframe"] == tf]
+    return (C.scorecard_note(sub),
+            charts.model_accuracy_bars(sc, timeframe=tf),
+            charts.calibration_scatter(sub),
+            charts.model_leaderboard_bars(lb),
+            scorecard_table(sub.sort_values("directional_accuracy", ascending=False)
+                            if not sub.empty else sub))
+def build_signals_view(store, asset: str, timeframe: str):
+    sc = catalog.load_scorecard(store)
+    cons = catalog.model_consensus(sc, store, asset, timeframe)
+    verdict = catalog.consensus_verdict(cons)
+    return C.consensus_panel(cons, verdict, asset, timeframe)

src/ui/components.py ADDED Viewed

	@@ -0,0 +1,359 @@

+"""HTML fragments in the Bit design language.
+Gradio gives us controls; the design needs panels, chips, stat bands and
+micro-labels that Gradio has no equivalent for. Those live here as small pure
+functions returning markup, so `app.py` stays layout and wiring only.
+Every fragment uses design-system tokens (`var(--...)`) rather than raw colour,
+so a theme change in `assets/tokens/colors.css` propagates without edits here.
+"""
+from __future__ import annotations
+import pandas as pd
+from .. import config
+from .format import EM, arrow, count, esc, money, num, pct, seg, sharpe_tone, tone
+# --------------------------------------------------------------------------
+# Primitives
+# --------------------------------------------------------------------------
+def micro(text: str, color: str = "var(--text-tertiary)") -> str:
+    return f'<div class="bit-micro" style="color:{color}">{esc(text)}</div>'
+def chip(text: str, kind: str = "") -> str:
+    cls = {"ok": " bit-chip-ok", "run": " bit-chip-run",
+           "warn": " bit-chip-warn", "accent": " bit-chip-accent"}.get(kind, "")
+    return f'<span class="bit-chip{cls}">{esc(text)}</span>'
+def note(text: str, danger: bool = False) -> str:
+    cls = "bit-note bit-note-danger" if danger else "bit-note"
+    return f'<div class="{cls}">{text}</div>'
+def panel_head(title: str, meta: str = "", right: str = "") -> str:
+    m = f'<span class="bit-micro">{esc(meta)}</span>' if meta else ""
+    r = f'<span style="margin-left:auto">{right}</span>' if right else ""
+    return (f'<div class="bit-panel-head"><span class="bit-h3">{esc(title)}</span>'
+            f'{m}{r}</div>')
+def panel(title: str, body: str, meta: str = "", right: str = "") -> str:
+    return f'<div class="bit-panel">{panel_head(title, meta, right)}{body}</div>'
+def kbd(text: str) -> str:
+    return f'<span class="bit-kbd">{esc(text)}</span>'
+# --------------------------------------------------------------------------
+# Top bar & footer
+# --------------------------------------------------------------------------
+def top_bar(status_text: str = "NO RUN LOADED", status_kind: str = "",
+            context: str = "", elapsed: str = "") -> str:
+    ctx = f'<span class="bit-chip bit-chip-ctx">{esc(context)}</span>' if context else ""
+    el = f'<span class="bit-micro" style="margin:0 4px">{esc(elapsed)}</span>' if elapsed else ""
+    return f"""
+<div class="bit-topbar">
+  <span class="bit-mark"></span>
+  <span class="bit-wordmark">BIT</span>
+  <span class="bit-slash">/</span>
+  <span class="bit-h1">Backtest Lab</span>
+  {ctx}
+  <span class="bit-spacer"></span>
+  {el}{chip(status_text, status_kind)}
+  <a class="bit-chip bit-link" target="_blank" rel="noopener"
+     href="https://huggingface.co/datasets/{config.STORE_REPO}">SIGNAL STORE ↗</a>
+  <a class="bit-chip bit-link" target="_blank" rel="noopener"
+     href="https://huggingface.co/spaces/{config.SPACE_REPO}">SPACE ↗</a>
+</div>"""
+def footer() -> str:
+    return f"""
+<div class="bit-footer">
+  <span>{esc(config.DISCLAIMER)}</span>
+  <span class="bit-footer-right">BITTRADING BACKTEST LAB v1.1.0</span>
+</div>"""
+# --------------------------------------------------------------------------
+# Empty & loading states
+# --------------------------------------------------------------------------
+def empty_state() -> str:
+    return f"""
+<div class="bit-empty">
+  <div class="bit-empty-glyph">◴</div>
+  <div class="bit-h2">No run loaded</div>
+  <div class="bit-empty-copy">
+    Configure a strategy on the left, or start from a worked example and edit it.
+  </div>
+  <div class="bit-kbd-row">
+    {kbd("RUN BACKTEST")}{kbd("COSTS DEFAULT ON")}{kbd("FILLS AT NEXT BAR OPEN")}
+  </div>
+</div>"""
+STAGE_LABELS = ("Reading cached slices", "Simulating trades",
+                "Walking forward", "Computing robustness")
+def loading_stages(active: int = 0) -> str:
+    rows = []
+    for i, label in enumerate(STAGE_LABELS):
+        mark = "✓" if i < active else ("▸" if i == active else "·")
+        color = ("var(--accent-moss-strong)" if i < active else
+                 "var(--text-primary)" if i == active else "var(--text-tertiary)")
+        rows.append(f'<div style="color:{color};font-size:11px">{mark} {esc(label)}</div>')
+    return f'<div class="bit-panel">{"".join(rows)}</div>'
+# --------------------------------------------------------------------------
+# Stat band
+# --------------------------------------------------------------------------
+def stat_band(rec) -> str:
+    """The design's stat band: big mono number, tiny label, IS/OOS underneath."""
+    if rec is None:
+        return ""
+    r = rec.result
+    a, i, o = r.metrics_all, r.metrics_is, r.metrics_oos
+    def isoos(fmt, key, *fargs):
+        return (f"IS {seg(i, fmt, getattr(i, key), *fargs)} · "
+                f"OOS {seg(o, fmt, getattr(o, key), *fargs)}")
+    bench = (float(r.benchmark_equity.iloc[-1] / r.benchmark_equity.iloc[0] - 1.0)
+             if len(r.benchmark_equity) else float("nan"))
+    gap = a.total_return - bench if pd.notna(bench) else float("nan")
+    cells = [
+        ("Total return", f"{pct(a.total_return)}{arrow(a.total_return)}",
+         tone(a.total_return), isoos(pct, "total_return"),
+         "Cumulative return of the strategy equity curve, costs included."),
+        ("CAGR", pct(a.cagr), tone(a.cagr), isoos(pct, "cagr"),
+         "Compound annual growth rate implied by the equity curve."),
+        ("Sharpe", num(a.sharpe), tone(a.sharpe), isoos(num, "sharpe"),
+         "Annualized mean excess return over return volatility."),
+        ("Sortino", num(a.sortino), tone(a.sortino), isoos(num, "sortino"),
+         "Sharpe with only downside deviation in the denominator."),
+        ("Max drawdown", pct(a.max_drawdown), "bit-down",
+         isoos(pct, "max_drawdown"), "Worst peak-to-trough decline."),
+        ("Win rate", pct(a.win_rate, 0, signed=False), "",
+         f"IS {seg(i, pct, i.win_rate, 0, False)} · OOS {seg(o, pct, o.win_rate, 0, False)}",
+         "Share of closed trades with positive net P&L."),
+        ("Profit factor", num(a.profit_factor), tone(a.profit_factor - 1.0),
+         isoos(num, "profit_factor"), "Gross profit over gross loss."),
+        ("Trades", count(a.trade_count), "",
+         f"IS {seg(i, count, i.trade_count)} · OOS {seg(o, count, o.trade_count)}",
+         "Closed round-trip trades."),
+        ("Exposure", pct(a.exposure, 0, signed=False), "",
+         f"IS {seg(i, pct, i.exposure, 0, False)} · OOS {seg(o, pct, o.exposure, 0, False)}",
+         "Fraction of bars holding a position."),
+        ("vs buy & hold", f"{pct(gap)}{arrow(gap)}", tone(gap),
+         f"costs paid {money(r.costs_paid)}",
+         "Strategy return minus buy-and-hold over the same window."),
+    ]
+    html = ['<div class="bit-statband">']
+    for label, value, cls, sub, tip in cells:
+        html.append(
+            f'<div class="bit-stat" title="{esc(tip)}">'
+            f'<div class="bit-stat-label">{esc(label)}</div>'
+            f'<div class="bit-stat-value {cls}">{value}</div>'
+            f'<div class="bit-stat-sub">{sub}</div></div>')
+    html.append("</div>")
+    for n in getattr(r.plan, "notes", []):
+        html.append(note(esc(n), danger=True))
+    if r.metrics_holdout is not None:
+        h = r.metrics_holdout
+        ok = h.total_return > 0
+        html.append(note(
+            f'<b>LOCKED HOLDOUT</b> · return {pct(h.total_return)} · '
+            f'Sharpe {num(h.sharpe)} · {h.bars} bars never used for any parameter '
+            f'choice.' + ("" if ok else " <b>It loses money here.</b>"),
+            danger=not ok))
+    return "".join(html)
+# --------------------------------------------------------------------------
+# Run manager
+# --------------------------------------------------------------------------
+def run_card(rec, selected: bool = False) -> str:
+    s = rec.sharpe
+    color = sharpe_tone(s)
+    mark = "✓" if selected else ""
+    return (
+        f'<div class="bit-run-card{" bit-run-card-sel" if selected else ""}">'
+        f'<div class="bit-run-top">'
+        f'<span class="bit-run-name">{esc(rec.label[:36])}</span>'
+        f'<span class="bit-run-mark">{mark}</span></div>'
+        f'<div class="bit-micro">{esc(rec.meta)}</div>'
+        f'<div class="bit-run-sharpe" style="color:{color}">'
+        f'SHARPE {num(s)}<span class="bit-run-ret">{pct(rec.result.metrics_all.total_return)}</span>'
+        f'</div></div>')
+def run_history(hist, selected_ids: set[str] | None = None) -> str:
+    if not hist:
+        return micro("no runs yet in this session")
+    sel = selected_ids or set()
+    head = micro(f"{len(hist)} run{'s' if len(hist) != 1 else ''} this session")
+    return head + "".join(run_card(r, r.run_id in sel) for r in hist[:14])
+def glossary(items) -> str:
+    rows = "".join(
+        f'<div class="bit-gloss"><div class="bit-micro">{esc(t)}</div>'
+        f'<div class="bit-gloss-def">{esc(d)}</div></div>'
+        for t, d in items)
+    return f'<div class="bit-panel">{rows}</div>'
+# --------------------------------------------------------------------------
+# Coverage
+# --------------------------------------------------------------------------
+def coverage_summary(cells) -> str:
+    if not cells:
+        return note("The signal store has no coverage yet.", danger=True)
+    models = sorted({c.model_slug for c in cells})
+    assets = sorted({c.asset for c in cells})
+    tfs = sorted({c.timeframe for c in cells})
+    rows = sum(c.rows for c in cells)
+    ph = sum(1 for c in cells if c.is_placeholder)
+    body = (
+        f'<div class="bit-kpi-row">'
+        f'{_kpi(len(cells), "slices")}{_kpi(len(models), "models")}'
+        f'{_kpi(len(assets), "assets")}{_kpi(len(tfs), "timeframes")}'
+        f'{_kpi(f"{rows:,}", "signal rows")}'
+        f'{_kpi(ph, "placeholder", danger=ph > 0)}</div>')
+    return panel("Coverage", body, meta="SIGNAL STORE")
+def _kpi(value, label, danger: bool = False) -> str:
+    color = "var(--fin-down)" if danger else "var(--text-primary)"
+    return (f'<div class="bit-kpi"><div class="bit-kpi-value" style="color:{color}">'
+            f'{esc(value)}</div><div class="bit-stat-label">{esc(label)}</div></div>')
+# --------------------------------------------------------------------------
+# Leaderboard & consensus (the global compare interface)
+# --------------------------------------------------------------------------
+def podium(df: pd.DataFrame, metric: str = "oos_sharpe") -> str:
+    """Top three rows of the leaderboard, called out above the table."""
+    if df is None or df.empty:
+        return micro("no catalog rows yet")
+    top = df.head(3)
+    cards = []
+    for rank, (_, r) in enumerate(top.iterrows(), start=1):
+        val = r.get(metric)
+        model = r.get("model_display") or EM
+        cards.append(
+            f'<div class="bit-podium bit-podium-{rank}">'
+            f'<div class="bit-podium-rank">#{rank}</div>'
+            f'<div class="bit-podium-name">{esc(r.get("strategy", ""))}</div>'
+            f'<div class="bit-micro">{esc(r.get("asset", ""))} · '
+            f'{esc(r.get("timeframe", ""))}'
+            + (f' · {esc(model)}' if r.get("model_slug") else "") + '</div>'
+            f'<div class="bit-podium-value" style="color:{sharpe_tone(val)}">'
+            f'{num(val)}<span class="bit-podium-unit">OOS SHARPE</span></div>'
+            f'<div class="bit-micro">ret {pct(r.get("total_return"))} · '
+            f'dd {pct(r.get("max_drawdown"))} · {count(r.get("trades"))} trades</div>'
+            f'</div>')
+    return f'<div class="bit-podium-row">{"".join(cards)}</div>'
+def consensus_panel(consensus: pd.DataFrame, verdict: dict,
+                    asset: str, timeframe: str) -> str:
+    """Signal Aggregator: every model's latest call for one asset, plus a
+    calibration-weighted consensus.
+    Weight is how far a model's realised directional accuracy sits above a coin
+    flip on this exact slice, so a confident but historically wrong model does
+    not get to shout.
+    """
+    if consensus is None or consensus.empty:
+        return note(f"No model signals cached for {esc(asset)} {esc(timeframe)}.")
+    rows = []
+    for _, r in consensus.iterrows():
+        d = r["direction"]
+        dcolor = ("var(--fin-up-strong)" if d == "LONG" else
+                  "var(--fin-down-strong)" if d == "SHORT" else "var(--text-tertiary)")
+        glyph = "▲" if d == "LONG" else ("▼" if d == "SHORT" else "■")
+        tag = ' <span class="bit-tag">BASELINE</span>' if r["is_baseline"] else ""
+        acc = r["directional_accuracy"]
+        rows.append(
+            f'<div class="bit-sig-row">'
+            f'<span class="bit-sig-name">{esc(r["model"])}{tag}</span>'
+            f'<span class="bit-sig-dir" style="color:{dcolor}">{glyph} {d}</span>'
+            f'<span class="bit-sig-edge" style="color:{dcolor}">{pct(r["edge"], 2)}</span>'
+            f'<span class="bit-sig-acc">acc {pct(acc, 0, signed=False)}</span>'
+            f'<span class="bit-sig-w">w{num(r["weight"])}</span>'
+            f'</div>')
+    d = verdict["direction"]
+    vcolor = ("var(--fin-up-strong)" if d == "LONG" else
+              "var(--fin-down-strong)" if d == "SHORT" else "var(--text-secondary)")
+    glyph = "▲" if d == "LONG" else ("▼" if d == "SHORT" else "■")
+    conf = verdict["confidence"]
+    verdict_box = (
+        f'<div class="bit-consensus">'
+        f'<div class="bit-micro">{esc(asset)} · {esc(timeframe)}</div>'
+        f'<div class="bit-consensus-dir" style="color:{vcolor}">{d} {glyph}</div>'
+        f'<div class="bit-micro">WEIGHTED EDGE {pct(verdict["edge"], 2)}</div>'
+        f'<div class="bit-conf-track"><div class="bit-conf-fill" '
+        f'style="width:{conf * 100:.0f}%;background:{vcolor}"></div></div>'
+        f'<div class="bit-micro">{verdict["agree"]}/{verdict["n_models"]} MODELS AGREE</div>'
+        f'</div>')
+    return panel(
+        "Signal aggregator",
+        f'<div class="bit-sig-grid"><div class="bit-sig-list">{"".join(rows)}</div>'
+        f'{verdict_box}</div>'
+        + micro("weight = realised directional accuracy above a coin flip, "
+                "on this asset and timeframe"),
+        meta=f"{len(consensus)} MODELS",
+    )
+def scorecard_note(sc: pd.DataFrame) -> str:
+    """One honest sentence about whether the models beat the baselines."""
+    if sc is None or sc.empty:
+        return ""
+    learned = sc[~sc["is_baseline"]] if "is_baseline" in sc.columns else sc
+    base = sc[sc["is_baseline"]] if "is_baseline" in sc.columns else pd.DataFrame()
+    if learned.empty or base.empty:
+        return ""
+    la = learned["directional_accuracy"].mean()
+    ba = base["directional_accuracy"].mean()
+    if pd.isna(la) or pd.isna(ba):
+        return ""
+    delta = la - ba
+    if delta > 0.01:
+        return note(f"Learned models call direction correctly "
+                    f"{pct(la, 1, signed=False)} of the time against "
+                    f"{pct(ba, 1, signed=False)} for naive baselines "
+                    f"({pct(delta, 1)} better).")
+    return note(f"Learned models call direction correctly "
+                f"{pct(la, 1, signed=False)} of the time; naive baselines manage "
+                f"{pct(ba, 1, signed=False)}. <b>The models are not beating the "
+                f"baselines on direction.</b> Read the leaderboard with that in mind.",
+                danger=True)

src/ui/format.py ADDED Viewed

	@@ -0,0 +1,76 @@

+"""Value formatting shared by every part of the UI.
+One rule runs through all of it: a number that does not exist must never render
+as a number that does. Empty segments become an em dash, not `0.00`.
+"""
+from __future__ import annotations
+import pandas as pd
+EM = "—"
+def pct(v, digits: int = 1, signed: bool = True) -> str:
+    if v is None or pd.isna(v):
+        return EM
+    return f"{v * 100:+.{digits}f}%" if signed else f"{v * 100:.{digits}f}%"
+def num(v, digits: int = 2) -> str:
+    if v is None or pd.isna(v):
+        return EM
+    return f"{v:.{digits}f}"
+def money(v) -> str:
+    if v is None or pd.isna(v):
+        return EM
+    return f"${v:,.0f}"
+def count(v) -> str:
+    if v is None or pd.isna(v):
+        return EM
+    return f"{int(v):,}"
+def seg(metrics, fmt, *args, **kwargs) -> str:
+    """Format a segment metric, or an em dash when that segment has no bars.
+    "The out-of-sample Sharpe is zero" and "there is no out-of-sample period"
+    are different claims. Only one of them is ever true here.
+    """
+    if metrics is None or getattr(metrics, "bars", 0) == 0:
+        return EM
+    return fmt(*args, **kwargs)
+def tone(v) -> str:
+    """CSS class for a signed value."""
+    if v is None or pd.isna(v) or v == 0:
+        return ""
+    return "bit-up" if v > 0 else "bit-down"
+def arrow(v) -> str:
+    """Direction as a glyph, so colour is never the only encoding."""
+    if v is None or pd.isna(v) or v == 0:
+        return ""
+    return " ▲" if v > 0 else " ▼"
+def sharpe_tone(v) -> str:
+    if v is None or pd.isna(v):
+        return "var(--text-tertiary)"
+    if v >= 1.0:
+        return "var(--accent-moss-strong)"
+    if v < 0:
+        return "var(--fin-down)"
+    return "var(--text-secondary)"
+def esc(s) -> str:
+    """Minimal HTML escaping for values interpolated into markup."""
+    return (str(s).replace("&", "&amp;").replace("<", "&lt;")
+            .replace(">", "&gt;").replace('"', "&quot;"))

src/ui/theme.py CHANGED Viewed

@@ -52,100 +52,212 @@ def _tokens_css() -> str:
 SHELL_CSS = """
-/* ---------- app shell ---------- */
 .gradio-container{
   max-width:100% !important; padding:0 !important;
   background:var(--bg-canvas) !important;
   font-family:var(--font-body); font-weight:var(--weight-body);
-  color:var(--text-primary);
 }
 .gradio-container *{ border-radius:var(--radius-sm) !important; }
 footer{ display:none !important; }
-/* ---------- typography ---------- */
 .bit-h1,.bit-h2,.bit-h3{
   font-family:var(--font-heading); font-weight:var(--weight-heading);
   letter-spacing:var(--tracking-wide); text-transform:uppercase;
   color:var(--text-primary); margin:0;
 }
-.bit-h1{ font-size:var(--text-lg); }
 .bit-h2{ font-size:var(--text-md); }
 .bit-h3{ font-size:var(--text-base); }
 .bit-micro{
   font-family:var(--font-tiny); font-size:var(--text-xs);
   letter-spacing:var(--tracking-wider); text-transform:uppercase;
-  color:var(--text-tertiary);
 }
-.bit-mono{ font-family:var(--font-mono); font-size:var(--text-base); }
-/* ---------- panels ---------- */
-.bit-panel{
-  background:var(--bg-panel); border:var(--border-width) solid var(--border-default);
-  padding:var(--space-4); margin-bottom:var(--space-3);
 }
-.bit-panel-head{
-  display:flex; align-items:center; gap:var(--space-3);
-  border-bottom:var(--border-width) solid var(--border-subtle);
-  padding-bottom:var(--space-2); margin-bottom:var(--space-3);
 }
-/* ---------- top bar ---------- */
-.bit-topbar{
-  display:flex; align-items:center; gap:var(--space-4);
-  background:var(--bg-panel); border-bottom:var(--border-width) solid var(--border-default);
-  padding:var(--space-3) var(--space-4); position:sticky; top:0; z-index:var(--z-header);
 }
-.bit-mark{ width:22px; height:22px; background:var(--accent-amber); display:inline-block; }
 .bit-chip{
-  display:inline-flex; align-items:center; gap:6px;
   font-family:var(--font-mono); font-size:var(--text-sm);
   border:var(--border-width) solid var(--border-default);
-  padding:3px var(--space-2); color:var(--text-secondary);
 }
 .bit-chip-ok{ color:var(--accent-moss-strong); border-color:var(--accent-moss-dim); }
 .bit-chip-run{ color:var(--accent-amber-strong); border-color:var(--accent-amber-dim); }
 .bit-chip-warn{ color:var(--fin-down); border-color:var(--fin-down); }
-/* ---------- stat band ---------- */
-.bit-statband{ display:flex; flex-wrap:wrap; border:var(--border-width) solid var(--border-default); }
 .bit-stat{
-  flex:1 1 150px; padding:var(--space-3) var(--space-4);
   border-right:var(--border-width) solid var(--border-subtle);
 }
-.bit-stat:last-child{ border-right:none; }
 .bit-stat-label{
   font-family:var(--font-tiny); font-size:var(--text-xs);
-  letter-spacing:var(--tracking-wider); text-transform:uppercase; color:var(--text-tertiary);
 }
 .bit-stat-value{
   font-family:var(--font-mono); font-size:var(--text-xl);
-  line-height:var(--leading-tight); color:var(--text-primary); margin:2px 0;
 }
-.bit-stat-sub{ font-family:var(--font-mono); font-size:var(--text-xs); color:var(--text-tertiary); }
 .bit-up{ color:var(--fin-up-strong); } .bit-down{ color:var(--fin-down-strong); }
-/* ---------- warnings / notes ---------- */
 .bit-note{
-  font-family:var(--font-mono); font-size:var(--text-sm);
   border-left:var(--border-width-strong) solid var(--accent-amber);
   background:var(--bg-raised); padding:var(--space-2) var(--space-3);
-  color:var(--text-secondary);
 }
 .bit-note-danger{ border-left-color:var(--fin-down); }
-/* ---------- footer disclaimer (always visible) ---------- */
-.bit-footer{
-  position:sticky; bottom:0; z-index:var(--z-header);
-  display:flex; justify-content:space-between; gap:var(--space-4);
-  background:var(--bg-panel); border-top:var(--border-width) solid var(--border-default);
-  padding:var(--space-2) var(--space-4);
-  font-family:var(--font-mono); font-size:var(--text-sm); color:var(--text-tertiary);
 }
-/* ---------- gradio overrides ---------- */
 .gradio-container .tabs > .tab-nav{
   border-bottom:var(--border-width) solid var(--border-default) !important;
-  background:transparent !important; gap:0 !important;
 }
 .gradio-container .tabs > .tab-nav > button{
   font-family:var(--font-heading) !important; text-transform:uppercase;
@@ -154,6 +266,7 @@ footer{ display:none !important; }
   border:none !important; border-bottom:2px solid transparent !important;
   padding:var(--space-2) var(--space-4) !important;
 }
 .gradio-container .tabs > .tab-nav > button.selected{
   color:var(--text-primary) !important;
   border-bottom-color:var(--accent-amber) !important;
@@ -161,7 +274,7 @@ footer{ display:none !important; }
 .gradio-container .form, .gradio-container .block{
   background:transparent !important; border:none !important;
 }
-/* Field captions only -- scoped so it never swallows radio/checkbox option text. */
 .block > label > span, .block > .form > label > span{
   font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
   letter-spacing:var(--tracking-wider) !important; text-transform:uppercase;
@@ -172,8 +285,11 @@ input, select, textarea{
   background:var(--bg-sunken) !important; color:var(--text-primary) !important;
   border:var(--border-width) solid var(--border-default) !important;
 }
-/* Radio / checkbox options read as segmented chips, per the design. */
 .gradio-container fieldset label,
 .gradio-container .wrap label:has(input[type="radio"]),
 .gradio-container .wrap label:has(input[type="checkbox"]){
@@ -192,8 +308,7 @@ input, select, textarea{
 .gradio-container fieldset label.selected,
 .gradio-container fieldset label:has(input:checked),
 .gradio-container .wrap label:has(input[type="radio"]:checked){
-  background:var(--accent-amber) !important;
-  border-color:var(--accent-amber) !important;
 }
 .gradio-container fieldset label.selected span,
 .gradio-container fieldset label:has(input:checked) span,
@@ -203,6 +318,22 @@ input, select, textarea{
 .gradio-container input[type="radio"], .gradio-container input[type="checkbox"]{
   accent-color:var(--accent-amber);
 }
 .bit-run-btn{
   background:var(--accent-amber) !important; color:var(--stone-950) !important;
   font-family:var(--font-heading) !important; text-transform:uppercase;
@@ -214,29 +345,34 @@ input, select, textarea{
   border:var(--border-width) solid var(--border-default) !important;
   font-family:var(--font-mono) !important; font-size:var(--text-sm) !important;
 }
-.bit-accordion{ border:var(--border-width) solid var(--border-subtle) !important; }
-/* ---------- tables ---------- */
 .bit-table table{ font-family:var(--font-mono) !important; font-size:var(--text-sm) !important; }
 .bit-table thead th{
   font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
   letter-spacing:var(--tracking-wider); text-transform:uppercase;
   color:var(--text-tertiary) !important; background:var(--bg-raised) !important;
 }
-/* ---------- empty state ---------- */
-.bit-empty{
-  display:flex; flex-direction:column; align-items:center; justify-content:center;
-  gap:var(--space-3); padding:var(--space-9) var(--space-4); text-align:center;
-  border:var(--border-width) dashed var(--border-default); background:var(--bg-panel);
-}
-.bit-kbd{
-  font-family:var(--font-mono); font-size:var(--text-xs); color:var(--text-tertiary);
-  border:var(--border-width) solid var(--border-default); padding:2px 6px;
 }
 @media (max-width: 900px){
   .bit-statband{ flex-direction:column; }
-  .bit-stat{ border-right:none; border-bottom:var(--border-width) solid var(--border-subtle); }
 }
 """

 SHELL_CSS = """
+/* ================= app shell ================= */
 .gradio-container{
   max-width:100% !important; padding:0 !important;
   background:var(--bg-canvas) !important;
   font-family:var(--font-body); font-weight:var(--weight-body);
+  color:var(--text-primary); font-size:var(--text-base);
 }
 .gradio-container *{ border-radius:var(--radius-sm) !important; }
 footer{ display:none !important; }
+.gap.svelte-vt1mxs, .gradio-container .gap{ gap:var(--space-2) !important; }
+/* ================= typography ================= */
 .bit-h1,.bit-h2,.bit-h3{
   font-family:var(--font-heading); font-weight:var(--weight-heading);
   letter-spacing:var(--tracking-wide); text-transform:uppercase;
   color:var(--text-primary); margin:0;
 }
+.bit-h1{ font-size:var(--text-md); }
 .bit-h2{ font-size:var(--text-md); }
 .bit-h3{ font-size:var(--text-base); }
 .bit-micro{
   font-family:var(--font-tiny); font-size:var(--text-xs);
   letter-spacing:var(--tracking-wider); text-transform:uppercase;
+  color:var(--text-tertiary); line-height:1.6;
 }
+.bit-spacer{ flex:1; }
+/* ================= top bar ================= */
+.bit-topbar{
+  display:flex; align-items:center; gap:var(--space-3); flex-wrap:wrap;
+  background:var(--bg-panel);
+  border-bottom:var(--border-width) solid var(--border-default);
+  padding:var(--space-3) var(--space-4);
+  position:sticky; top:0; z-index:var(--z-header);
 }
+.bit-mark{
+  width:20px; height:20px; background:var(--accent-amber);
+  display:inline-block; flex:0 0 20px;
+  box-shadow:inset 0 0 0 3px var(--bg-panel), inset 0 0 0 6px var(--accent-amber);
 }
+.bit-wordmark{
+  font-family:var(--font-heading); font-size:var(--text-md); font-weight:500;
+  letter-spacing:var(--tracking-wide); color:var(--text-primary);
 }
+.bit-slash{ color:var(--text-tertiary); }
 .bit-chip{
+  display:inline-flex; align-items:center; gap:6px; white-space:nowrap;
   font-family:var(--font-mono); font-size:var(--text-sm);
   border:var(--border-width) solid var(--border-default);
+  padding:3px var(--space-2); color:var(--text-secondary); background:transparent;
 }
 .bit-chip-ok{ color:var(--accent-moss-strong); border-color:var(--accent-moss-dim); }
 .bit-chip-run{ color:var(--accent-amber-strong); border-color:var(--accent-amber-dim); }
 .bit-chip-warn{ color:var(--fin-down); border-color:var(--fin-down); }
+.bit-chip-accent{ color:var(--stone-950); background:var(--accent-amber);
+  border-color:var(--accent-amber); }
+.bit-chip-ctx{ color:var(--text-tertiary); }
+a.bit-link{ text-decoration:none; }
+a.bit-link:hover{ color:var(--accent-amber-strong);
+  border-color:var(--accent-amber-dim); }
+/* ================= panels ================= */
+.bit-panel{
+  background:var(--bg-panel);
+  border:var(--border-width) solid var(--border-default);
+  padding:var(--space-4); margin-bottom:var(--space-3);
+}
+.bit-panel-head{
+  display:flex; align-items:baseline; gap:var(--space-3);
+  border-bottom:var(--border-width) solid var(--border-subtle);
+  padding-bottom:var(--space-2); margin-bottom:var(--space-3);
+}
+.bit-zone-title{
+  font-family:var(--font-heading); text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); font-size:var(--text-md);
+  color:var(--text-primary); padding:var(--space-3) var(--space-1) var(--space-2);
+  display:flex; align-items:baseline; gap:var(--space-2);
+}
+.bit-zone-title .bit-micro{ margin-left:auto; }
+/* ================= stat band ================= */
+.bit-statband{
+  display:flex; flex-wrap:wrap;
+  border:var(--border-width) solid var(--border-default);
+  background:var(--bg-panel); margin-bottom:var(--space-3);
+}
 .bit-stat{
+  flex:1 1 132px; padding:var(--space-3) var(--space-4);
   border-right:var(--border-width) solid var(--border-subtle);
+  border-bottom:var(--border-width) solid var(--border-subtle);
 }
 .bit-stat-label{
   font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); text-transform:uppercase;
+  color:var(--text-tertiary);
 }
 .bit-stat-value{
   font-family:var(--font-mono); font-size:var(--text-xl);
+  line-height:var(--leading-tight); color:var(--text-primary);
+  margin:3px 0 2px; letter-spacing:var(--tracking-tight);
+}
+.bit-stat-sub{
+  font-family:var(--font-mono); font-size:var(--text-xs);
+  color:var(--text-tertiary);
 }
 .bit-up{ color:var(--fin-up-strong); } .bit-down{ color:var(--fin-down-strong); }
+/* ================= notes ================= */
 .bit-note{
+  font-family:var(--font-mono); font-size:var(--text-sm); line-height:1.5;
   border-left:var(--border-width-strong) solid var(--accent-amber);
   background:var(--bg-raised); padding:var(--space-2) var(--space-3);
+  color:var(--text-secondary); margin-bottom:var(--space-2);
 }
 .bit-note-danger{ border-left-color:var(--fin-down); }
+.bit-tag{
+  font-family:var(--font-tiny); font-size:var(--text-2xs);
+  letter-spacing:var(--tracking-wider); border:1px solid var(--border-default);
+  padding:1px 4px; color:var(--text-tertiary); margin-left:4px;
+}
+/* ================= empty state ================= */
+.bit-empty{
+  display:flex; flex-direction:column; align-items:center; justify-content:center;
+  gap:var(--space-3); padding:var(--space-8) var(--space-4); text-align:center;
+  border:var(--border-width) dashed var(--border-default); background:var(--bg-panel);
+}
+.bit-empty-glyph{ font-size:34px; color:var(--accent-amber-dim); line-height:1; }
+.bit-empty-copy{ color:var(--text-secondary); max-width:46ch; font-size:var(--text-base); }
+.bit-kbd-row{ display:flex; gap:var(--space-2); flex-wrap:wrap; justify-content:center; }
+.bit-kbd{
+  font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); color:var(--text-tertiary);
+  border:var(--border-width) solid var(--border-default); padding:3px 7px;
+}
+/* ================= run manager ================= */
+.bit-run-card{
+  border:var(--border-width) solid var(--border-subtle);
+  padding:var(--space-2); margin-bottom:var(--space-1); background:var(--bg-panel);
+}
+.bit-run-card-sel{ border-color:var(--accent-amber-dim); }
+.bit-run-top{ display:flex; align-items:center; gap:var(--space-2); }
+.bit-run-name{ font-size:var(--text-sm); color:var(--text-primary);
+  overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
+.bit-run-mark{ margin-left:auto; color:var(--accent-amber-strong); font-size:11px; }
+.bit-run-sharpe{ font-family:var(--font-mono); font-size:var(--text-base);
+  display:flex; gap:var(--space-2); }
+.bit-run-ret{ margin-left:auto; color:var(--text-tertiary); }
+.bit-gloss{ margin-bottom:var(--space-2); }
+.bit-gloss-def{ font-size:var(--text-sm); color:var(--text-secondary); line-height:1.45; }
+/* ================= KPI row ================= */
+.bit-kpi-row{ display:flex; flex-wrap:wrap; gap:var(--space-4); }
+.bit-kpi{ min-width:78px; }
+.bit-kpi-value{ font-family:var(--font-mono); font-size:var(--text-lg);
+  line-height:1.1; color:var(--text-primary); }
+/* ================= podium ================= */
+.bit-podium-row{ display:flex; gap:var(--space-3); flex-wrap:wrap;
+  margin-bottom:var(--space-3); }
+.bit-podium{
+  flex:1 1 200px; border:var(--border-width) solid var(--border-default);
+  background:var(--bg-panel); padding:var(--space-3);
+  border-top:var(--border-width-strong) solid var(--border-strong);
 }
+.bit-podium-1{ border-top-color:var(--accent-amber); }
+.bit-podium-2{ border-top-color:var(--accent-moss-strong); }
+.bit-podium-3{ border-top-color:var(--mute-teal); }
+.bit-podium-rank{ font-family:var(--font-tiny); font-size:var(--text-xs);
+  color:var(--text-tertiary); letter-spacing:var(--tracking-wider); }
+.bit-podium-name{ font-size:var(--text-md); color:var(--text-primary);
+  font-family:var(--font-heading); letter-spacing:var(--tracking-tight); }
+.bit-podium-value{ font-family:var(--font-mono); font-size:var(--text-2xl);
+  line-height:1.1; margin-top:var(--space-2); display:flex;
+  align-items:baseline; gap:var(--space-2); }
+.bit-podium-unit{ font-family:var(--font-tiny); font-size:var(--text-xs);
+  letter-spacing:var(--tracking-wider); color:var(--text-tertiary); }
+/* ================= signal aggregator ================= */
+.bit-sig-grid{ display:flex; gap:var(--space-4); flex-wrap:wrap; align-items:flex-start; }
+.bit-sig-list{ flex:1 1 340px; min-width:0; }
+.bit-sig-row{
+  display:flex; align-items:center; gap:var(--space-3);
+  border:var(--border-width) solid var(--border-subtle);
+  padding:5px var(--space-2); margin-bottom:3px;
+  font-family:var(--font-mono); font-size:var(--text-sm);
+}
+.bit-sig-name{ flex:1 1 auto; color:var(--text-secondary);
+  overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
+.bit-sig-dir{ width:74px; text-align:right; }
+.bit-sig-edge{ width:64px; text-align:right; }
+.bit-sig-acc,.bit-sig-w{ width:64px; text-align:right; color:var(--text-tertiary); }
+.bit-consensus{
+  flex:0 0 210px; border:var(--border-width) solid var(--accent-moss-dim);
+  background:var(--bg-raised); padding:var(--space-3);
+}
+.bit-consensus-dir{ font-family:var(--font-heading); font-size:var(--text-2xl);
+  line-height:1.1; margin:var(--space-1) 0; letter-spacing:var(--tracking-tight); }
+.bit-conf-track{ height:4px; background:var(--bg-sunken); margin:var(--space-2) 0 4px; }
+.bit-conf-fill{ height:100%; }
+/* ================= gradio overrides ================= */
 .gradio-container .tabs > .tab-nav{
   border-bottom:var(--border-width) solid var(--border-default) !important;
+  background:transparent !important; gap:0 !important; flex-wrap:wrap;
 }
 .gradio-container .tabs > .tab-nav > button{
   font-family:var(--font-heading) !important; text-transform:uppercase;
   border:none !important; border-bottom:2px solid transparent !important;
   padding:var(--space-2) var(--space-4) !important;
 }
+.gradio-container .tabs > .tab-nav > button:hover{ color:var(--text-secondary) !important; }
 .gradio-container .tabs > .tab-nav > button.selected{
   color:var(--text-primary) !important;
   border-bottom-color:var(--accent-amber) !important;
 .gradio-container .form, .gradio-container .block{
   background:transparent !important; border:none !important;
 }
+/* Field captions only -- scoped so it never swallows option text. */
 .block > label > span, .block > .form > label > span{
   font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
   letter-spacing:var(--tracking-wider) !important; text-transform:uppercase;
   background:var(--bg-sunken) !important; color:var(--text-primary) !important;
   border:var(--border-width) solid var(--border-default) !important;
 }
+input:focus, select:focus, textarea:focus{
+  outline:1px solid var(--focus-ring) !important; outline-offset:-1px;
+}
+/* Radio / checkbox options read as segmented chips. */
 .gradio-container fieldset label,
 .gradio-container .wrap label:has(input[type="radio"]),
 .gradio-container .wrap label:has(input[type="checkbox"]){
 .gradio-container fieldset label.selected,
 .gradio-container fieldset label:has(input:checked),
 .gradio-container .wrap label:has(input[type="radio"]:checked){
+  background:var(--accent-amber) !important; border-color:var(--accent-amber) !important;
 }
 .gradio-container fieldset label.selected span,
 .gradio-container fieldset label:has(input:checked) span,
 .gradio-container input[type="radio"], .gradio-container input[type="checkbox"]{
   accent-color:var(--accent-amber);
 }
+/* Accordions read as the design's numbered sections. */
+.gradio-container .label-wrap, .bit-accordion > .label-wrap{
+  font-family:var(--font-heading) !important; text-transform:uppercase;
+  letter-spacing:var(--tracking-wide); font-size:var(--text-base) !important;
+  color:var(--text-primary) !important;
+  border-bottom:var(--border-width) solid var(--border-subtle) !important;
+  padding:var(--space-2) 0 !important;
+}
+.bit-accordion{
+  border:none !important;
+  border-top:var(--border-width) solid var(--border-subtle) !important;
+  padding:0 !important; margin-bottom:var(--space-1) !important;
+}
+/* Buttons */
 .bit-run-btn{
   background:var(--accent-amber) !important; color:var(--stone-950) !important;
   font-family:var(--font-heading) !important; text-transform:uppercase;
   border:var(--border-width) solid var(--border-default) !important;
   font-family:var(--font-mono) !important; font-size:var(--text-sm) !important;
 }
+.bit-ghost-btn:hover{ border-color:var(--border-strong) !important;
+  color:var(--text-primary) !important; }
+/* Tables */
 .bit-table table{ font-family:var(--font-mono) !important; font-size:var(--text-sm) !important; }
 .bit-table thead th{
   font-family:var(--font-tiny) !important; font-size:var(--text-xs) !important;
   letter-spacing:var(--tracking-wider); text-transform:uppercase;
   color:var(--text-tertiary) !important; background:var(--bg-raised) !important;
+  border-bottom:var(--border-width) solid var(--border-default) !important;
 }
+.bit-table tbody td{ border-color:var(--border-subtle) !important; }
+/* Footer */
+.bit-footer{
+  position:sticky; bottom:0; z-index:var(--z-header);
+  display:flex; justify-content:space-between; gap:var(--space-4);
+  background:var(--bg-panel); border-top:var(--border-width) solid var(--border-default);
+  padding:var(--space-2) var(--space-4);
+  font-family:var(--font-mono); font-size:var(--text-sm); color:var(--text-tertiary);
 }
+.bit-footer-right{ white-space:nowrap; }
 @media (max-width: 900px){
   .bit-statband{ flex-direction:column; }
+  .bit-stat{ border-right:none; }
+  .bit-sig-grid{ flex-direction:column; }
+  .bit-consensus{ flex:1 1 auto; width:100%; }
 }
 """

tests/test_adapters.py CHANGED Viewed

@@ -60,7 +60,7 @@ def test_unknown_adapter_family_is_refused():
 def test_allowed_families_are_exactly_the_configured_set():
-    assert set(config.ALLOWED_ADAPTER_FAMILIES) == {"chronos", "timesfm"}
 @pytest.mark.parametrize("bad", [

 def test_allowed_families_are_exactly_the_configured_set():
+    assert set(config.ALLOWED_ADAPTER_FAMILIES) == {"chronos", "timesfm", "baseline"}
 @pytest.mark.parametrize("bad", [

tests/test_catalog.py ADDED Viewed

	@@ -0,0 +1,312 @@

+"""Tests for the catalog: the global comparison backend.
+Covers the leaderboard sweep, precomputed equity curves, the signal scorecard,
+the baseline adapters, and the consensus aggregator.
+"""
+from __future__ import annotations
+import sys
+from pathlib import Path
+import numpy as np
+import pandas as pd
+import pytest
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+from src import catalog, charts, config
+from src.adapters import AdapterError, BaselineAdapter, build_windows, get_adapter
+from src.metrics import directional_accuracy
+from src.store import SignalStore, validate_signal_frame
+from src.ui import compare_tab as CT
+def prices_frame(n=500, seed=3, start="2024-01-01"):
+    rng = np.random.default_rng(seed)
+    close = pd.Series(100 * np.exp(np.cumsum(rng.normal(0.0004, 0.02, n))))
+    ts = pd.date_range(start, periods=n, freq="D", tz="UTC")
+    return pd.DataFrame({
+        "ts": ts, "open": close * 0.999, "high": close * 1.02,
+        "low": close * 0.98, "close": close, "volume": 1000.0, "source": "test",
+    })
+@pytest.fixture
+def seeded_store(tmp_path):
+    """A small offline store with prices and two baseline models' signals."""
+    store = SignalStore(repo_id=None, local_root=tmp_path / "store", offline=True)
+    for asset in ("BTC-USD", "ETH-USD"):
+        px = prices_frame()
+        store.write_prices(asset, "1d", px)
+        close = px.set_index("ts")["close"]
+        for slug, method in (("baseline-naive", "naive"), ("baseline-drift", "drift")):
+            a = BaselineAdapter(f"baseline/{method}", context_len=100).load()
+            stamps, wins = build_windows(close, 100)
+            frame = a.predict(wins).as_frame(stamps, a.inference_version())
+            store.write_signals(slug, f"baseline/{method}", a.resolved_revision,
+                                asset, "1d", frame,
+                                inference_version=a.inference_version())
+    return store
+# ---------------------------- baselines ----------------------------
+@pytest.mark.parametrize("method", ["naive", "drift", "seasonal"])
+def test_baselines_produce_valid_signals(method):
+    a = get_adapter("baseline", f"baseline/{method}", context_len=64).load()
+    s = pd.Series(np.linspace(100, 200, 300),
+                  index=pd.date_range("2024-01-01", periods=300, freq="D", tz="UTC"))
+    stamps, wins = build_windows(s, 64)
+    out = validate_signal_frame(
+        a.predict(wins[:40]).as_frame(stamps[:40], a.inference_version()))
+    assert len(out) == 40
+    assert ((out["q10"] <= out["q50"]) & (out["q50"] <= out["q90"])).all()
+def test_baseline_rejects_unknown_method():
+    with pytest.raises(AdapterError, match="unknown baseline method"):
+        get_adapter("baseline", "baseline/crystal-ball")
+def test_naive_baseline_predicts_the_last_value():
+    a = BaselineAdapter("baseline/naive", context_len=10).load()
+    assert a.predict(np.arange(1, 11, dtype="float64")[None, :]).q50[0] == pytest.approx(10.0)
+def test_drift_baseline_extrapolates_the_window_slope():
+    a = BaselineAdapter("baseline/drift", context_len=11).load()
+    assert a.predict(np.arange(0, 11, dtype="float64")[None, :]).q50[0] == pytest.approx(11.0)
+def test_baselines_are_deterministic():
+    a = BaselineAdapter("baseline/drift", context_len=32).load()
+    _, wins = build_windows(pd.Series(
+        np.random.default_rng(1).normal(100, 5, 200),
+        index=pd.date_range("2024-01-01", periods=200, freq="D", tz="UTC")), 32)
+    assert np.allclose(a.predict(wins[:20]).q50, a.predict(wins[:20]).q50)
+def test_baseline_revision_is_pinned():
+    a = BaselineAdapter("baseline/naive").load()
+    assert a.resolved_revision.startswith("baseline-")
+    assert a.inference_version() != config.PLACEHOLDER_VERSION
+def test_flat_forecast_has_undefined_direction_not_zero_accuracy():
+    """A random walk never claims a direction, so accuracy is NaN, not 0%."""
+    idx = pd.date_range("2024-01-01", periods=80, freq="D", tz="UTC")
+    ref = pd.Series(np.linspace(100, 160, 80), index=idx)
+    assert pd.isna(directional_accuracy(ref.shift(-1).ffill(), ref.copy(), ref))
+def test_chronos_chunking_is_smaller_for_sampling_models():
+    bolt = get_adapter("chronos", "amazon/chronos-bolt-small")
+    t5 = get_adapter("chronos", "amazon/chronos-t5-small")
+    assert bolt.chunk_size > t5.chunk_size
+    assert bolt._is_bolt and not t5._is_bolt
+# ---------------------------- catalog build ----------------------------
+def test_build_produces_all_three_artifacts(seeded_store):
+    rep = catalog.build(seeded_store)
+    assert rep.rows > 0 and rep.curves > 0 and rep.scorecard_rows > 0
+    assert not rep.failed
+    assert not catalog.load_leaderboard(seeded_store).empty
+    assert not catalog.load_equity_curves(seeded_store).empty
+    assert not catalog.load_scorecard(seeded_store).empty
+def test_leaderboard_covers_signal_and_non_signal_strategies(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    assert (lb["model_slug"] == "").any()
+    assert (lb["model_slug"] != "").any()
+def test_signal_strategies_fan_out_over_every_model(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    ff = lb[lb["strategy"] == "Chronos Forecast Follower"]
+    assert set(ff["model_slug"]) == {"baseline-naive", "baseline-drift"}
+def test_every_row_carries_a_unique_key(seeded_store):
+    catalog.build(seeded_store)
+    assert catalog.load_leaderboard(seeded_store)["key"].is_unique
+def test_thin_evidence_is_flagged(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    assert (lb["significant"] == (lb["trades"] >= catalog.MIN_MEANINGFUL_TRADES)).all()
+def test_ranking_excludes_thin_rows_when_asked(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    strict = catalog.rank(lb, "oos_sharpe", significant_only=True)
+    assert (strict["trades"] >= catalog.MIN_MEANINGFUL_TRADES).all()
+def test_least_bad_drawdown_ranks_first():
+    """Drawdowns are negative, so descending order puts the shallowest on top."""
+    df = pd.DataFrame({"max_drawdown": [-0.5, -0.1, -0.3],
+                       "significant": True, "trades": 50})
+    assert catalog.rank(df, "max_drawdown").iloc[0]["max_drawdown"] == -0.1
+def test_drawdown_ranking_in_the_compare_view_puts_shallowest_first(seeded_store):
+    catalog.build(seeded_store)
+    _p, table, *_ = CT.build_leaderboard_view(
+        seeded_store, assets=None, timeframes=None, strategies_=None, models=None,
+        metric_label="Max drawdown (least bad)", min_trades=0,
+        hide_baselines=False, require_oos=False, top_n=5)
+    vals = [float(v.rstrip("%").replace("+", "")) for v in table["Max DD"]]
+    assert vals == sorted(vals, reverse=True)
+def test_curves_for_returns_named_series(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    curves = catalog.curves_for(catalog.load_equity_curves(seeded_store),
+                                list(lb["key"])[:3])
+    assert len(curves) == 3
+    for s in curves.values():
+        assert isinstance(s, pd.Series) and len(s) > 0
+def test_curves_for_unknown_key_is_empty(seeded_store):
+    catalog.build(seeded_store)
+    assert catalog.curves_for(catalog.load_equity_curves(seeded_store), ["nope"]) == {}
+def test_filters_narrow_the_board(seeded_store):
+    catalog.build(seeded_store)
+    lb = catalog.load_leaderboard(seeded_store)
+    assert set(catalog.filter_leaderboard(lb, assets=["BTC-USD"])["asset"]) == {"BTC-USD"}
+    assert not catalog.filter_leaderboard(lb, hide_baselines=True)["is_baseline_model"].any()
+def test_catalog_meta_records_the_canonical_config(seeded_store):
+    catalog.build(seeded_store)
+    meta = catalog.catalog_meta(seeded_store)
+    assert meta["leaderboard_rows"] > 0
+    assert "walk-forward" in meta["canonical_config"]
+# ---------------------------- scorecard & consensus ----------------------------
+def test_scorecard_labels_baselines(seeded_store):
+    catalog.build(seeded_store)
+    sc = catalog.load_scorecard(seeded_store)
+    assert sc["is_baseline"].all()
+    assert {"coverage_q10_q90", "directional_accuracy", "beats_momentum"} <= set(sc.columns)
+def test_consensus_lists_every_model_with_a_direction(seeded_store):
+    catalog.build(seeded_store)
+    cons = catalog.model_consensus(catalog.load_scorecard(seeded_store),
+                                   seeded_store, "BTC-USD", "1d")
+    assert len(cons) == 2
+    assert set(cons["direction"]) <= {"LONG", "SHORT", "FLAT"}
+    assert (cons["weight"] >= 0).all()
+def test_consensus_verdict_aggregates(seeded_store):
+    catalog.build(seeded_store)
+    cons = catalog.model_consensus(catalog.load_scorecard(seeded_store),
+                                   seeded_store, "BTC-USD", "1d")
+    v = catalog.consensus_verdict(cons)
+    assert v["direction"] in ("LONG", "SHORT", "FLAT")
+    assert 0.0 <= v["confidence"] <= 1.0
+    assert v["n_models"] == len(cons)
+def test_consensus_on_an_uncovered_slice_is_empty(seeded_store):
+    sc = catalog.load_scorecard(seeded_store)
+    assert catalog.model_consensus(sc, seeded_store, "SOL-USD", "1d").empty
+    assert catalog.consensus_verdict(pd.DataFrame())["direction"] == "NO DATA"
+# ---------------------------- compare tab ----------------------------
+def test_leaderboard_view_renders_end_to_end(seeded_store):
+    catalog.build(seeded_store)
+    podium, table, overlay, scatter, meta = CT.build_leaderboard_view(
+        seeded_store, assets=None, timeframes=None, strategies_=None, models=None,
+        metric_label="OOS Sharpe", min_trades=0, hide_baselines=False,
+        require_oos=False, top_n=10)
+    assert "bit-podium" in podium
+    assert not table.empty
+    assert len(overlay.data) > 0 and len(scatter.data) > 0
+    assert "rows match" in meta
+def test_leaderboard_view_handles_an_empty_catalog(tmp_path):
+    empty = SignalStore(repo_id=None, local_root=tmp_path / "s", offline=True)
+    podium, table, *_ = CT.build_leaderboard_view(
+        empty, assets=None, timeframes=None, strategies_=None, models=None,
+        metric_label="OOS Sharpe", min_trades=0, hide_baselines=False,
+        require_oos=False, top_n=10)
+    assert "not been generated" in podium and table.empty
+def test_models_view_renders(seeded_store):
+    catalog.build(seeded_store)
+    _note, acc, cal, _bars, table = CT.build_models_view(seeded_store, "1d")
+    assert acc is not None and cal is not None
+    assert not table.empty
+def test_signals_view_renders(seeded_store):
+    catalog.build(seeded_store)
+    html = CT.build_signals_view(seeded_store, "BTC-USD", "1d")
+    assert "bit-sig-row" in html or "No model signals" in html
+def test_runs_table_merges_sources():
+    saved = pd.DataFrame([{
+        "created_at": "2026-01-01T00:00:00", "label": "saved one",
+        "strategy": "SMA Crossover", "asset": "BTC-USD", "timeframe": "1d",
+        "total_return": 0.2, "sharpe": 1.1, "oos_sharpe": 0.9,
+        "max_drawdown": -0.1, "trades": 30,
+    }])
+    out = CT.runs_table([], saved)
+    assert len(out) == 1 and out["Source"].iloc[0] == "signal store"
+def test_leaderboard_table_flags_thin_rows():
+    df = pd.DataFrame([{
+        "strategy": "X", "model_display": "—", "asset": "BTC-USD", "timeframe": "1d",
+        "oos_sharpe": 4.0, "sharpe": 4.0, "total_return": 1.0, "cagr": 0.5,
+        "max_drawdown": -0.1, "win_rate": 0.6, "profit_factor": 2.0, "trades": 3,
+        "excess_vs_hold": 0.2, "holdout_sharpe": 1.0, "costs_paid": 10.0,
+        "significant": False,
+    }])
+    assert "trades" in CT.leaderboard_table(df)["Note"].iloc[0]
+# ---------------------------- charts ----------------------------
+def test_catalog_charts_survive_empty_input():
+    import plotly.graph_objects as go
+    for fig in (charts.multi_return_overlay({}),
+                charts.risk_return_scatter(pd.DataFrame()),
+                charts.model_accuracy_bars(pd.DataFrame()),
+                charts.calibration_scatter(pd.DataFrame()),
+                charts.model_leaderboard_bars(pd.DataFrame())):
+        assert isinstance(fig, go.Figure)
+def test_overlay_caps_the_number_of_series():
+    idx = pd.date_range("2024-01-01", periods=50, freq="D", tz="UTC")
+    curves = {f"s{i}": pd.Series(np.linspace(0, 1, 50), index=idx) for i in range(40)}
+    assert len(charts.multi_return_overlay(curves, max_series=8).data) == 8

tests/test_ui.py CHANGED Viewed

@@ -18,6 +18,8 @@ sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 import app as bitapp
 from src import charts, comparisons, config, runtime, strategies
 from src.runtime import RunRequest
@@ -70,14 +72,14 @@ def test_theme_css_carries_the_design_tokens():
 def test_disclaimer_is_present_in_the_footer():
-    assert "not indicative of future results" in bitapp.FOOTER_HTML
-    assert "Not financial advice" in bitapp.FOOTER_HTML or \
-           "not a licensed investment adviser" in bitapp.FOOTER_HTML
 def test_empty_state_offers_the_worked_example():
-    assert "No run loaded" in bitapp.EMPTY_HTML
-    assert "worked example" in bitapp.EMPTY_HTML
 def test_glossary_covers_every_design_term():
@@ -92,29 +94,29 @@ def test_glossary_covers_every_design_term():
 def test_stat_band_values_match_engine_metrics(rec):
-    html = bitapp.stat_band_html(rec)
     m = rec.result.metrics_all
-    assert bitapp.pct(m.total_return) in html
-    assert bitapp.pct(m.cagr) in html
-    assert bitapp.num(m.sharpe) in html
-    assert bitapp.num(m.sortino) in html
-    assert bitapp.pct(m.max_drawdown) in html
-    assert bitapp.num(m.profit_factor) in html
     assert f">{m.trade_count}<" in html
 def test_stat_band_shows_is_and_oos_for_every_stat(rec):
-    html = bitapp.stat_band_html(rec)
     assert html.count("IS ") >= 9
     assert html.count("· OOS") >= 9
-    assert bitapp.num(rec.result.metrics_oos.sharpe) in html
-    assert bitapp.num(rec.result.metrics_is.sharpe) in html
 def test_stat_band_reports_costs_actually_paid(rec):
-    html = bitapp.stat_band_html(rec)
-    assert bitapp.money(rec.result.costs_paid) in html
     assert rec.result.costs_paid > 0, "costs default to ON, so this must be positive"
@@ -124,7 +126,7 @@ def test_empty_segment_renders_an_em_dash_not_a_zero():
         strategy="SMA Crossover", asset="BTC-USD", timeframe="1h", date_range="1Y"))
     if short.result.metrics_oos.bars:
         pytest.skip("this range did produce OOS windows")
-    html = bitapp.stat_band_html(short)
     assert "· OOS —" in html
     assert "no out-of-sample windows" in html
@@ -141,8 +143,8 @@ def test_trade_table_rows_match_the_engine_trade_list(rec):
 def test_report_quotes_the_same_numbers_as_the_stat_band(rec):
     md = bitapp.report_markdown(rec)
-    assert bitapp.pct(rec.result.metrics_all.total_return) in md
-    assert bitapp.money(rec.result.costs_paid) in md
     assert rec.run_id in md

 import app as bitapp
 from src import charts, comparisons, config, runtime, strategies
+from src.ui import components as C
+from src.ui.format import money, num, pct
 from src.runtime import RunRequest
 def test_disclaimer_is_present_in_the_footer():
+    assert "not indicative of future results" in C.footer()
+    assert "Not financial advice" in C.footer() or \
+           "not a licensed investment adviser" in C.footer()
 def test_empty_state_offers_the_worked_example():
+    assert "No run loaded" in C.empty_state()
+    assert "worked example" in C.empty_state()
 def test_glossary_covers_every_design_term():
 def test_stat_band_values_match_engine_metrics(rec):
+    html = C.stat_band(rec)
     m = rec.result.metrics_all
+    assert pct(m.total_return) in html
+    assert pct(m.cagr) in html
+    assert num(m.sharpe) in html
+    assert num(m.sortino) in html
+    assert pct(m.max_drawdown) in html
+    assert num(m.profit_factor) in html
     assert f">{m.trade_count}<" in html
 def test_stat_band_shows_is_and_oos_for_every_stat(rec):
+    html = C.stat_band(rec)
     assert html.count("IS ") >= 9
     assert html.count("· OOS") >= 9
+    assert num(rec.result.metrics_oos.sharpe) in html
+    assert num(rec.result.metrics_is.sharpe) in html
 def test_stat_band_reports_costs_actually_paid(rec):
+    html = C.stat_band(rec)
+    assert money(rec.result.costs_paid) in html
     assert rec.result.costs_paid > 0, "costs default to ON, so this must be positive"
         strategy="SMA Crossover", asset="BTC-USD", timeframe="1h", date_range="1Y"))
     if short.result.metrics_oos.bars:
         pytest.skip("this range did produce OOS windows")
+    html = C.stat_band(short)
     assert "· OOS —" in html
     assert "no out-of-sample windows" in html
 def test_report_quotes_the_same_numbers_as_the_stat_band(rec):
     md = bitapp.report_markdown(rec)
+    assert pct(rec.result.metrics_all.total_return) in md
+    assert money(rec.result.costs_paid) in md
     assert rec.run_id in md