Datasets:
File size: 16,918 Bytes
149513d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 | #!/usr/bin/env python3
"""Build the artifact release bundle.
Copies all paper-referenced run artifacts into release_data/ with a
human-readable directory structure and produces:
- release_data/manifest.json
- release_data/derived/all_trades.csv
- release_data/derived/all_metrics.csv
By default, the script infers the repository root from its installed
location under release_data/tools/. For nonstandard layouts, set
QUANTARENA_REPO or QUANTARENA_INGEST_DIR explicitly.
"""
from __future__ import annotations
import csv
import json
import os
import shutil
from pathlib import Path
from typing import Iterable
SCRIPT_DIR = Path(__file__).resolve().parent
REPO = Path(os.environ.get("QUANTARENA_REPO", SCRIPT_DIR.parents[1])).resolve()
SRC_BACKTEST = REPO / "reports" / "backtest"
SRC_MP = REPO / "reports" / "multi_personality"
SRC_INGEST = Path(
os.environ.get("QUANTARENA_INGEST_DIR", REPO / "latex" / "data" / "artifact_ingest")
).resolve()
SRC_DOCS = REPO / "latex" / "docs"
SRC_AUDIT = REPO / "latex" / "audit"
RELEASE = Path(os.environ.get("QUANTARENA_RELEASE_DIR", REPO / "release_data")).resolve()
# (display_name, mandate_dirname, run_id)
EXP1_US = [
("Fundamental Value", "fundamental_value", "mp_fundamental_value_20260409_165131_876468"),
("Macro Tactical", "macro_tactical", "mp_macro_tactical_20260409_165131_876505"),
("Behavioral Momentum", "behavioral_momentum", "mp_behavioral_momentum_20260409_165131_876499"),
("Low-Volatility", "low_volatility", "mp_smart_beta_passive_20260409_165131_876510"),
("Equal-Weight", "equal_weight", "mp_equal_weight_index_20260409_165131_876514"),
]
EXP1_CN = [
("Fundamental Value", "fundamental_value", "mp_fundamental_value_20260413_163905_878961"),
("Macro Tactical", "macro_tactical", "mp_macro_tactical_20260413_163905_878979"),
("Behavioral Momentum", "behavioral_momentum", "mp_behavioral_momentum_20260413_163905_878974"),
("Low-Volatility", "low_volatility", "mp_smart_beta_passive_20260413_163905_878984"),
("Equal-Weight", "equal_weight", "mp_equal_weight_index_20260413_163905_878987"),
]
EXP2_US_R2 = [
("Fundamental Value", "fundamental_value", "mp_fundamental_value_20260425_035029_413185"),
("Macro Tactical", "macro_tactical", "mp_macro_tactical_20260425_035029_413208"),
("Behavioral Momentum", "behavioral_momentum", "mp_behavioral_momentum_20260425_035029_413202"),
("Low-Volatility", "low_volatility", "mp_smart_beta_passive_20260425_035029_413212"),
("Equal-Weight", "equal_weight", "mp_equal_weight_index_20260425_035029_413216"),
]
# (display, mandate-variant dir, run_id, mandate, variant)
EXP3 = [
("FV Full", "fundamental_value_full", "20260417_160942", "fundamental_value", "full"),
("FV No filter", "fundamental_value_no_filter", "20260417_191443", "fundamental_value", "no_filter"),
("BM Full", "behavioral_momentum_full", "20260417_220721", "behavioral_momentum", "full"),
("BM No guardrails", "behavioral_momentum_no_guardrails","20260418_004927", "behavioral_momentum", "no_guardrails"),
("MT Full", "macro_tactical_full", "20260418_033940", "macro_tactical", "full"),
("MT No tilt", "macro_tactical_no_tilt", "20260418_072633", "macro_tactical", "no_tilt"),
("LV Reference", "low_volatility_reference", "20260418_110947", "low_volatility", "reference"),
("EqW Reference", "equal_weight_reference", "20260418_111210", "equal_weight", "reference"),
]
EXP4 = [
("Fundamental Value", "fundamental_value", "20260423_124148"),
("Macro Tactical", "macro_tactical", "20260424_094955"),
("Behavioral Momentum", "behavioral_momentum", "20260424_132728_372104"),
("Low-Volatility", "low_volatility", "20260424_094956"),
("Equal-Weight", "equal_weight", "20260424_132723_271683"),
]
def copy_run(src_run_id: str, dst_dir: Path) -> dict:
"""Copy a run dir's artifacts; return summary dict."""
src = SRC_BACKTEST / src_run_id
if not src.exists():
return {"src_run_id": src_run_id, "status": "missing"}
dst_dir.mkdir(parents=True, exist_ok=True)
files_copied = []
for f in src.iterdir():
if f.is_file():
shutil.copy2(f, dst_dir / f.name)
files_copied.append(f.name)
# Record raw run metadata
metrics_file = dst_dir / "metrics.json"
if metrics_file.exists():
m = json.loads(metrics_file.read_text())
return {
"src_run_id": src_run_id,
"start_date": m.get("start_date"),
"end_date": m.get("end_date"),
"market": m.get("market"),
"personality": m.get("config", {}).get("personality"),
"total_return": m["metrics"].get("total_return"),
"total_trades": m["metrics"].get("total_trades"),
"files": files_copied,
"status": "ok",
}
return {"src_run_id": src_run_id, "files": files_copied, "status": "ok"}
def build_manifest() -> dict:
"""Build the full manifest while copying everything."""
manifest: dict = {
"version": "1.0",
"title": "QuantArena artifact bundle",
"source_commit": "a4bf9e6 (sector matrix fix) / f2c9921 (audit manifest)",
"experiments": {},
}
# --- Exp 1 US ---
exp1_us = {"description": "Six-month main case study (US, Sep 1 2025 – Feb 28 2026, 124 trading days, $100,000 initial)", "runs": {}}
for label, dirn, rid in EXP1_US:
info = copy_run(rid, RELEASE / "runs" / "exp1_caseStudy_us_6m" / dirn)
info["display_name"] = label
info["bundle_path"] = f"runs/exp1_caseStudy_us_6m/{dirn}"
exp1_us["runs"][dirn] = info
manifest["experiments"]["exp1_caseStudy_us_6m"] = exp1_us
# --- Exp 1 CN ---
exp1_cn = {"description": "Six-month main case study (CN A-share, Sep 1 2025 – Feb 28 2026, 102 trading days, $100,000 initial)", "runs": {}}
for label, dirn, rid in EXP1_CN:
info = copy_run(rid, RELEASE / "runs" / "exp1_caseStudy_cn_6m" / dirn)
info["display_name"] = label
info["bundle_path"] = f"runs/exp1_caseStudy_cn_6m/{dirn}"
exp1_cn["runs"][dirn] = info
manifest["experiments"]["exp1_caseStudy_cn_6m"] = exp1_cn
# --- Exp 2 reproducibility ---
exp2 = {"description": "US 6M reproducibility re-run (Run 2). Independent re-run of all five mandates with identical config to Exp 1.", "runs": {}}
for label, dirn, rid in EXP2_US_R2:
info = copy_run(rid, RELEASE / "runs" / "exp2_reproducibility_us_6m_run2" / dirn)
info["display_name"] = label
info["bundle_path"] = f"runs/exp2_reproducibility_us_6m_run2/{dirn}"
exp2["runs"][dirn] = info
manifest["experiments"]["exp2_reproducibility_us_6m_run2"] = exp2
# --- Exp 3 mechanism ablation ---
exp3 = {"description": "US 3M mechanism ablation (Dec 1 2025 – Feb 28 2026). For each LLM-conditioned mandate, a Full variant and one Ablated variant.", "runs": {}}
for label, dirn, rid, mandate, variant in EXP3:
info = copy_run(rid, RELEASE / "runs" / "exp3_mechanism_ablation_us_3m" / dirn)
info["display_name"] = label
info["mandate"] = mandate
info["variant"] = variant
info["bundle_path"] = f"runs/exp3_mechanism_ablation_us_3m/{dirn}"
exp3["runs"][dirn] = info
manifest["experiments"]["exp3_mechanism_ablation_us_3m"] = exp3
# --- Exp 4 backend robustness ---
exp4 = {"description": "US 3M backend robustness: GPT-5.4 replaces DeepSeek-V3.2 reasoning backend; everything else fixed.", "runs": {}}
for label, dirn, rid in EXP4:
info = copy_run(rid, RELEASE / "runs" / "exp4_backend_robustness_us_3m_gpt54" / dirn)
info["display_name"] = label
info["bundle_path"] = f"runs/exp4_backend_robustness_us_3m_gpt54/{dirn}"
exp4["runs"][dirn] = info
manifest["experiments"]["exp4_backend_robustness_us_3m_gpt54"] = exp4
# --- Exp 5: docs only ---
exp5_dir = RELEASE / "exp5_efficiency_ablation_cn_10t_6m"
exp5_dir.mkdir(parents=True, exist_ok=True)
src_md = SRC_DOCS / "2026-03-23-cn-10tickers-6m-efficiency-ablation.md"
if src_md.exists():
shutil.copy2(src_md, exp5_dir / "source_doc.md")
(exp5_dir / "README.md").write_text(
"# Exp 5 — Execution efficiency ablation (CN, 10 tickers, 6M)\n\n"
"This experiment compares three execution paths (E0 baseline / E1 cold-cache / E2 warm-cache) "
"to demonstrate runtime/cost reductions enabled by shared phase-one preprocessing.\n\n"
"The original run artifacts are not redistributed in this bundle. "
"All numbers cited in the paper come from `source_doc.md` (copied here verbatim from "
"`latex/docs/2026-03-23-cn-10tickers-6m-efficiency-ablation.md`).\n"
)
manifest["experiments"]["exp5_efficiency_ablation_cn_10t_6m"] = {
"description": "Execution efficiency ablation (E0/E1/E2). Documented only — run artifacts not redistributed.",
"source_doc": "exp5_efficiency_ablation_cn_10t_6m/source_doc.md",
}
# --- Comparison bundles ---
comp_us = SRC_MP / "20260409_165121_176781"
comp_cn = SRC_MP / "20260413_163854_715331"
for label, src in [("us_6m", comp_us), ("cn_6m", comp_cn)]:
dst = RELEASE / "comparisons" / label
dst.mkdir(parents=True, exist_ok=True)
for f in src.iterdir():
if f.is_file():
shutil.copy2(f, dst / f.name)
manifest["comparisons"] = {
"us_6m": {
"description": "US 6M main case-study cross-mandate comparison bundle",
"source": "reports/multi_personality/20260409_165121_176781",
"bundle_path": "comparisons/us_6m",
},
"cn_6m": {
"description": "CN 6M main case-study cross-mandate comparison bundle",
"source": "reports/multi_personality/20260413_163854_715331",
"bundle_path": "comparisons/cn_6m",
},
}
# --- Universe ---
universe_dst = RELEASE / "universe"
universe_dst.mkdir(parents=True, exist_ok=True)
universe_src = SRC_INGEST / "universe" / "sector_style_universe.csv"
if universe_src.exists():
shutil.copy2(universe_src, universe_dst / "sector_style_universe.csv")
manifest["universe"] = {
"description": "5x4 sector/style universe (20 US + 20 CN tickers)",
"bundle_path": "universe/sector_style_universe.csv",
}
# --- GPT-5.4 robustness CSVs ---
gpt54_src = SRC_INGEST / "gpt54_robustness"
gpt54_dst = RELEASE / "derived" / "gpt54_robustness"
gpt54_dst.mkdir(parents=True, exist_ok=True)
if gpt54_src.exists():
for f in gpt54_src.iterdir():
if f.is_file():
shutil.copy2(f, gpt54_dst / f.name)
manifest["derived"] = {
"gpt54_robustness": "Pre-aggregated CSVs for the GPT-5.4 backend robustness analysis",
}
# --- Audit (mirror) ---
audit_dst = RELEASE / "audit"
audit_dst.mkdir(parents=True, exist_ok=True)
if SRC_AUDIT.exists():
for f in SRC_AUDIT.iterdir():
if f.is_file():
shutil.copy2(f, audit_dst / f.name)
manifest["audit"] = "Reproducibility manifest (mirror of latex/audit/)"
return manifest
def write_all_metrics(manifest: dict) -> int:
"""Write derived/all_metrics.csv as a long-format flat table."""
rows = []
metric_keys = [
"total_return", "annualized_return", "max_drawdown", "max_drawdown_duration",
"sharpe_ratio", "sortino_ratio", "volatility", "total_trades", "trading_days",
"avg_position_days", "win_rate", "initial_cash", "final_value", "final_cash",
"benchmark_total_return", "excess_return", "tracking_error", "information_ratio",
"beta", "alpha", "avg_cash_ratio", "avg_gross_exposure",
"value_filter_pass_rate", "value_consistency_score",
]
for exp_name, exp in manifest["experiments"].items():
if "runs" not in exp:
continue
for run_dirname, info in exp["runs"].items():
metrics_path = RELEASE / info["bundle_path"] / "metrics.json"
if not metrics_path.exists():
continue
m = json.loads(metrics_path.read_text())
base = {
"experiment": exp_name,
"mandate_dir": run_dirname,
"display_name": info.get("display_name"),
"src_run_id": info.get("src_run_id"),
"market": m.get("market"),
"start_date": m.get("start_date"),
"end_date": m.get("end_date"),
"personality": m.get("config", {}).get("personality"),
}
mandate = info.get("mandate")
variant = info.get("variant")
if mandate is not None:
base["mandate"] = mandate
if variant is not None:
base["variant"] = variant
for k in metric_keys:
base[k] = m["metrics"].get(k)
rows.append(base)
if not rows:
return 0
out = RELEASE / "derived" / "all_metrics.csv"
out.parent.mkdir(parents=True, exist_ok=True)
fieldnames = list(rows[0].keys())
# Ensure stable column order across rows
all_keys = sorted({k for r in rows for k in r.keys()},
key=lambda k: (k not in fieldnames, fieldnames.index(k) if k in fieldnames else 0))
# simpler: use union with priority order
priority = ["experiment", "market", "mandate_dir", "display_name", "mandate", "variant",
"src_run_id", "personality", "start_date", "end_date"] + metric_keys
all_keys = [k for k in priority if any(k in r for r in rows)]
with out.open("w") as f:
w = csv.DictWriter(f, fieldnames=all_keys, extrasaction="ignore")
w.writeheader()
for r in rows:
w.writerow(r)
return len(rows)
def write_all_trades(manifest: dict) -> int:
"""Concatenate all trades.csv into a single flat table."""
rows = []
for exp_name, exp in manifest["experiments"].items():
if "runs" not in exp:
continue
for run_dirname, info in exp["runs"].items():
trades_path = RELEASE / info["bundle_path"] / "trades.csv"
if not trades_path.exists():
continue
with trades_path.open() as f:
for r in csv.DictReader(f):
rows.append({
"experiment": exp_name,
"mandate_dir": run_dirname,
"display_name": info.get("display_name"),
"src_run_id": info.get("src_run_id"),
**r,
})
if not rows:
return 0
out = RELEASE / "derived" / "all_trades.csv"
out.parent.mkdir(parents=True, exist_ok=True)
fieldnames = ["experiment", "mandate_dir", "display_name", "src_run_id",
"date", "ticker", "action", "shares", "price", "value", "justification"]
with out.open("w") as f:
w = csv.DictWriter(f, fieldnames=fieldnames, extrasaction="ignore")
w.writeheader()
for r in rows:
w.writerow(r)
return len(rows)
def main() -> None:
if RELEASE.exists():
print(f"Cleaning existing {RELEASE}")
shutil.rmtree(RELEASE)
RELEASE.mkdir(parents=True, exist_ok=True)
print("Building manifest and copying artifacts...")
manifest = build_manifest()
(RELEASE / "manifest.json").write_text(json.dumps(manifest, indent=2, default=str))
print("Building derived/all_metrics.csv...")
n_metrics = write_all_metrics(manifest)
print(f" -> {n_metrics} rows")
print("Building derived/all_trades.csv...")
n_trades = write_all_trades(manifest)
print(f" -> {n_trades} rows")
# Final tree summary
print("\nFinal layout:")
total_size = 0
total_files = 0
for p in sorted(RELEASE.rglob("*")):
if p.is_file():
total_files += 1
total_size += p.stat().st_size
print(f" {total_files} files, {total_size/1024/1024:.2f} MB")
# v1 intentionally excludes benchmark-relative fields; see audit/README.md
# "Metric scope (v1)".
import subprocess as _sp
import sys as _sys
strip_script = SCRIPT_DIR / "strip_benchmark_relative_metrics.py"
if strip_script.exists():
print()
print("Running strip_benchmark_relative_metrics.py to enforce v1 schema...")
_sp.run([_sys.executable, str(strip_script)], check=True)
if __name__ == "__main__":
main()
|