feat: bearish neutralization guard, T7 trigger, WhatsApp alerts, equity curve + portfolio review

ai_forecast: add T7 (slow positive drift) trigger to 1D/3D/5D; loosen T4/guard RSI+BB thresholds
to 50/45% (from 46/40%); add bearish neutralization guard (BEARISH → NEUTRAL when momentum > -3%
and BB < 65% and RSI < 60 and no bull trigger fired); Nifty-EMA200 context in bearish guard.

app.py: WhatsApp HIGH-confidence alerts on watchlist refresh (no-op if env not set);
/api/equity-curve endpoint; /api/portfolio-review batch AI trade coach endpoint.

risk_engine: include_curve=True support for equity curve data.

top5_picker: refactored for reliability.

research: backtest + loop_backtest + validate_on_trades updates; confidence calibration tuned.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ai_forecast.py

@@ -974,6 +974,9 @@ def _build_synthesis_prompt(
             "  [B1] Price below VWAP AND price broke the opening-range (ORB) low AND RSI(5) falling\n"
             "  [B2] Gap down (<-0.3%) failing to reclaim the opening price  [failed gap]\n"
             "BEARISH GUARD: If RSI < 44 AND BB < 35% → call BULLISH not BEARISH (oversold bounce).\n"
+            "BEARISH NEUTRALIZATION GUARD (prevents wrong BEARISH only — does NOT override BULLISH):\n"
+            "  If NO bullish trigger fired AND you are about to call BEARISH AND 10D momentum > -3%\n"
+            "  AND BB < 65% AND RSI < 60 → call NEUTRAL instead (insufficient bearish evidence for intraday)\n"
             "NEUTRAL: no trigger fires AND price is hovering around VWAP with flat RSI(5).\n"
             "Confidence: HIGH = price + VWAP + ORB all aligned. Conflicting VWAP/ORB → MEDIUM. "
             "Mixed/at-VWAP → LOW → output NEUTRAL.\n"
@@ -984,15 +987,20 @@ def _build_synthesis_prompt(
             "  [T1] Price above EMA50 AND MACD > 0\n"
             "  [T2] Price above EMA50 AND 10D momentum > +3% AND BB position < 85%\n"
             "  [T3] Price above EMA50 AND 3+ consecutive up days AND 20D momentum > 0%\n"
-            "  [T4] RSI < 46 AND BB position < 38% AND 10D momentum > -2%  [mild oversold + flat momentum]\n"
+            "  [T4] RSI < 50 AND BB position < 45% AND 10D momentum > -2%  [mild oversold + flat momentum]\n"
             "  [T5] 10D momentum > +7% AND BB position < 80%  [strong momentum breakout]\n"
             "  [T6] RSI < 44 AND BB position < 35%  [deeply oversold — expect intraday bounce]\n"
+            "  [T7] Price above EMA50 AND 20D momentum between +1% and +5% AND RSI < 62  [slow positive drift]\n"
             "BEARISH trigger — ALL conditions required (confirmed bear market, not just correction):\n"
             "  [B2] Below EMA50 AND below EMA200 AND MACD < 0 AND 10D momentum < -5%\n"
             "       AND RSI > 50 AND BB > 40%  [established bear trend, not oversold]\n"
             "OVERBOUGHT stocks (BB > 90%, RSI > 63): call NEUTRAL — NSE stocks in strong uptrends\n"
             "  continue rallying; overbought alone is NOT a reversal signal.\n"
-            "BEARISH GUARD: If RSI < 46 AND BB < 40% → call BULLISH (oversold stocks bounce intraday)\n"
+            "BEARISH GUARD: If RSI < 50 AND BB < 45% AND Nifty ABOVE EMA200 → call BULLISH not BEARISH\n"
+            "  (If Nifty BELOW EMA200, oversold stocks may continue falling — output NEUTRAL instead)\n"
+            "BEARISH NEUTRALIZATION GUARD (prevents wrong BEARISH calls only — does NOT override BULLISH):\n"
+            "  If NO bullish trigger fired AND you are about to call BEARISH AND 10D momentum > -3%\n"
+            "  AND BB < 65% AND RSI < 60 → call NEUTRAL instead (insufficient bearish evidence)\n"
             "POST-SELLOFF GUARD: If Streak shows 3+ consecutive DOWN days → call NEUTRAL, not BEARISH.\n"
             "  After a multi-day selloff, NSE stocks mean-revert strongly. B2 cannot fire in this state.\n"
             "  Exception: B1 (BB>95% AND RSI>64 — stock JUST peaked before falling) can still be BEARISH.\n"
@@ -1007,14 +1015,19 @@ def _build_synthesis_prompt(
             "  [T1] Price above EMA50 AND MACD > 0\n"
             "  [T2] Price above EMA50 AND (10D momentum > +3% OR 20D momentum > +2%)\n"
             "  [T3] Price above EMA50 AND 3+ consecutive up days AND 20D momentum > 0%\n"
-            "  [T4] RSI < 46 AND BB position < 38% AND 10D momentum > -2%  [mild oversold bounce]\n"
+            "  [T4] RSI < 50 AND BB position < 45% AND 10D momentum > -2%  [mild oversold bounce]\n"
             "  [T5] 10D momentum > +6% AND BB position < 75%  [strong breakout]\n"
             "  [T6] RSI < 44 AND BB position < 35%  [deeply oversold — high bounce probability over 3D]\n"
+            "  [T7] Price above EMA50 AND 20D momentum between +1% and +5% AND RSI < 62  [slow positive drift]\n"
             "BEARISH trigger — ALL conditions required:\n"
             "  [B2] Below EMA50 AND below EMA200 AND MACD < 0 AND 20D momentum < -5%\n"
             "       AND RSI > 50 AND BB > 40%  [confirmed bear market downtrend, not correction]\n"
             "OVERBOUGHT stocks (BB > 90%, RSI > 62): call NEUTRAL — momentum stocks continue higher.\n"
-            "BEARISH GUARD: If RSI < 46 AND BB < 40%: call BULLISH not BEARISH (oversold reversal)\n"
+            "BEARISH GUARD: If RSI < 50 AND BB < 45% AND Nifty ABOVE EMA200 → call BULLISH not BEARISH\n"
+            "  (If Nifty BELOW EMA200, oversold stocks may continue falling — output NEUTRAL instead)\n"
+            "BEARISH NEUTRALIZATION GUARD (prevents wrong BEARISH calls only — does NOT override BULLISH):\n"
+            "  If NO bullish trigger fired AND you are about to call BEARISH AND 10D momentum > -3%\n"
+            "  AND BB < 65% AND RSI < 60 → call NEUTRAL instead (insufficient bearish evidence)\n"
             "POST-SELLOFF GUARD: If Streak shows 3+ consecutive DOWN days → call NEUTRAL, not BEARISH.\n"
             "  After a multi-day selloff, NSE stocks mean-revert. 3D BEARISH after 3+ down days fails ~97%.\n"
             "  Exception: B1 (stock JUST reversed from an extreme peak — BB>95%, RSI>64) can be BEARISH.\n"
@@ -1029,14 +1042,19 @@ def _build_synthesis_prompt(
             "  [T1] Price above EMA50 AND 20D momentum > 0%\n"
             "  [T2] Price above EMA200 AND MACD > 0  [medium-term trend intact]\n"
             "  [T3] 10D momentum > +5% AND BB position < 70%  [trend with room to run]\n"
-            "  [T4] RSI < 46 AND BB position < 35% AND 10D momentum > -3%\n"
+            "  [T4] RSI < 50 AND BB position < 45% AND 10D momentum > -3%\n"
             "  [T5] RSI < 44 AND BB position < 30%  [deeply oversold — strong 5D bounce likely]\n"
+            "  [T6] Price above EMA50 AND 20D momentum between +1% and +5% AND RSI < 62  [slow positive drift]\n"
             "BEARISH trigger — ALL conditions required:\n"
             "  [B2] Below EMA50 AND below EMA200 AND 20D momentum < -6%\n"
             "       AND MACD < 0 AND RSI > 52 AND BB > 40%  [genuine bear trend, not oversold dip]\n"
             "OVERBOUGHT stocks (BB > 90%, RSI > 60): call NEUTRAL — high-momentum NSE stocks\n"
             "  overshoot and keep running; overbought is not a timing signal over 5 days.\n"
-            "BEARISH GUARD: If RSI < 46 AND BB < 40%: call BULLISH not BEARISH\n"
+            "BEARISH GUARD: If RSI < 50 AND BB < 45% AND Nifty ABOVE EMA200 → call BULLISH not BEARISH\n"
+            "  (If Nifty BELOW EMA200, oversold stocks may continue falling — output NEUTRAL instead)\n"
+            "BEARISH NEUTRALIZATION GUARD (prevents wrong BEARISH calls only — does NOT override BULLISH):\n"
+            "  If NO bullish trigger fired AND you are about to call BEARISH AND 10D momentum > -3%\n"
+            "  AND BB < 65% AND RSI < 60 → call NEUTRAL instead (insufficient bearish evidence)\n"
             "POST-SELLOFF GUARD: If Streak shows 3+ consecutive DOWN days → call NEUTRAL, not BEARISH.\n"
             "  Over 5D horizon, stocks recovering from multi-day selloffs outperform BEARISH predictions.\n"
             "  Exception: B1 (extreme overbought before the drop — BB>95%, RSI>64) can still be BEARISH.\n"

app.py

@@ -1403,6 +1403,14 @@ def watchlist_picks():
         if not market and pick_market:
             market = pick_market
 
+    # Send WhatsApp alerts for HIGH-confidence predictions (no-op if env vars not set)
+    try:
+        from whatsapp_alerts import send_bulk_alerts
+        all_tf_preds = [tf_pred for pick in picks for tf_pred in pick.get("timeframes", {}).values() if isinstance(tf_pred, dict)]
+        send_bulk_alerts(all_tf_preds)
+    except Exception:
+        pass
+
     mkt = nse_market_status()
     resp: dict = {"picks": picks, "market": market,
                   "generated_at": datetime.now().strftime("%Y-%m-%d %H:%M")}
@@ -1803,6 +1811,77 @@ def portfolio():
     return _json_no_store(summary)
 
 
+@app.route("/api/equity-curve")
+def equity_curve():
+    """Return the equity curve (running portfolio value starting at 100) from closed trades."""
+    try:
+        from risk_engine import get_portfolio_risk
+        data = get_portfolio_risk(include_curve=True)
+        curve = data.get("equity_curve") or []
+        return _json_no_store({"equity_curve": curve, "trade_count": data.get("trade_count", 0)})
+    except Exception as e:
+        return _json_no_store({"equity_curve": [], "error": str(e)}), 500
+
+
+@app.route("/api/portfolio-review")
+def portfolio_review():
+    """Batch AI review of last N closed trades — surfaces systematic patterns and biases."""
+    from datetime import datetime as _dt
+    n = request.args.get("n", 20, type=int)
+    n = max(5, min(n, 100))
+
+    trades = db.get_trade_history()
+    closed = [t for t in trades if t.get("pnl_pct") is not None][-n:]
+    if len(closed) < 3:
+        return _json_no_store({"error": "Need at least 3 closed trades for a meaningful review", "trade_count": len(closed)})
+
+    lines = []
+    for t in closed:
+        outcome = "WIN" if _safe_float(t.get("pnl_pct"), 0) >= 0 else "LOSS"
+        lines.append(
+            f"- {t.get('ticker','?')} | {t.get('direction','?')} | "
+            f"Entry ₹{_safe_float(t.get('entry_price'),0):.0f} → Exit ₹{_safe_float(t.get('exit_price') or t.get('current_price'),0):.0f} | "
+            f"P&L {_safe_float(t.get('pnl_pct'),0):+.2f}% | {outcome}"
+        )
+    trade_list = "\n".join(lines)
+
+    prompt = (
+        f"You are a senior trading coach reviewing {len(closed)} recent paper trades on NSE Indian equities.\n\n"
+        f"TRADES:\n{trade_list}\n\n"
+        "Analyze these trades holistically and provide:\n"
+        "1. SYSTEMATIC BIASES: any patterns in what types of trades consistently win or lose\n"
+        "2. SECTOR / TIMING PATTERNS: any sector or time-based tendencies\n"
+        "3. SIZING / RISK MISTAKES: any position sizing or stop-loss issues visible\n"
+        "4. ONE CONCRETE PROCESS FIX: the single most impactful change to improve outcomes\n\n"
+        "Be specific — name tickers and P&L figures. Keep total response under 300 words."
+    )
+
+    review_text = None
+    try:
+        from ai_forecast import _make_chat_call
+        content, provider, model = _make_chat_call(
+            messages=[{"role": "user", "content": prompt}],
+            max_tokens=512,
+            temperature=0.4,
+        )
+        review_text = content.strip()
+    except Exception as e:
+        review_text = f"AI review unavailable: {e}"
+
+    return _json_no_store({
+        "review_text": review_text,
+        "trade_count": len(closed),
+        "generated_at": _dt.now().strftime("%Y-%m-%d %H:%M"),
+    })
+
+
+def _safe_float(v, default=0.0):
+    try:
+        return float(v) if v is not None else default
+    except Exception:
+        return default
+
+
 # ── SIGNAL ACCURACY ───────────────────────────────────────────────────────────
 
 @app.route("/api/signal-accuracy")

research/ai_prompt_accuracy_trades.csv

@@ -1,73 +1,73 @@
-date,ticker,timeframe,confidence,direction,matched_strategy,ml_prob,vix,nifty_ok,source,source_provider,source_model,entry_price,target_price_lo,target_price_hi,ret_intraday,ret_1d,ret_3d,ret_5d,ret_for_tf,max_up_0d,min_down_0d,max_up_1d,min_down_1d,max_up_3d,min_down_3d,max_up_5d,min_down_5d,max_up_for_tf,min_down_for_tf,intraday_hit_for_tf,target_hit_for_tf
-2026-06-16,HINDALCO.NS,INTRADAY,MEDIUM,BULLISH,,0.78,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,982.4,982.69,984.46,0.0,2.596,2.809,0.448,0.0,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,0.774,-1.262,1,1
-2026-06-16,HINDALCO.NS,1D,MEDIUM,NEUTRAL,,0.78,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,982.4,932.3,1032.5,0.0,2.596,2.809,0.448,2.596,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,2.972,-1.14,1,1
-2026-06-16,HINDALCO.NS,3D,MEDIUM,NEUTRAL,,0.78,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,982.4,932.3,1032.5,0.0,2.596,2.809,0.448,2.809,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,3.563,-1.14,1,1
-2026-06-16,HINDALCO.NS,5D,MEDIUM,BULLISH,,0.78,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,982.4,982.5,983.68,0.0,2.596,2.809,0.448,0.448,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,3.634,-1.14,1,1
-2026-06-16,IPCALAB.NS,INTRADAY,LOW,NEUTRAL,,0.64,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1550.2,1536.25,1564.15,0.0,-0.742,2.296,4.825,0.0,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,2.174,-0.335,1,1
-2026-06-16,IPCALAB.NS,1D,MEDIUM,NEUTRAL,,0.64,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1550.2,1471.14,1629.26,0.0,-0.742,2.296,4.825,-0.742,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,0.548,-1.135,1,1
-2026-06-16,IPCALAB.NS,3D,MEDIUM,NEUTRAL,,0.64,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1550.2,1471.14,1629.26,0.0,-0.742,2.296,4.825,2.296,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,3.8,-1.303,1,1
-2026-06-16,IPCALAB.NS,5D,MEDIUM,NEUTRAL,,0.64,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1550.2,1452.54,1647.86,0.0,-0.742,2.296,4.825,4.825,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,5.883,-1.303,1,1
-2026-06-16,POLYCAB.NS,INTRADAY,MEDIUM,BULLISH,,0.75,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,9546.205,9549.07,9566.25,0.0,3.493,5.623,3.921,0.0,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,0.829,-1.215,1,1
-2026-06-16,POLYCAB.NS,1D,MEDIUM,BULLISH,,0.75,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,9546.205,9547.16,9558.62,0.0,3.493,5.623,3.921,3.493,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,4.129,0.089,1,1
-2026-06-16,POLYCAB.NS,3D,MEDIUM,BULLISH,,0.75,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,9546.205,9547.16,9558.62,0.0,3.493,5.623,3.921,5.623,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,6.011,0.089,1,1
-2026-06-16,POLYCAB.NS,5D,MEDIUM,BULLISH,,0.75,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,9546.205,9547.16,9558.62,0.0,3.493,5.623,3.921,3.921,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,6.074,0.089,1,1
-2026-06-16,DLF.NS,INTRADAY,MEDIUM,BULLISH,,0.65,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,629.3,629.49,630.62,0.0,-0.914,-0.763,-2.693,0.0,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,0.429,-2.797,1,1
-2026-06-16,DLF.NS,1D,MEDIUM,BULLISH,,0.65,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,629.3,629.36,630.12,0.0,-0.914,-0.763,-2.693,-0.914,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,0.707,-1.764,1,1
-2026-06-16,DLF.NS,3D,MEDIUM,BULLISH,,0.65,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,629.3,629.36,630.12,0.0,-0.914,-0.763,-2.693,-0.763,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,2.336,-1.764,1,1
-2026-06-16,DLF.NS,5D,MEDIUM,BULLISH,,0.65,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,629.3,629.36,630.12,0.0,-0.914,-0.763,-2.693,-2.693,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,2.336,-3.059,1,1
-2026-06-16,SHRIRAMFIN.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1000.093,1000.39,1002.19,0.0,0.174,-0.383,-1.233,0.0,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,0.428,-1.362,1,1
-2026-06-16,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,0.174,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,0.994,-0.378,1,1
-2026-06-16,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,-0.383,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,1.148,-1.337,1,1
-2026-06-16,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,-1.233,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,1.148,-2.182,1,1
-2026-06-16,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1825.3,1825.85,1829.13,0.0,4.996,6.459,0.904,0.0,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,5.057,-0.493,1,1
-2026-06-16,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,4.996,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,4.996,0.038,1,1
-2026-06-16,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,6.459,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,9.023,0.038,1,1
-2026-06-16,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.71,13.4,False,github:gpt-4o-mini,github,gpt-4o-mini,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,0.904,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,9.023,0.038,1,1
-2026-06-19,HINDALCO.NS,INTRADAY,MEDIUM,BULLISH,,0.74,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1010.0,1010.3,1012.12,0.0,0.416,-3.307,-5.624,0.0,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.673,-2.455,1,1
-2026-06-19,HINDALCO.NS,1D,MEDIUM,NEUTRAL,,0.74,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1010.0,958.49,1061.51,0.0,0.416,-3.307,-5.624,0.416,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-0.634,1,1
-2026-06-19,HINDALCO.NS,3D,MEDIUM,NEUTRAL,,0.74,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1010.0,958.49,1061.51,0.0,0.416,-3.307,-5.624,-3.307,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-3.931,1,1
-2026-06-19,HINDALCO.NS,5D,MEDIUM,NEUTRAL,,0.74,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1010.0,946.37,1073.63,0.0,0.416,-3.307,-5.624,-5.624,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-5.921,1,1
-2026-06-19,IPCALAB.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1585.8,1586.28,1589.13,0.0,0.725,2.232,2.049,0.0,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,1.469,-2.73,1,1
-2026-06-19,IPCALAB.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1585.8,1585.96,1587.86,0.0,0.725,2.232,2.049,0.725,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,1.204,-0.214,1,1
-2026-06-19,IPCALAB.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1585.8,1585.96,1587.86,0.0,0.725,2.232,2.049,2.232,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,3.506,-0.214,1,1
-2026-06-19,IPCALAB.NS,5D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1585.8,1585.96,1587.86,0.0,0.725,2.232,2.049,2.049,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,4.364,-0.214,1,1
-2026-06-19,POLYCAB.NS,INTRADAY,MEDIUM,BULLISH,,0.72,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,10083.0,10086.02,10104.17,0.0,-0.6,-3.873,-5.475,0.0,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.367,-2.499,1,1
-2026-06-19,POLYCAB.NS,1D,MEDIUM,BULLISH,,0.72,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,10083.0,10084.01,10096.11,0.0,-0.6,-3.873,-5.475,-0.6,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-0.903,1,1
-2026-06-19,POLYCAB.NS,3D,MEDIUM,BULLISH,,0.72,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,10083.0,10084.01,10096.11,0.0,-0.6,-3.873,-5.475,-3.873,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-4.046,1,1
-2026-06-19,POLYCAB.NS,5D,MEDIUM,BULLISH,,0.72,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,10083.0,10084.01,10096.11,0.0,-0.6,-3.873,-5.475,-5.475,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-5.618,1,1
-2026-06-19,DLF.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,624.5,624.69,625.81,0.0,0.496,-1.017,-0.504,0.0,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,2.322,-0.777,1,1
-2026-06-19,DLF.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,0.496,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.169,0.096,1,1
-2026-06-19,DLF.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,-1.017,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.906,-2.418,1,1
-2026-06-19,DLF.NS,5D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,-0.504,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.906,-2.418,1,1
-2026-06-19,SHRIRAMFIN.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,996.265,996.56,998.36,0.0,-0.903,1.707,2.984,0.0,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,0.579,-0.958,1,1
-2026-06-19,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,-0.903,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,0.384,-1.807,1,1
-2026-06-19,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,1.707,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,2.74,-2.131,1,1
-2026-06-19,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.71,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,2.984,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,5.0,-2.131,1,1
-2026-06-19,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.78,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1943.2,1943.78,1947.28,0.0,-3.355,-9.536,-12.979,0.0,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,1.379,-1.662,1,1
-2026-06-19,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.78,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-3.355,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-3.767,1,1
-2026-06-19,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.78,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-9.536,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-9.953,1,1
-2026-06-19,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.78,13.0,False,github:gpt-4o-mini,github,gpt-4o-mini,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-12.979,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-13.601,1,1
-2026-06-22,HINDALCO.NS,INTRADAY,LOW,NEUTRAL,,0.64,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1014.2,1005.07,1023.33,0.0,-2.702,-6.015,-4.969,0.0,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,0.385,-1.045,1,1
-2026-06-22,HINDALCO.NS,1D,MEDIUM,NEUTRAL,,0.64,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1014.2,962.48,1065.92,0.0,-2.702,-6.015,-4.969,-2.702,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-3.569,1,0
-2026-06-22,HINDALCO.NS,3D,MEDIUM,BULLISH,,0.64,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1014.2,1014.3,1015.52,0.0,-2.702,-6.015,-4.969,-6.015,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-6.31,0,0
-2026-06-22,HINDALCO.NS,5D,MEDIUM,NEUTRAL,,0.64,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1014.2,950.31,1078.09,0.0,-2.702,-6.015,-4.969,-4.969,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-6.31,1,0
-2026-06-22,IPCALAB.NS,INTRADAY,LOW,NEUTRAL,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1597.3,1582.92,1611.68,0.0,1.734,1.315,3.124,0.0,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,0.476,-0.933,1,1
-2026-06-22,IPCALAB.NS,1D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1597.3,1597.46,1599.38,0.0,1.734,1.315,3.124,1.734,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,2.761,0.163,1,1
-2026-06-22,IPCALAB.NS,3D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1597.3,1597.46,1599.38,0.0,1.734,1.315,3.124,1.315,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,3.612,0.163,1,1
-2026-06-22,IPCALAB.NS,5D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1597.3,1597.46,1599.38,0.0,1.734,1.315,3.124,3.124,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,4.232,0.163,1,1
-2026-06-22,POLYCAB.NS,INTRADAY,LOW,NEUTRAL,,0.68,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,10022.5,9932.3,10112.7,0.0,-1.018,-4.904,-2.43,0.0,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,1.033,-0.304,1,1
-2026-06-22,POLYCAB.NS,1D,MEDIUM,BULLISH,,0.68,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,10022.5,10023.5,10035.53,0.0,-1.018,-4.904,-2.43,-1.018,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-1.307,1,1
-2026-06-22,POLYCAB.NS,3D,MEDIUM,BULLISH,,0.68,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,10022.5,10023.5,10035.53,0.0,-1.018,-4.904,-2.43,-4.904,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-5.049,1,1
-2026-06-22,POLYCAB.NS,5D,MEDIUM,BULLISH,,0.68,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,10022.5,10023.5,10035.53,0.0,-1.018,-4.904,-2.43,-2.43,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-6.006,1,1
-2026-06-22,DLF.NS,INTRADAY,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,627.6,627.79,628.92,0.0,-2.43,-0.996,-2.047,0.0,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,0.669,-0.398,1,1
-2026-06-22,DLF.NS,1D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-2.43,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.796,1,1
-2026-06-22,DLF.NS,3D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-0.996,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.9,1,1
-2026-06-22,DLF.NS,5D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-2.047,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.9,1,1
-2026-06-22,SHRIRAMFIN.NS,INTRADAY,MEDIUM,BULLISH,,0.78,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,987.265,987.56,989.34,0.0,0.05,3.923,4.009,0.0,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,1.299,-0.912,1,1
-2026-06-22,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.78,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,987.265,987.36,988.55,0.0,0.05,3.923,4.009,0.05,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,1.496,-0.841,1,1
-2026-06-22,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.78,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,987.265,987.36,988.55,0.0,0.05,3.923,4.009,3.923,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,5.958,-1.239,1,1
-2026-06-22,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.78,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,987.265,987.36,988.55,0.0,0.05,3.923,4.009,4.009,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,6.3,-1.239,1,1
-2026-06-22,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1878.0,1878.56,1881.94,0.0,-1.928,-9.957,-9.526,0.0,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,3.573,-0.426,1,1
-2026-06-22,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-1.928,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-2.391,1,1
-2026-06-22,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-9.957,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-10.602,1,1
-2026-06-22,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.71,12.8,False,github:gpt-4o-mini,github,gpt-4o-mini,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-9.526,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-12.875,1,1
+date,ticker,timeframe,confidence,direction,matched_strategy,ml_prob,vix,nifty_ok,source,source_provider,source_model,entry_price,target_price_lo,target_price_hi,ret_intraday,ret_1d,ret_3d,ret_5d,ret_for_tf,max_up_0d,min_down_0d,max_up_1d,min_down_1d,max_up_3d,min_down_3d,max_up_5d,min_down_5d,max_up_for_tf,min_down_for_tf,intraday_hit_for_tf,target_hit_for_tf,trigger_T1,trigger_T2,trigger_T3,trigger_T4,trigger_T5,trigger_T6,trigger_B1,trigger_B2
+2026-06-16,HINDALCO.NS,INTRADAY,MEDIUM,BULLISH,,0.78,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,982.4,982.69,984.46,0.0,2.596,2.809,0.448,0.0,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,0.774,-1.262,1,1,0,0,0,0,0,1,0,0
+2026-06-16,HINDALCO.NS,1D,MEDIUM,BULLISH,,0.78,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,982.4,982.5,983.68,0.0,2.596,2.809,0.448,2.596,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,2.972,-1.14,1,1,0,0,0,0,0,1,0,0
+2026-06-16,HINDALCO.NS,3D,MEDIUM,BULLISH,,0.78,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,982.4,982.5,983.68,0.0,2.596,2.809,0.448,2.809,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,3.563,-1.14,1,1,0,0,0,0,0,1,0,0
+2026-06-16,HINDALCO.NS,5D,MEDIUM,BULLISH,,0.78,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,982.4,982.5,983.68,0.0,2.596,2.809,0.448,0.448,0.774,-1.262,2.972,-1.14,3.563,-1.14,3.634,-1.14,3.634,-1.14,1,1,0,0,0,0,0,1,0,0
+2026-06-16,IPCALAB.NS,INTRADAY,LOW,NEUTRAL,,0.64,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1550.2,1536.25,1564.15,0.0,-0.742,2.296,4.825,0.0,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,2.174,-0.335,1,1,0,0,0,1,0,0,0,0
+2026-06-16,IPCALAB.NS,1D,MEDIUM,NEUTRAL,,0.64,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1550.2,1471.14,1629.26,0.0,-0.742,2.296,4.825,-0.742,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,0.548,-1.135,1,1,0,0,0,1,0,0,0,0
+2026-06-16,IPCALAB.NS,3D,MEDIUM,NEUTRAL,,0.64,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1550.2,1471.14,1629.26,0.0,-0.742,2.296,4.825,2.296,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,3.8,-1.303,1,1,0,0,0,1,0,0,0,0
+2026-06-16,IPCALAB.NS,5D,MEDIUM,NEUTRAL,,0.64,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1550.2,1452.54,1647.86,0.0,-0.742,2.296,4.825,4.825,2.174,-0.335,0.548,-1.135,3.8,-1.303,5.883,-1.303,5.883,-1.303,1,1,0,0,0,1,0,0,0,0
+2026-06-16,POLYCAB.NS,INTRADAY,LOW,NEUTRAL,,0.75,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,9546.205,9460.29,9632.12,0.0,3.493,5.623,3.921,0.0,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,0.829,-1.215,1,1,0,0,1,0,0,0,0,0
+2026-06-16,POLYCAB.NS,1D,MEDIUM,NEUTRAL,,0.75,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,9546.205,9059.35,10033.06,0.0,3.493,5.623,3.921,3.493,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,4.129,0.089,1,0,0,0,1,0,0,0,0,0
+2026-06-16,POLYCAB.NS,3D,LOW,NEUTRAL,,0.75,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,9546.205,9059.35,10033.06,0.0,3.493,5.623,3.921,5.623,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,6.011,0.089,0,0,0,0,1,0,0,0,0,0
+2026-06-16,POLYCAB.NS,5D,LOW,NEUTRAL,,0.75,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,9546.205,8944.79,10147.62,0.0,3.493,5.623,3.921,3.921,0.829,-1.215,4.129,0.089,6.011,0.089,6.074,0.089,6.074,0.089,1,0,0,0,1,0,0,0,0,0
+2026-06-16,DLF.NS,INTRADAY,LOW,NEUTRAL,,0.65,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,629.3,623.64,634.96,0.0,-0.914,-0.763,-2.693,0.0,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,0.429,-2.797,1,1,1,0,1,0,0,0,0,0
+2026-06-16,DLF.NS,1D,LOW,NEUTRAL,,0.65,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,629.3,597.21,661.39,0.0,-0.914,-0.763,-2.693,-0.914,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,0.707,-1.764,1,1,1,0,1,0,0,0,0,0
+2026-06-16,DLF.NS,3D,MEDIUM,NEUTRAL,,0.65,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,629.3,597.21,661.39,0.0,-0.914,-0.763,-2.693,-0.763,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,2.336,-1.764,1,1,1,0,1,0,0,0,0,0
+2026-06-16,DLF.NS,5D,LOW,NEUTRAL,,0.65,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,629.3,589.65,668.95,0.0,-0.914,-0.763,-2.693,-2.693,0.429,-2.797,0.707,-1.764,2.336,-1.764,2.336,-3.059,2.336,-3.059,1,1,1,0,1,0,0,0,0,0
+2026-06-16,SHRIRAMFIN.NS,INTRADAY,LOW,NEUTRAL,,0.71,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1000.093,991.09,1009.09,0.0,0.174,-0.383,-1.233,0.0,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,0.428,-1.362,1,1,1,0,1,0,0,0,0,0
+2026-06-16,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.71,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,0.174,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,0.994,-0.378,1,1,1,0,1,0,0,0,0,0
+2026-06-16,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.71,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,-0.383,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,1.148,-1.337,1,1,1,0,1,0,0,0,0,0
+2026-06-16,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.71,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1000.093,1000.19,1001.39,0.0,0.174,-0.383,-1.233,-1.233,0.428,-1.362,0.994,-0.378,1.148,-1.337,1.148,-2.182,1.148,-2.182,1,1,1,0,1,0,0,0,0,0
+2026-06-16,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.4,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1825.3,1825.85,1829.13,0.0,4.996,6.459,0.904,0.0,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,5.057,-0.493,1,1,0,0,0,1,1,0,0,0
+2026-06-16,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.71,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,4.996,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,4.996,0.038,1,1,0,0,0,1,1,0,0,0
+2026-06-16,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.71,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,6.459,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,9.023,0.038,1,1,0,0,0,1,1,0,0,0
+2026-06-16,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.71,13.4,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1825.3,1825.48,1827.67,0.0,4.996,6.459,0.904,0.904,5.057,-0.493,4.996,0.038,9.023,0.038,9.023,0.038,9.023,0.038,1,1,0,0,0,1,1,0,0,0
+2026-06-19,HINDALCO.NS,INTRADAY,LOW,NEUTRAL,,0.74,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1010.0,1000.91,1019.09,0.0,0.416,-3.307,-5.624,0.0,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.673,-2.455,1,1,0,0,0,0,0,1,0,0
+2026-06-19,HINDALCO.NS,1D,MEDIUM,NEUTRAL,,0.74,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1010.0,958.49,1061.51,0.0,0.416,-3.307,-5.624,0.416,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-0.634,1,1,0,0,0,0,0,1,0,0
+2026-06-19,HINDALCO.NS,3D,MEDIUM,NEUTRAL,,0.74,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1010.0,958.49,1061.51,0.0,0.416,-3.307,-5.624,-3.307,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-3.931,1,1,0,0,0,0,0,1,0,0
+2026-06-19,HINDALCO.NS,5D,MEDIUM,NEUTRAL,,0.74,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1010.0,946.37,1073.63,0.0,0.416,-3.307,-5.624,-5.624,0.673,-2.455,0.802,-0.634,0.802,-3.931,0.802,-5.921,0.802,-5.921,1,1,0,0,0,0,0,1,0,0
+2026-06-19,IPCALAB.NS,INTRADAY,LOW,NEUTRAL,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1585.8,1571.53,1600.07,0.0,0.725,2.232,2.049,0.0,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,1.469,-2.73,1,1,0,0,0,0,0,0,0,0
+2026-06-19,IPCALAB.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1585.8,1585.96,1587.86,0.0,0.725,2.232,2.049,0.725,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,1.204,-0.214,1,1,0,0,0,0,0,0,0,0
+2026-06-19,IPCALAB.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1585.8,1585.96,1587.86,0.0,0.725,2.232,2.049,2.232,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,3.506,-0.214,1,1,0,0,0,0,0,0,0,0
+2026-06-19,IPCALAB.NS,5D,MEDIUM,NEUTRAL,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,1585.8,1485.89,1685.71,0.0,0.725,2.232,2.049,2.049,1.469,-2.73,1.204,-0.214,3.506,-0.214,4.364,-0.214,4.364,-0.214,1,1,0,0,0,0,0,0,0,0
+2026-06-19,POLYCAB.NS,INTRADAY,MEDIUM,BULLISH,,0.72,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10083.0,10086.02,10104.17,0.0,-0.6,-3.873,-5.475,0.0,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.367,-2.499,1,1,1,0,1,0,0,0,0,0
+2026-06-19,POLYCAB.NS,1D,LOW,NEUTRAL,,0.72,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10083.0,9568.77,10597.23,0.0,-0.6,-3.873,-5.475,-0.6,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-0.903,1,1,1,0,1,0,0,0,0,0
+2026-06-19,POLYCAB.NS,3D,MEDIUM,BULLISH,,0.72,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10083.0,10084.01,10096.11,0.0,-0.6,-3.873,-5.475,-3.873,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-4.046,1,1,1,0,1,0,0,0,0,0
+2026-06-19,POLYCAB.NS,5D,LOW,NEUTRAL,,0.72,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10083.0,9447.77,10718.23,0.0,-0.6,-3.873,-5.475,-5.475,0.367,-2.499,0.426,-0.903,0.426,-4.046,0.426,-5.618,0.426,-5.618,1,1,1,0,1,0,0,0,0,0
+2026-06-19,DLF.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,624.5,624.69,625.81,0.0,0.496,-1.017,-0.504,0.0,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,2.322,-0.777,1,1,1,1,0,0,0,0,0,0
+2026-06-19,DLF.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,0.496,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.169,0.096,1,1,1,1,0,0,0,0,0,0
+2026-06-19,DLF.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,-1.017,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.906,-2.418,1,1,1,1,0,0,0,0,0,0
+2026-06-19,DLF.NS,5D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,624.5,624.56,625.31,0.0,0.496,-1.017,-0.504,-0.504,2.322,-0.777,1.169,0.096,1.906,-2.418,1.906,-2.418,1.906,-2.418,1,1,1,1,0,0,0,0,0,0
+2026-06-19,SHRIRAMFIN.NS,INTRADAY,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,996.265,996.56,998.36,0.0,-0.903,1.707,2.984,0.0,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,0.579,-0.958,1,1,1,1,0,0,0,0,0,0
+2026-06-19,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,-0.903,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,0.384,-1.807,1,1,1,1,0,0,0,0,0,0
+2026-06-19,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.71,13.0,False,huggingface:meta-llama/llama-3.1-8b-instruct,huggingface,meta-llama/llama-3.1-8b-instruct,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,1.707,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,2.74,-2.131,1,1,1,1,0,0,0,0,0,0
+2026-06-19,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.71,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,996.265,996.36,997.56,0.0,-0.903,1.707,2.984,2.984,0.579,-0.958,0.384,-1.807,2.74,-2.131,5.0,-2.131,5.0,-2.131,1,1,1,1,0,0,0,0,0,0
+2026-06-19,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.78,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1943.2,1943.78,1947.28,0.0,-3.355,-9.536,-12.979,0.0,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,1.379,-1.662,1,1,1,1,0,0,0,0,0,0
+2026-06-19,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.78,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-3.355,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-3.767,1,1,1,1,0,0,0,0,0,0
+2026-06-19,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.78,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-9.536,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-9.953,1,1,1,1,0,0,0,0,0,0
+2026-06-19,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.78,13.0,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1943.2,1943.39,1945.73,0.0,-3.355,-9.536,-12.979,-12.979,1.379,-1.662,0.098,-3.767,0.098,-9.953,0.098,-13.601,0.098,-13.601,1,1,1,1,0,0,0,0,0,0
+2026-06-22,HINDALCO.NS,INTRADAY,MEDIUM,BEARISH,,0.64,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1014.2,1012.98,1013.79,0.0,-2.702,-6.015,-4.969,0.0,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,0.385,-1.045,1,1,0,0,0,0,0,1,0,0
+2026-06-22,HINDALCO.NS,1D,LOW,NEUTRAL,,0.64,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1014.2,962.48,1065.92,0.0,-2.702,-6.015,-4.969,-2.702,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-3.569,1,0,0,0,0,0,0,1,0,0
+2026-06-22,HINDALCO.NS,3D,MEDIUM,NEUTRAL,,0.64,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1014.2,962.48,1065.92,0.0,-2.702,-6.015,-4.969,-6.015,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-6.31,0,0,0,0,0,0,0,1,0,0
+2026-06-22,HINDALCO.NS,5D,MEDIUM,NEUTRAL,,0.64,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1014.2,950.31,1078.09,0.0,-2.702,-6.015,-4.969,-4.969,0.385,-1.045,-1.499,-3.569,-1.499,-6.31,-1.499,-6.31,-1.499,-6.31,1,0,0,0,0,0,0,1,0,0
+2026-06-22,IPCALAB.NS,INTRADAY,LOW,NEUTRAL,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1597.3,1582.92,1611.68,0.0,1.734,1.315,3.124,0.0,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,0.476,-0.933,1,1,0,0,1,0,0,0,0,0
+2026-06-22,IPCALAB.NS,1D,LOW,NEUTRAL,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1597.3,1515.84,1678.76,0.0,1.734,1.315,3.124,1.734,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,2.761,0.163,1,0,0,0,1,0,0,0,0,0
+2026-06-22,IPCALAB.NS,3D,LOW,NEUTRAL,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1597.3,1515.84,1678.76,0.0,1.734,1.315,3.124,1.315,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,3.612,0.163,1,0,0,0,1,0,0,0,0,0
+2026-06-22,IPCALAB.NS,5D,LOW,NEUTRAL,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1597.3,1496.67,1697.93,0.0,1.734,1.315,3.124,3.124,0.476,-0.933,2.761,0.163,3.612,0.163,4.232,0.163,4.232,0.163,1,0,0,0,1,0,0,0,0,0
+2026-06-22,POLYCAB.NS,INTRADAY,MEDIUM,BEARISH,,0.68,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10022.5,10010.47,10018.49,0.0,-1.018,-4.904,-2.43,0.0,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,1.033,-0.304,1,1,1,0,0,0,0,0,0,0
+2026-06-22,POLYCAB.NS,1D,LOW,NEUTRAL,,0.68,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10022.5,9511.35,10533.65,0.0,-1.018,-4.904,-2.43,-1.018,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-1.307,1,1,1,0,0,0,0,0,0,0
+2026-06-22,POLYCAB.NS,3D,MEDIUM,NEUTRAL,,0.68,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10022.5,9511.35,10533.65,0.0,-1.018,-4.904,-2.43,-4.904,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-5.049,1,1,1,0,0,0,0,0,0,0
+2026-06-22,POLYCAB.NS,5D,LOW,NEUTRAL,,0.68,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,10022.5,9391.08,10653.92,0.0,-1.018,-4.904,-2.43,-2.43,1.033,-0.304,0.524,-1.307,0.524,-5.049,0.524,-6.006,0.524,-6.006,1,1,1,0,0,0,0,0,0,0
+2026-06-22,DLF.NS,INTRADAY,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,627.6,627.79,628.92,0.0,-2.43,-0.996,-2.047,0.0,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,0.669,-0.398,1,1,1,1,0,0,0,0,0,0
+2026-06-22,DLF.NS,1D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-2.43,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.796,1,1,1,1,0,0,0,0,0,0
+2026-06-22,DLF.NS,3D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-0.996,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.9,1,1,1,1,0,0,0,0,0,0
+2026-06-22,DLF.NS,5D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,627.6,627.66,628.42,0.0,-2.43,-0.996,-2.047,-2.047,0.669,-0.398,1.402,-2.796,1.402,-2.9,1.402,-2.9,1.402,-2.9,1,1,1,1,0,0,0,0,0,0
+2026-06-22,SHRIRAMFIN.NS,INTRADAY,MEDIUM,BULLISH,,0.78,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,987.265,987.56,989.34,0.0,0.05,3.923,4.009,0.0,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,1.299,-0.912,1,1,1,1,0,0,1,0,0,0
+2026-06-22,SHRIRAMFIN.NS,1D,MEDIUM,BULLISH,,0.78,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,987.265,987.36,988.55,0.0,0.05,3.923,4.009,0.05,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,1.496,-0.841,1,1,1,1,0,0,1,0,0,0
+2026-06-22,SHRIRAMFIN.NS,3D,MEDIUM,BULLISH,,0.78,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,987.265,987.36,988.55,0.0,0.05,3.923,4.009,3.923,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,5.958,-1.239,1,1,1,1,0,0,1,0,0,0
+2026-06-22,SHRIRAMFIN.NS,5D,MEDIUM,BULLISH,,0.78,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,987.265,987.36,988.55,0.0,0.05,3.923,4.009,4.009,1.299,-0.912,1.496,-0.841,5.958,-1.239,6.3,-1.239,6.3,-1.239,1,1,1,1,0,0,1,0,0,0
+2026-06-22,AXISCADES.NS,INTRADAY,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.3-70b-versatile,groq,llama-3.3-70b-versatile,1878.0,1878.56,1881.94,0.0,-1.928,-9.957,-9.526,0.0,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,3.573,-0.426,1,1,1,0,0,0,0,0,0,0
+2026-06-22,AXISCADES.NS,1D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-1.928,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-2.391,1,1,1,0,0,0,0,0,0,0
+2026-06-22,AXISCADES.NS,3D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-9.957,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-10.602,1,1,1,0,0,0,0,0,0,0
+2026-06-22,AXISCADES.NS,5D,MEDIUM,BULLISH,,0.71,12.8,False,groq:llama-3.1-8b-instant,groq,llama-3.1-8b-instant,1878.0,1878.19,1880.44,0.0,-1.928,-9.957,-9.526,-9.526,3.573,-0.426,1.587,-2.391,1.587,-10.602,1.587,-12.875,1.587,-12.875,1,1,1,0,0,0,0,0,0,0

research/backtest.py

@@ -49,6 +49,13 @@ VIX_TK     = "^INDIAVIX"
 STEP       = 60       # every 60 trading days → ~22 dates × 15 tickers × 3 TFs = ~990 work items
 WORKERS    = 1        # single worker — avoids 429 burst; rate limiter still controls pace
 
+# ── HOLD-OUT SPLIT ────────────────────────────────────────────────────────────
+# Training window: START → TRAIN_END  (loop_backtest.py optimizes here)
+# Hold-out window: HOLDOUT_START → HOLDOUT_END  (--eval flag; never used for optimization)
+TRAIN_END     = "2024-12-31"
+HOLDOUT_START = "2025-01-01"
+HOLDOUT_END   = "2025-06-01"
+
 TIMEFRAMES = ["INTRADAY", "1D", "3D", "5D"]
 _TF_COL    = {"INTRADAY": "ret_intraday", "1D": "ret_1d", "3D": "ret_3d", "5D": "ret_5d"}
 
@@ -312,6 +319,63 @@ def _compute_indicators(sc_tk, sh_tk, sl_tk, sv_tk, date):
     return inds
 
 
+# ── TRIGGER FLAG EVALUATION ───────────────────────────────────────────────────
+
+def _compute_trigger_flags(inds: dict, price: float) -> dict:
+    """Evaluate which synthesis-prompt triggers fired for a given indicator snapshot.
+
+    Uses the 1D trigger conditions as the canonical set (they represent all TFs).
+    Adds 9 boolean columns: trigger_T1 … trigger_T7, trigger_B1, trigger_B2.
+    """
+    rsi    = inds.get("rsi14", 50.0)
+    bb     = inds.get("bb_pct", 50.0)
+    r10    = inds.get("return_10d", 0.0)
+    r20    = inds.get("return_20d", 0.0)
+    macd   = inds.get("macd_signal", 0.0)
+    ema50  = inds.get("ema50")
+    ema200 = inds.get("ema200")
+
+    above_ema50  = (ema50  is not None) and (price > ema50)
+    above_ema200 = (ema200 is not None) and (price > ema200)
+
+    streak = inds.get("consec_days", "")
+    consec_up = 0
+    if isinstance(streak, str) and streak.startswith("+"):
+        try: consec_up = int(streak.split()[0].lstrip("+"))
+        except Exception: pass
+
+    T1 = bool(above_ema50 and macd > 0)
+    T2 = bool(above_ema50 and r10 > 3.0 and bb < 85.0)
+    T3 = bool(above_ema50 and consec_up >= 3 and r20 > 0.0)
+    T4 = bool(rsi < 50 and bb < 45.0 and r10 > -2.0)   # widened from rsi<46 bb<38
+    T5 = bool(r10 > 7.0 and bb < 80.0)
+    T6 = bool(rsi < 44 and bb < 35.0)
+    T7 = bool(above_ema50 and 1.0 <= r20 <= 5.0 and rsi < 62.0)  # slow positive drift
+
+    # B1 removed from production (overbought reversal); kept as placeholder False for CSV schema stability
+    B1 = False
+    B2 = bool(
+        (not above_ema50)
+        and (not above_ema200)
+        and macd < 0
+        and r10 < -5.0
+        and rsi > 50
+        and bb > 40.0
+    )
+
+    return {
+        "trigger_T1": int(T1),
+        "trigger_T2": int(T2),
+        "trigger_T3": int(T3),
+        "trigger_T4": int(T4),
+        "trigger_T5": int(T5),
+        "trigger_T6": int(T6),
+        "trigger_T7": int(T7),
+        "trigger_B1": int(B1),
+        "trigger_B2": int(B2),
+    }
+
+
 # ── MAIN BACKTEST ──────────────────────────────────────────────────────────────
 
 def _build_indicator_snapshots(sc, sh, sl, sv, nc, vc):
@@ -618,6 +682,7 @@ def run_backtest(work_items: list[dict], csv_path: str | None = None, limit_work
                 w["tf"],
                 ret_tf,
             )
+            trigger_flags = _compute_trigger_flags(w.get("inds", {}), w["price"])
             return {
                 "date":             str(w["date"].date()),
                 "ticker":           w["ticker"],
@@ -651,6 +716,7 @@ def run_backtest(work_items: list[dict], csv_path: str | None = None, limit_work
                 "min_down_for_tf":  round(min_down_tf, 3),
                 "intraday_hit_for_tf": int(direction_hit),
                 "target_hit_for_tf": int(target_hit),
+                **trigger_flags,
             }
         except Exception as e:
             print(f"  SKIP {w['ticker']} @ {w['date']} [{w['tf']}]: {e}")
@@ -770,6 +836,54 @@ def print_results(df: pd.DataFrame) -> dict:
             for k, v in provider_counts.items():
                 print(f"    - {k}: {v}")
 
+    # TABLE 4 — Per-trigger accuracy breakdown (only if columns present in CSV)
+    _trigger_cols = [c for c in ["trigger_T1","trigger_T2","trigger_T3","trigger_T4","trigger_T5","trigger_T6","trigger_T7","trigger_B2"] if c in df.columns]
+    if _trigger_cols:
+        _sep()
+        print("TABLE 4 — Per-Trigger Accuracy Breakdown (how often predictions that fired each trigger were correct)")
+        _sep()
+        print(f"  {'Trigger':<12} {'Fired':>7} {'Correct':>9} {'HitRate':>9}  Direction")
+        _sep("-")
+        _dir_df = df[df["direction"].isin(["BULLISH", "BEARISH"])].copy()
+        _bull_triggers = ["trigger_T1","trigger_T2","trigger_T3","trigger_T4","trigger_T5","trigger_T6","trigger_T7"]
+        _bear_triggers = ["trigger_B2"]
+        for col in _trigger_cols:
+            fired = _dir_df[_dir_df[col] == 1]
+            if len(fired) < 3:
+                continue
+            n_fired = len(fired)
+            n_correct = int(fired["intraday_hit_for_tf"].sum())
+            rate = n_correct / n_fired * 100
+            dirn = "BULLISH" if col in _bull_triggers else "BEARISH"
+            print(f"  {col:<12} {n_fired:>7} {n_correct:>9} {rate:>8.1f}%  {dirn}")
+
+    # TABLE 5 — Per-regime accuracy slice (requires nifty_ok and vix columns)
+    if "nifty_ok" in df.columns and "vix" in df.columns:
+        _sep()
+        print("TABLE 5 — Per-Regime Accuracy Slice (directional predictions only)")
+        _sep()
+        print(f"  {'Regime':<22} {'N':>6} {'DirHit':>9} {'TgtHit':>9}")
+        _sep("-")
+        _dir_df2 = df[df["direction"].isin(["BULLISH","BEARISH"])].copy()
+        _dir_df2["vix_band"] = pd.cut(_dir_df2["vix"], bins=[0, 15, 20, 100], labels=["VIX<15","VIX 15-20","VIX>20"])
+        for regime_label, mask in [
+            ("Nifty Bull (above EMA200)", _dir_df2["nifty_ok"] == True),
+            ("Nifty Bear (below EMA200)", _dir_df2["nifty_ok"] == False),
+        ]:
+            sub = _dir_df2[mask]
+            if len(sub) < 3:
+                continue
+            dh = sub["intraday_hit_for_tf"].mean() * 100
+            th = sub["target_hit_for_tf"].mean() * 100
+            print(f"  {regime_label:<22} {len(sub):>6} {dh:>8.1f}% {th:>8.1f}%")
+        for band in ["VIX<15","VIX 15-20","VIX>20"]:
+            sub = _dir_df2[_dir_df2["vix_band"] == band]
+            if len(sub) < 3:
+                continue
+            dh = sub["intraday_hit_for_tf"].mean() * 100
+            th = sub["target_hit_for_tf"].mean() * 100
+            print(f"  {band:<22} {len(sub):>6} {dh:>8.1f}% {th:>8.1f}%")
+
     _write_calibration_artifact(df)
     _sep()
     return acc
@@ -829,10 +943,26 @@ if __name__ == "__main__":
     p.add_argument("--timeframes", nargs="+", choices=TIMEFRAMES, help="Run only selected timeframe(s)")
     p.add_argument("--csv-out", default=os.path.join(os.path.dirname(__file__), "ai_prompt_accuracy.csv"), help="Output CSV path")
     p.add_argument("--limit-work-items", type=int, default=0, help="Run only first N work items (quick smoke runs)")
+    p.add_argument("--eval", action="store_true",
+                   help=f"Run on hold-out dates only ({HOLDOUT_START} → {HOLDOUT_END}) — never used for prompt optimization")
+    p.add_argument("--start", default=None, help="Override start date (YYYY-MM-DD)")
+    p.add_argument("--end",   default=None, help="Override end date (YYYY-MM-DD)")
     args = p.parse_args()
 
     csv_path = args.csv_out
 
+    # Resolve date range
+    if args.eval:
+        _run_start = HOLDOUT_START
+        _run_end   = HOLDOUT_END
+        _eval_csv  = csv_path.replace(".csv", "_holdout_eval.csv")
+        csv_path   = _eval_csv
+        print(f"  *** HOLD-OUT EVAL MODE: {_run_start} → {_run_end} ***")
+        print(f"  Output: {csv_path}")
+    else:
+        _run_start = args.start or START
+        _run_end   = args.end   or END
+
     if args.print_only:
         if not os.path.exists(csv_path):
             print("No CSV found."); sys.exit(1)
@@ -842,11 +972,23 @@ if __name__ == "__main__":
         print_results(df)
     else:
         _sep()
-        print("LLM Backtest — 1D/3D/5D  |  fast mode (1 call/prediction)  |  actual NSE data 2024–2025")
+        mode_label = "HOLD-OUT EVAL" if args.eval else "TRAINING"
+        print(f"LLM Backtest [{mode_label}] — {_run_start} → {_run_end}  |  fast mode  |  actual NSE data")
         _sep()
-        sc, sh, sl, sv, nc, vc = fetch_data(LLM_UNIVERSE, DATA_START, END)
+
+        # Temporarily override module-level START/END so build_work_items uses the right range
+        _g = globals()
+        _orig_start, _orig_end = _g["START"], _g["END"]
+        _g["START"] = _run_start
+        _g["END"]   = _run_end
+
+        sc, sh, sl, sv, nc, vc = fetch_data(LLM_UNIVERSE, DATA_START, _run_end)
         indicator_cache = _build_indicator_snapshots(sc, sh, sl, sv, nc, vc)
         work_items = build_work_items(sc, sh, sl, sv, nc, vc, feature_cache=indicator_cache)
+
+        _g["START"] = _orig_start
+        _g["END"]   = _orig_end
+
         if args.timeframes:
             selected = set(args.timeframes)
             work_items = [item for item in work_items if item["tf"] in selected]

research/confidence_calibration.json

@@ -1,26 +1,33 @@
 {
-  "generated_rows": 672,
+  "generated_rows": 72,
   "timeframes": {
+    "INTRADAY": {
+      "n_total": 18,
+      "high_rate_pct": 0.0,
+      "high_hit_pct": null,
+      "medium_hit_pct": 100.0,
+      "recommendation": "promote_medium_to_high"
+    },
     "1D": {
-      "n_total": 232,
-      "high_rate_pct": 7.8,
-      "high_hit_pct": 83.3,
-      "medium_hit_pct": 76.2,
-      "recommendation": "hold"
+      "n_total": 18,
+      "high_rate_pct": 0.0,
+      "high_hit_pct": null,
+      "medium_hit_pct": 100.0,
+      "recommendation": "promote_medium_to_high"
     },
     "3D": {
-      "n_total": 232,
-      "high_rate_pct": 9.5,
-      "high_hit_pct": 100.0,
-      "medium_hit_pct": 82.6,
-      "recommendation": "hold"
+      "n_total": 18,
+      "high_rate_pct": 0.0,
+      "high_hit_pct": null,
+      "medium_hit_pct": 93.8,
+      "recommendation": "promote_medium_to_high"
     },
     "5D": {
-      "n_total": 208,
-      "high_rate_pct": 13.9,
-      "high_hit_pct": 96.6,
-      "medium_hit_pct": 78.4,
-      "recommendation": "hold"
+      "n_total": 18,
+      "high_rate_pct": 0.0,
+      "high_hit_pct": null,
+      "medium_hit_pct": 100.0,
+      "recommendation": "promote_medium_to_high"
     }
   }
 }

research/loop_backtest.py

@@ -29,6 +29,10 @@ TARGET      = 90.0
 BACKTEST_CACHE_DIR = os.path.join(os.path.dirname(__file__), "cache")
 ITERATION_BACKUP_RE = re.compile(r"ai_prompt_accuracy_iter(\d+)\.csv$")
 
+# Hold-out boundary — must match backtest.py HOLDOUT_START.
+# Training window for loop optimization = rolling 18 months ending the day before this.
+HOLDOUT_START = "2025-01-01"
+
 
 # ── ANALYSIS ──────────────────────────────────────────────────────────────────
 
@@ -430,6 +434,17 @@ def main():
             print(f"  Backed up previous results -> {os.path.basename(backup)}")
             backup_iteration += 1
 
+        # Rolling 18-month training window: optimize on recent data only, never touch hold-out.
+        # holdout_dt is 2025-01-01; rolling window ends 2024-12-31, starts 18 months earlier.
+        from datetime import datetime as _dt, timedelta as _td
+        _holdout_dt   = _dt.strptime(HOLDOUT_START, "%Y-%m-%d")
+        _train_end_dt = _holdout_dt - _td(days=1)          # 2024-12-31
+        _rolling_days = 18 * 30                            # ~18 months
+        _rolling_start_dt = _train_end_dt - _td(days=_rolling_days)
+        _rolling_start = _rolling_start_dt.strftime("%Y-%m-%d")
+        _rolling_end   = _train_end_dt.strftime("%Y-%m-%d")
+        print(f"  Rolling window: {_rolling_start} → {_rolling_end} (18-month training set, hold-out locked)")
+
         # Run backtest — explicit file handles so subprocess doesn't inherit
         # nohup's broken fds (avoids "Bad file descriptor" crash on macOS)
         print(f"\n  Running backtest (fresh historical data + prompt-only calibration)...")
@@ -441,6 +456,8 @@ def main():
                 [
                     sys.executable,
                     os.path.join(os.path.dirname(__file__), "backtest.py"),
+                    "--start", _rolling_start,
+                    "--end",   _rolling_end,
                 ],
                 cwd=os.path.dirname(__file__) + "/..",
                 stdin=subprocess.DEVNULL,

research/validate_on_trades.py

@@ -22,7 +22,7 @@ warnings.filterwarnings("ignore")
 
 from backtest import (
     fetch_data, _compute_indicators, _fwd_intraday_moves, _fwd_returns,
-    _vix_nifty_series, _simple_ml_prob, run_backtest, TIMEFRAMES,
+    _vix_nifty_series, _simple_ml_prob, run_backtest, print_results, TIMEFRAMES,
 )
 
 # ── TRADED TICKERS ─────────────────────────────────────────────────────────────
@@ -238,6 +238,11 @@ def main():
     _print_summary(df)
     print(f"\nSaved → {csv_out}")
 
+    # Full breakdown — Tables 1-5 (trigger accuracy, per-regime slice, etc.)
+    print("\n" + "=" * 70)
+    print("FULL BREAKDOWN (shared with backtest.py output)")
+    print_results(df)
+
 
 if __name__ == "__main__":
     main()

risk_engine.py

@@ -58,6 +58,23 @@ def _annualized_vol(std_per_trade: float, avg_holding_days: float) -> float:
 
 # ── EQUITY CURVE & DRAWDOWN ───────────────────────────────────────────────────
 
+def _build_equity_series(sorted_trades: list[dict]) -> list[dict]:
+    """Build equity curve as [{date, equity, trade_id}] starting at 100."""
+    series = []
+    equity = 100.0
+    for t in sorted_trades:
+        pnl = _safe(t.get("pnl_pct"), 0.0)
+        equity *= (1 + pnl / 100)
+        series.append({
+            "date":     (t.get("closed_at") or "")[:10],
+            "equity":   round(equity, 4),
+            "trade_id": t.get("id"),
+            "ticker":   t.get("ticker"),
+            "pnl_pct":  round(pnl, 3),
+        })
+    return series
+
+
 def _max_drawdown(pnl_pct_list: list[float]) -> float:
     """Max drawdown % from an ordered series of per-trade P&L %."""
     if not pnl_pct_list:
@@ -176,7 +193,7 @@ def _avg_holding_days(trades: list[dict]) -> float:
 
 # ── PUBLIC API ────────────────────────────────────────────────────────────────
 
-def get_portfolio_risk() -> dict:
+def get_portfolio_risk(include_curve: bool = False) -> dict:
     """
     Compute risk metrics from the paper trading database.
 
@@ -237,6 +254,9 @@ def get_portfolio_risk() -> dict:
     gross_loss   = abs(sum(losers))
     profit_factor = round(gross_profit / gross_loss, 3) if gross_loss > 0 else float("inf")
 
+    loss_rate = 1 - win_rate
+    expectancy = round(win_rate * avg_win - loss_rate * avg_loss, 3)
+
     # Holding period & volatility
     avg_hold = _avg_holding_days(closed)
 
@@ -265,6 +285,9 @@ def get_portfolio_risk() -> dict:
     ordered_pnl = [_safe(t["pnl_pct"]) for t in sorted_trades]
     max_dd = _max_drawdown(ordered_pnl)
 
+    # Equity curve (exposed when include_curve=True)
+    equity_series = _build_equity_series(sorted_trades) if include_curve else None
+
     # Beta (best-effort; may return None)
     beta = _compute_beta(sorted_trades)
 
@@ -279,6 +302,7 @@ def get_portfolio_risk() -> dict:
         "beta_vs_nifty":             beta,
         "portfolio_volatility_ann":  round(ann_vol * 100, 2) if ann_vol else None,
         "profit_factor":             profit_factor,
+        "expectancy":                expectancy,
         "kelly_fraction":            round(kf, 4),
         "suggested_position_size_pct": suggested_position,
         "trade_count":               n,
@@ -287,6 +311,7 @@ def get_portfolio_risk() -> dict:
         "avg_win_pct":               round(avg_win, 2),
         "avg_loss_pct":              round(avg_loss, 2),
         "computed_at":               datetime.now().isoformat(),
+        **({"equity_curve": equity_series} if include_curve else {}),
     }
 
 

static/app.js

@@ -549,6 +549,10 @@ function renderPickCard(pick, idx, idPrefix = 'pick', mode = 'top5') {
   const pickPrice = pick.price || 0;
   const sl3d  = (tfs['3D'] || {}).stop_loss  || 0;
   const tgt3d = (tfs['3D'] || {}).expected_target_price || (tfs['3D'] || {}).min_target || 0;
+  const bestTf         = pick.best_tf || '3D';
+  const slBest         = (tfs[bestTf] || {}).stop_loss || 0;
+  const tgtBest        = (tfs[bestTf] || {}).expected_target_price || (tfs[bestTf] || {}).min_target || 0;
+  const planDataBestJSON = JSON.stringify(Object.assign({}, tfs[bestTf] || {}, { timeframe: bestTf }));
   const NO_TRADE_LABELS = {
     'no_signal':       '— No signal',
     'wrong_timeframe': '— Signal ≠ horizon',
@@ -677,7 +681,9 @@ function renderPickCard(pick, idx, idPrefix = 'pick', mode = 'top5') {
       ${hasSl ? `<div class="rr-bar"><div class="rr-bar-fill" style="width:${rrPct}%"></div></div>` : ''}
     </div>`;
 
-    return `<div class="tf-cell">
+    const isBestTf = (mode !== 'watchlist') && (tf === bestTf);
+    return `<div class="tf-cell${isBestTf ? ' tf-cell--best' : ''}">
+      ${isBestTf ? '<span class="best-tf-badge">Best Bet</span>' : ''}
       <div class="tf-label">${tf === 'INTRADAY' ? 'Today' : tf}</div>
       <div class="tf-return" style="color:${isNoTrade ? 'var(--text-muted)' : retColor(d.midpoint||0)}">${retLabel}</div>
       ${noTradeDetail}
@@ -698,7 +704,7 @@ function renderPickCard(pick, idx, idPrefix = 'pick', mode = 'top5') {
     ? `<button class="btn-danger btn-sm" onclick="removeFromWatchlist('${pick.ticker}')">✕ Remove</button>
        <button class="btn-primary btn-sm" onclick='openTradeModal(${JSON.stringify(pick.ticker)},${JSON.stringify(safeCompany)},${pick.price||0},${sl3d},${tgt3d},${JSON.stringify(tfs["3D"] || {})})'>Trade</button>`
     : `<button class="btn-ghost btn-sm" onclick="addToWatchlist('${pick.ticker}','${safeCompany}')">+ Watch</button>
-       <button class="btn-primary btn-sm" onclick='openTradeModal(${JSON.stringify(pick.ticker)},${JSON.stringify(safeCompany)},${pick.price||0},${sl3d},${tgt3d},${JSON.stringify(tfs["3D"] || {})})'>Trade</button>`;
+       <button class="btn-primary btn-sm" onclick='openTradeModal(${JSON.stringify(pick.ticker)},${JSON.stringify(safeCompany)},${pick.price||0},${slBest},${tgtBest},${planDataBestJSON})'>Trade</button>`;
 
   const bareSym  = pick.ticker.replace(/\.(NS|BO)$/i, '');
   const exchange = pick.ticker.endsWith('.BO') ? 'BSE' : 'NSE';

static/style.css

@@ -396,6 +396,17 @@ input::placeholder { color: var(--text-dim); }
   display: flex; flex-direction: column; align-items: center; justify-content: flex-start;
   gap: 4px;
 }
+.tf-cell--best {
+  border-color: var(--green);
+  background: rgba(34, 197, 94, 0.06);
+}
+.best-tf-badge {
+  display: inline-flex; align-items: center;
+  font-size: 9px; font-weight: 700; letter-spacing: 0.5px; text-transform: uppercase;
+  color: var(--green);
+  background: rgba(34, 197, 94, 0.15); border: 1px solid rgba(34, 197, 94, 0.3);
+  border-radius: 3px; padding: 1px 5px; margin-bottom: 2px;
+}
 .tf-label  { font-size: 11px; color: var(--text-muted); text-transform: uppercase; letter-spacing: .6px; font-weight: 600; }
 .tf-return { font-size: 16px; font-weight: 700; font-variant-numeric: tabular-nums; line-height: 1.1; }
 .no-trade-label { font-size: 13px; font-weight: 500; color: var(--text-muted); font-style: italic; }

top5_picker.py

@@ -18,7 +18,7 @@ from typing import Optional
 
 from predictor_core import predict_stock_v2, DEFAULT_UNIVERSE, timeframe_to_dates
 
-TIMEFRAMES = ["1D", "3D", "5D"]
+TIMEFRAMES = ["INTRADAY", "1D", "3D", "5D"]
 
 # Curated liquid universe for top5 — top 75 stocks by market cap from cache.
 # Using the full 491-stock universe × 3 timeframes is too slow for interactive use.
@@ -131,8 +131,8 @@ def get_top5_picks(
     _CONF_MULT = {"HIGH": 1.0, "MEDIUM": 0.80}
     _ACCEPTED_DIRECTIONS = {"BULLISH", "SLIGHTLY BULLISH"}
 
-    def _score_5d(p: dict) -> float:
-        """Composite profit score for a 5D scan result. Higher is better."""
+    def _score_tf(p: dict) -> float:
+        """Composite profit score for any TF prediction. Higher is better."""
         base_ret = float(p.get("ret_hi") or 0.0)
 
         conf = p.get("confidence", "LOW")
@@ -166,7 +166,7 @@ def get_top5_picks(
 
     # Filter: BULLISH direction, MEDIUM+ confidence, positive return magnitude.
     # Magnitude-based thresholds (1%, R:R ≥ 1.2) are removed: incompatible with
-    # AI calibrated ranges (~0.18%); _score_5d handles ranking instead.
+    # AI calibrated ranges (~0.18%); _score_tf handles ranking instead.
     bullish = []
     for p in scan_preds.values():
         if not p:
@@ -179,9 +179,9 @@ def get_top5_picks(
             continue
         bullish.append(p)
 
-    # Sort by composite 5D profit score descending.
+    # Sort by composite profit score descending.
     # Expand candidate pool to 2× top_n so the full 1D/3D rerun has room to re-rank.
-    bullish.sort(key=_score_5d, reverse=True)
+    bullish.sort(key=_score_tf, reverse=True)
     candidate_pool_size = top_n * 2  # fetch 10 full predictions, trim to top_n at end
     candidates = bullish[:candidate_pool_size]
 
@@ -204,12 +204,12 @@ def get_top5_picks(
         }
 
     # ATR multipliers and R:R targets per timeframe (must match predictor_core.py)
-    _ATR_MULT = {"1D": 0.7, "3D": 1.1, "5D": 1.5}
-    _RR_MULT  = {"1D": 1.5, "3D": 1.7, "5D": 2.0}
+    _ATR_MULT = {"INTRADAY": 0.4, "1D": 0.7, "3D": 1.1, "5D": 1.5}
+    _RR_MULT  = {"INTRADAY": 1.2, "1D": 1.5, "3D": 1.7, "5D": 2.0}
 
     def _derive_risk(price, atr14, tf):
         """Derive SL/target from ATR when AI response risk fields are missing."""
-        if not price or not atr14:
+        if not price or not atr14 or tf not in _ATR_MULT:
             return None, None, None
         sl_risk  = _ATR_MULT[tf] * atr14
         sl_price = round(price - sl_risk, 2)
@@ -263,9 +263,26 @@ def get_top5_picks(
             "risk": {},
         }
 
-    # Step 3: Assemble picks for full candidate pool, then re-rank by 5D profit score
-    # and trim to top_n. This ensures the final list maximises 5D return even after
-    # 1D/3D full predictions might have changed direction/confidence for some stocks.
+    def _pick_best_tf(ticker: str) -> tuple[str, float]:
+        """Return the TF (1D/3D/5D) with the highest profit score for this ticker."""
+        best, best_score = "3D", 0.0
+        for tf in ["INTRADAY", "1D", "3D", "5D"]:
+            pred = ai_preds.get((ticker, tf), {})
+            if not pred or pred.get("error"):
+                continue
+            if pred.get("direction") not in _ACCEPTED_DIRECTIONS:
+                continue
+            if pred.get("confidence") not in _CONF_MULT:
+                continue
+            if float(pred.get("ret_hi") or 0.0) <= 0:
+                continue
+            s = _score_tf(pred)
+            if s > best_score:
+                best, best_score = tf, s
+        return best, best_score
+
+    # Step 3: Assemble picks for full candidate pool, then re-rank by best-TF profit score
+    # and trim to top_n. Stocks that shine in any TF (not just 5D) can now win.
     picks = []
     for stock in candidates:
         ticker  = stock["ticker"]
@@ -313,10 +330,11 @@ def get_top5_picks(
                 "actual_rr":     actual_rr,
             }
 
-        # Keep card metadata, but expose AI-led primary call and avoid
-        # strategy-derived chips/details in top5 cards.
+        # Keep card metadata, but expose AI-led primary call from the best TF.
         pick = dict(stock)
-        ai_anchor = ai_preds.get((ticker, "5D"), {})
+        best_tf, best_score = _pick_best_tf(ticker)
+        pick["best_tf"] = best_tf
+        ai_anchor = ai_preds.get((ticker, best_tf), {})
         if ai_anchor and not ai_anchor.get("error"):
             pick["direction"] = ai_anchor.get("direction", pick.get("direction"))
             pick["confidence"] = ai_anchor.get("confidence", pick.get("confidence"))
@@ -326,18 +344,18 @@ def get_top5_picks(
         pick["signal_count"] = 0
         pick["timeframes"] = timeframe_data
 
-        # Attach composite 5D score using the full (debate) prediction as anchor
-        pick["_score_5d"] = _score_5d(ai_anchor if (ai_anchor and not ai_anchor.get("error")) else stock)
+        # Rank by best-TF score so stocks strong in any TF surface to the top
+        pick["_score_best"] = best_score
         picks.append(pick)
 
-    # Re-rank by 5D composite profit score (uses full debate predictions where available)
-    picks.sort(key=lambda x: x.get("_score_5d", 0.0), reverse=True)
+    # Re-rank by best-TF composite profit score
+    picks.sort(key=lambda x: x.get("_score_best", 0.0), reverse=True)
     picks = picks[:top_n]
 
     # Assign final ranks and remove internal score field
     for i, p in enumerate(picks):
         p["rank"] = i + 1
-        p.pop("_score_5d", None)
+        p.pop("_score_best", None)
 
     return {
         "picks": picks,

whatsapp_alerts.py

@@ -0,0 +1,80 @@
+"""
+whatsapp_alerts.py — Send WhatsApp alerts for HIGH-confidence predictions via CallMeBot.
+
+One-time self-setup (done once by you in WhatsApp):
+  1. Save +34 644 60 49 11 as a contact named "CallMeBot"
+  2. Send "I allow callmebot to send me messages" to that number
+  3. You receive an API key back (e.g. 123456)
+  4. Set env vars: WHATSAPP_PHONE=91XXXXXXXXXX  WHATSAPP_APIKEY=123456
+
+No extra packages needed — uses requests (already a dependency).
+"""
+from __future__ import annotations
+import logging
+import os
+import urllib.parse
+
+logger = logging.getLogger(__name__)
+
+_CALLMEBOT_URL = "https://api.callmebot.com/whatsapp.php"
+
+
+def _build_message(pred: dict) -> str:
+    ticker = pred.get("ticker", "?").replace(".NS", "")
+    direction = pred.get("direction", "?")
+    confidence = pred.get("confidence", "?")
+    tf = pred.get("timeframe", "?")
+    entry = pred.get("entry_price") or pred.get("current_price")
+    lo = pred.get("target_price_lo")
+    hi = pred.get("target_price_hi")
+    stop = pred.get("stop_loss")
+
+    parts = [f"{ticker} | {direction} {confidence} | {tf}"]
+    if entry:
+        parts.append(f"Entry ₹{entry:.0f}")
+    if lo and hi:
+        parts.append(f"Target ₹{lo:.0f}–{hi:.0f}")
+    if stop:
+        parts.append(f"Stop ₹{stop:.0f}")
+    return " | ".join(parts)
+
+
+def send_prediction_alert(pred: dict) -> bool:
+    """
+    Send a WhatsApp message for a HIGH-confidence prediction.
+    No-ops silently if WHATSAPP_PHONE or WHATSAPP_APIKEY are not set.
+    Returns True if the message was sent successfully.
+    """
+    phone = os.environ.get("WHATSAPP_PHONE", "").strip()
+    apikey = os.environ.get("WHATSAPP_APIKEY", "").strip()
+    if not phone or not apikey:
+        return False
+
+    if (pred.get("confidence") or "").upper() != "HIGH":
+        return False
+
+    text = _build_message(pred)
+    try:
+        import requests
+        resp = requests.get(
+            _CALLMEBOT_URL,
+            params={"phone": phone, "text": text, "apikey": apikey},
+            timeout=10,
+        )
+        if resp.status_code == 200:
+            logger.info("WhatsApp alert sent: %s", text)
+            return True
+        logger.warning("WhatsApp alert HTTP %d: %s", resp.status_code, resp.text[:200])
+        return False
+    except Exception as e:
+        logger.warning("WhatsApp alert failed: %s", e)
+        return False
+
+
+def send_bulk_alerts(predictions: list[dict]) -> int:
+    """Send alerts for all HIGH-confidence predictions. Returns count sent."""
+    sent = 0
+    for pred in predictions:
+        if send_prediction_alert(pred):
+            sent += 1
+    return sent