Intraday: re-evaluate AI + ML calls when live price hits target

perf(ml-predict): skip intraday 15m download when market closed

fix: stop ThreadPoolExecutor wait=True hang on watchlist news pre-warm and predict I/O pool

Add stale-PENDING give-up safeguard for validation backlog

Fix Flask single-threaded blocking and penny-stock picks

Fix stuck backdated INTRADAY validations: add daily-OHLC fallback

AI range: containment mode + Target-below-range fix + ML un-floor

Fix wide NEUTRAL bands, live/pre-market target re-anchoring, stop-loss guard; add ₹ range to ML row

Force 1D range-only, volatility-scaled ML caps, sector-mode Top Picks, self-learning throttle + collapsible panel

Intraday: 1% min-move gate (AI+ML), pre-market preview, momentum override, top-pick INTRADAY auto-refresh; retire 3D/5D from live view+API

Self-learning: fix N=0 (missing validation cols), all-history window, why-missed notes; INTRADAY final-call cache after close

ML intraday-high guard, yfinance thread-safety, validation consistency

Restructure ML card, split AI/ML self-learning, fix Kelly block + news order + intraday-only refresh

Fix ML intraday cache, add ML vs AI risk/confidence UI, floor Kelly limit

Block INTRADAY ML after market close + clearer ML unavailable messaging

Fix AI conflicts: persistent retry (no terminal 'AI unavailable'), consistent direction/reasoning, live-anchored 1D/3D targets + gap note

Top Picks: progressive streaming with per-card spinners + HF logs

Fix empty universe + full-market top picks; drop dead NSE price APIs

fix: live price AI context, top5 IST midnight reset, remove Primary Call UI

feat: Gemini/SambaNova providers, ATR power-law range model, crash-guard triggers, backtest retry rounds

fix: realistic AI ranges, Ollama cold-start warmup, price-target prompt

fix: 320s TF timeout, per-TF cell IDs, auto-retry timed-out cells

fix: handle timeout reason in UI, extend deadline for Ollama-only mode

feat: cost-aware predictions, graded validation, per-TF AI, drop 5D

fix: deadlock in top5 route — _start_top5_background re-acquired the lock held by caller

fix: add has_ai_unavailable field to bulk watchlist-picks response

fix: limit bulk watchlist to 3 concurrent tickers to avoid provider rate-limit cascade

fix: remove news LLM from warm phase in bulk watchlist — was exhausting providers before AI forecast

fix: cap watchlist warm phase at 40s and prediction deadline at 160s to stay under timeout

fix: add 200s server-side deadline to bulk watchlist-picks, return partial results

fix: use fast_mode=True for bulk watchlist-picks to avoid 108-call rate limit storm

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

fix: yfinance first on HF Spaces; raise executor timeout 90→150s

fix: dynamic provider reorder + OHLCV cache path + all-TF banner

fix: has_ai_unavailable only when 3D or ALL TFs failed, not any single TF

fix: executor shutdown(wait=False) so Flask returns at 90s deadline

refactor: Ollama 2nd in chain, AI all TFs, max concurrency, JSON mode

fix: skip AI entirely when Ollama is the only provider (instant heuristic)

fix: limit AI TFs to 3D+INTRADAY to prevent Ollama timeout cascade

fix: resolve AI unavailable cascade on watchlist predictions

perf: move Ollama to fast-path, raise concurrency limits for watchlist/top5

feat: add loophole checking to all prediction APIs

feat: add loophole detection and specialist analysis

fix: watchlist slowness, low confidence, low ranges, missing banners

fix: meaningful INTRADAY ranges + 14:15 IST prediction cutoff

fix: normalize news payload shape + clearer watchlist error messages

fix: meaningful INTRADAY ranges, confidence not defaulting to LOW, best_tf not hardcoded 3D

fix: top5 timeout loop — cap live-price fetch at 10s, extend frontend abort to 5min

fix: prevent cache poisoning of watchlist predictions + OHLCV thundering herd