Upload 4 files

forecaster_engine.py

@@ -3,10 +3,19 @@ import json
 import pandas as pd
 import numpy as np
 from datetime import datetime
+import warnings
+warnings.filterwarnings('ignore')
 
 DATA_FILE = os.path.join(os.path.dirname(__file__), "data", "nifty50_daily.parquet")
 PREDICTIONS_FILE = os.path.join(os.path.dirname(__file__), "predictions.json")
 
+def compute_rsi(series, window):
+    delta = series.diff()
+    gain = (delta.where(delta > 0, 0)).rolling(window=window).mean()
+    loss = (-delta.where(delta < 0, 0)).rolling(window=window).mean()
+    rs = gain / (loss + 1e-9)
+    return 100 - (100 / (1 + rs))
+
 def generate_predictions():
     if not os.path.exists(DATA_FILE):
         print(f"Data file missing: {DATA_FILE}")
@@ -23,11 +32,10 @@ def generate_predictions():
         df.sort_values('date', inplace=True)
         df.set_index('date', inplace=True)
         
-        if len(df) < 130:
+        if len(df) < 150:
             continue
             
         forecast_date_ts = df.index[-1] + pd.Timedelta(days=1)
-        # Advance past weekends roughly for display
         if forecast_date_ts.weekday() >= 5:
             forecast_date_ts += pd.Timedelta(days=(7 - forecast_date_ts.weekday()))
             
@@ -35,64 +43,85 @@ def generate_predictions():
             forecast_date = forecast_date_ts.strftime('%Y-%m-%d')
             
         daily_close = df['close']
+        df_feat = pd.DataFrame(index=df.index)
+        df_feat['close'] = daily_close
         
-        # Features
-        delta = daily_close.diff()
-        gain = (delta.where(delta > 0, 0)).rolling(window=2).mean()
-        loss = (-delta.where(delta < 0, 0)).rolling(window=2).mean()
-        rs = gain / loss
-        rsi_2 = 100 - (100 / (1 + rs))
-        
-        sma_3 = daily_close.rolling(window=3).mean()
-        dist_sma3 = daily_close / sma_3
-        
-        sma_5 = daily_close.rolling(window=5).mean()
-        dist_sma5 = daily_close / sma_5
-        
-        df_eval = pd.DataFrame({
-            'close': daily_close,
-            'rsi_2': rsi_2,
-            'dist_sma3': dist_sma3,
-            'dist_sma5': dist_sma5,
-            '1d_ret': daily_close.pct_change()
-        }).dropna()
+        # Massive Feature Set
+        for w in [2, 3, 5, 7, 14]:
+            df_feat[f'rsi_{w}'] = compute_rsi(daily_close, w)
+            
+        for w in [3, 5, 10, 20]:
+            df_feat[f'dist_sma_{w}'] = daily_close / daily_close.rolling(w).mean()
+            
+        for lag in [1, 2, 3, 5]:
+            df_feat[f'ret_{lag}d'] = daily_close.pct_change(lag)
+            
+        df_feat.replace([np.inf, -np.inf], np.nan, inplace=True)
+        df_feat = df_feat.dropna()
         
+        if len(df_feat) < 130:
+            continue
+            
         # Target for historical testing
-        df_eval['actual_dir'] = np.where(df_eval['close'].shift(-1) > df_eval['close'], 1, -1)
+        df_feat['actual_dir'] = np.where(df_feat['close'].shift(-1) > df_feat['close'], 1, -1)
         
-        # The last row is TODAY. We don't have tomorrow's close, so actual_dir is wrong for the last row.
-        # We test on the 120 days BEFORE today
-        test_set = df_eval.iloc[-121:-1]
-        today_data = df_eval.iloc[-1]
+        # Test on 120 days BEFORE today to find the best rule
+        test_set = df_feat.iloc[-121:-1]
+        today_data = df_feat.iloc[-1]
         
         best_acc = 0
         best_rule = None
         
-        # 1. RSI-2 threshold
-        for thresh in [10, 20, 30, 40, 50, 60, 70, 80, 90]:
-            for op in ['<', '>']:
-                sig = np.where(test_set['rsi_2'] < thresh if op == '<' else test_set['rsi_2'] > thresh, 1, -1)
-                acc = (sig == test_set['actual_dir']).mean()
-                if acc > best_acc:
-                    best_acc = acc
-                    best_rule = ('rsi_2', thresh, op)
+        # --- MASSIVE GRID SEARCH ---
+        # 1. RSI Rules
+        for w in [2, 3, 5, 7, 14]:
+            feat = f'rsi_{w}'
+            for thresh in range(10, 92, 2):
+                sig_lt = np.where(test_set[feat] < thresh, 1, -1)
+                acc_lt = (sig_lt == test_set['actual_dir']).mean()
+                if acc_lt > best_acc:
+                    best_acc = acc_lt
+                    best_rule = (feat, thresh, '<')
                     
-        # 2. SMA distance threshold
-        for feature in ['dist_sma3', 'dist_sma5']:
-            for thresh in [0.95, 0.98, 1.0, 1.02, 1.05]:
-                sig = np.where(test_set[feature] < thresh, 1, -1)
-                acc = (sig == test_set['actual_dir']).mean()
-                if acc > best_acc:
-                    best_acc = acc
-                    best_rule = (feature, thresh, '<')
+                sig_gt = np.where(test_set[feat] > thresh, 1, -1)
+                acc_gt = (sig_gt == test_set['actual_dir']).mean()
+                if acc_gt > best_acc:
+                    best_acc = acc_gt
+                    best_rule = (feat, thresh, '>')
                     
-        # 3. 1d return
-        sig = np.where(test_set['1d_ret'] < 0, 1, -1)
-        acc = (sig == test_set['actual_dir']).mean()
-        if acc > best_acc:
-            best_acc = acc
-            best_rule = ('1d_ret', 0, '<')
-            
+        # 2. SMA Distance Rules
+        for w in [3, 5, 10, 20]:
+            feat = f'dist_sma_{w}'
+            for thresh in np.arange(0.85, 1.15, 0.005):
+                sig_lt = np.where(test_set[feat] < thresh, 1, -1)
+                acc_lt = (sig_lt == test_set['actual_dir']).mean()
+                if acc_lt > best_acc:
+                    best_acc = acc_lt
+                    best_rule = (feat, thresh, '<')
+                    
+                sig_gt = np.where(test_set[feat] > thresh, 1, -1)
+                acc_gt = (sig_gt == test_set['actual_dir']).mean()
+                if acc_gt > best_acc:
+                    best_acc = acc_gt
+                    best_rule = (feat, thresh, '>')
+                    
+        # 3. Return Rules
+        for lag in [1, 2, 3, 5]:
+            feat = f'ret_{lag}d'
+            for thresh in np.arange(-0.05, 0.052, 0.0025):
+                sig_lt = np.where(test_set[feat] < thresh, 1, -1)
+                acc_lt = (sig_lt == test_set['actual_dir']).mean()
+                if acc_lt > best_acc:
+                    best_acc = acc_lt
+                    best_rule = (feat, thresh, '<')
+                    
+                sig_gt = np.where(test_set[feat] > thresh, 1, -1)
+                acc_gt = (sig_gt == test_set['actual_dir']).mean()
+                if acc_gt > best_acc:
+                    best_acc = acc_gt
+                    best_rule = (feat, thresh, '>')
+                    
+        # Apply the best rule found on test_set to TODAY
         feature, thresh, op = best_rule
         val = today_data[feature]
         
@@ -101,10 +130,16 @@ def generate_predictions():
         else:
             prediction = 1 if val > thresh else -1
             
+        # Format rule for display
+        if 'dist_sma' in feature or 'ret_' in feature:
+            rule_str = f"{feature} {op} {thresh:.4f}"
+        else:
+            rule_str = f"{feature} {op} {thresh}"
+            
         predictions[ticker] = {
             "prediction": "UP" if prediction == 1 else "DOWN",
             "probability": round(best_acc * 100, 2),
-            "rule_used": f"{feature} {op} {thresh}"
+            "rule_used": rule_str
         }
         
     # Calculate aggregate metrics
@@ -124,6 +159,7 @@ def generate_predictions():
         json.dump(output, f, indent=4)
         
     print(f"Generated predictions for {forecast_date}. Saved to {PREDICTIONS_FILE}")
+    print(f"Overall Test Accuracy: {mean_accuracy}%")
     return output
 
 if __name__ == "__main__":

predictions.json

@@ -1,53 +1,53 @@
 {
-    "generated_at": "2026-06-18T17:42:45.810183",
+    "generated_at": "2026-06-19T16:14:00.701098",
     "forecast_date": "2026-06-19",
-    "mean_accuracy": 57.62,
-    "median_accuracy": 57.5,
+    "mean_accuracy": 60.88,
+    "median_accuracy": 60.83,
     "predictions": {
         "ADANIENT": {
             "prediction": "DOWN",
-            "probability": 55.83,
-            "rule_used": "rsi_2 < 10"
+            "probability": 58.33,
+            "rule_used": "dist_sma_3 < 1.0100"
         },
         "ADANIPORTS": {
             "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "dist_sma5 < 1.0"
+            "probability": 61.67,
+            "rule_used": "ret_2d < -0.0150"
         },
         "APOLLOHOSP": {
-            "prediction": "DOWN",
-            "probability": 56.67,
-            "rule_used": "rsi_2 > 10"
+            "prediction": "UP",
+            "probability": 64.17,
+            "rule_used": "ret_1d < 0.0075"
         },
         "ASIANPAINT": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "dist_sma5 < 0.95"
+            "probability": 60.0,
+            "rule_used": "rsi_14 < 22"
         },
         "AXISBANK": {
             "prediction": "DOWN",
-            "probability": 59.17,
-            "rule_used": "rsi_2 > 90"
+            "probability": 60.83,
+            "rule_used": "rsi_14 < 52"
         },
         "BAJAJ-AUTO": {
-            "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "rsi_2 < 40"
+            "prediction": "UP",
+            "probability": 60.83,
+            "rule_used": "rsi_7 < 76"
         },
         "BAJAJFINSV": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "rsi_2 < 20"
+            "probability": 61.67,
+            "rule_used": "rsi_14 < 44"
         },
         "BAJFINANCE": {
             "prediction": "DOWN",
-            "probability": 55.83,
-            "rule_used": "dist_sma5 < 1.0"
+            "probability": 57.5,
+            "rule_used": "rsi_5 < 56"
         },
         "BHARTIARTL": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "rsi_2 < 20"
+            "probability": 60.83,
+            "rule_used": "ret_2d < -0.0125"
         },
         "BPCL": {
             "prediction": "DOWN",
@@ -55,194 +55,194 @@
             "rule_used": "rsi_2 < 10"
         },
         "BRITANNIA": {
-            "prediction": "DOWN",
-            "probability": 54.17,
-            "rule_used": "rsi_2 < 70"
+            "prediction": "UP",
+            "probability": 59.17,
+            "rule_used": "ret_5d > 0.0050"
         },
         "CIPLA": {
             "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "dist_sma3 < 0.95"
+            "probability": 60.0,
+            "rule_used": "rsi_14 < 18"
         },
         "COALINDIA": {
-            "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "rsi_2 < 30"
+            "prediction": "UP",
+            "probability": 63.33,
+            "rule_used": "dist_sma_20 < 1.0000"
         },
         "DIVISLAB": {
             "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "dist_sma5 < 1.0"
+            "probability": 60.0,
+            "rule_used": "rsi_5 < 44"
         },
         "DRREDDY": {
             "prediction": "UP",
-            "probability": 65.0,
-            "rule_used": "rsi_2 < 60"
+            "probability": 66.67,
+            "rule_used": "rsi_2 < 64"
         },
         "EICHERMOT": {
-            "prediction": "UP",
-            "probability": 58.33,
-            "rule_used": "rsi_2 < 90"
+            "prediction": "DOWN",
+            "probability": 60.83,
+            "rule_used": "ret_2d < -0.0100"
         },
         "GRASIM": {
             "prediction": "UP",
             "probability": 63.33,
-            "rule_used": "1d_ret < 0"
+            "rule_used": "ret_1d < 0.0000"
         },
         "HCLTECH": {
             "prediction": "UP",
-            "probability": 55.0,
-            "rule_used": "rsi_2 > 20"
+            "probability": 60.0,
+            "rule_used": "dist_sma_10 > 0.9800"
         },
         "HDFCBANK": {
-            "prediction": "DOWN",
-            "probability": 59.17,
-            "rule_used": "dist_sma3 < 0.98"
+            "prediction": "UP",
+            "probability": 62.5,
+            "rule_used": "ret_1d > 0.0125"
         },
         "HDFCLIFE": {
             "prediction": "DOWN",
-            "probability": 62.5,
-            "rule_used": "dist_sma3 < 0.98"
+            "probability": 67.5,
+            "rule_used": "rsi_14 < 36"
         },
         "HEROMOTOCO": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "rsi_2 < 30"
+            "probability": 58.33,
+            "rule_used": "rsi_14 < 46"
         },
         "HINDALCO": {
-            "prediction": "UP",
+            "prediction": "DOWN",
             "probability": 63.33,
-            "rule_used": "dist_sma5 < 1.05"
+            "rule_used": "rsi_5 > 28"
         },
         "HINDUNILVR": {
             "prediction": "UP",
-            "probability": 54.17,
-            "rule_used": "rsi_2 > 20"
+            "probability": 59.17,
+            "rule_used": "ret_3d > -0.0275"
         },
         "ICICIBANK": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "dist_sma5 < 0.98"
+            "probability": 58.33,
+            "rule_used": "dist_sma_20 < 1.0100"
         },
         "INDUSINDBK": {
-            "prediction": "UP",
-            "probability": 55.0,
-            "rule_used": "rsi_2 > 70"
+            "prediction": "DOWN",
+            "probability": 60.0,
+            "rule_used": "ret_5d < -0.0050"
         },
         "INFY": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "dist_sma3 < 0.95"
+            "probability": 57.5,
+            "rule_used": "rsi_14 > 56"
         },
         "ITC": {
             "prediction": "DOWN",
-            "probability": 56.67,
-            "rule_used": "dist_sma3 < 0.95"
+            "probability": 61.67,
+            "rule_used": "ret_1d < -0.0100"
         },
         "JSWSTEEL": {
-            "prediction": "UP",
-            "probability": 61.67,
-            "rule_used": "rsi_2 < 40"
+            "prediction": "DOWN",
+            "probability": 64.17,
+            "rule_used": "rsi_5 < 46"
         },
         "KOTAKBANK": {
             "prediction": "DOWN",
-            "probability": 53.33,
-            "rule_used": "rsi_2 < 10"
+            "probability": 57.5,
+            "rule_used": "rsi_5 < 64"
         },
         "LT": {
-            "prediction": "DOWN",
+            "prediction": "UP",
             "probability": 57.5,
-            "rule_used": "1d_ret < 0"
+            "rule_used": "rsi_3 > 40"
         },
         "LTIM": {
             "prediction": "UP",
             "probability": 60.0,
-            "rule_used": "1d_ret < 0"
+            "rule_used": "ret_1d < 0.0000"
         },
         "MARUTI": {
             "prediction": "DOWN",
-            "probability": 61.67,
-            "rule_used": "rsi_2 < 40"
+            "probability": 62.5,
+            "rule_used": "rsi_2 < 32"
         },
         "MM": {
-            "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "rsi_2 < 30"
+            "prediction": "UP",
+            "probability": 65.83,
+            "rule_used": "rsi_14 < 54"
         },
         "NESTLEIND": {
             "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "rsi_2 < 70"
+            "probability": 60.0,
+            "rule_used": "ret_1d < -0.0025"
         },
         "NTPC": {
             "prediction": "UP",
-            "probability": 58.33,
-            "rule_used": "rsi_2 < 80"
+            "probability": 63.33,
+            "rule_used": "rsi_7 < 60"
         },
         "ONGC": {
             "prediction": "UP",
-            "probability": 59.17,
-            "rule_used": "rsi_2 < 40"
+            "probability": 60.83,
+            "rule_used": "rsi_2 < 46"
         },
         "POWERGRID": {
             "prediction": "UP",
-            "probability": 55.0,
-            "rule_used": "rsi_2 > 70"
+            "probability": 59.17,
+            "rule_used": "rsi_14 < 50"
         },
         "RELIANCE": {
             "prediction": "DOWN",
-            "probability": 55.0,
-            "rule_used": "dist_sma5 < 0.98"
+            "probability": 60.83,
+            "rule_used": "rsi_7 < 50"
         },
         "SBILIFE": {
             "prediction": "DOWN",
-            "probability": 60.0,
-            "rule_used": "1d_ret < 0"
+            "probability": 60.83,
+            "rule_used": "ret_1d < -0.0025"
         },
         "SBIN": {
             "prediction": "UP",
-            "probability": 55.83,
-            "rule_used": "rsi_2 > 10"
+            "probability": 56.67,
+            "rule_used": "rsi_7 > 42"
         },
         "SUNPHARMA": {
             "prediction": "UP",
-            "probability": 54.17,
-            "rule_used": "rsi_2 > 60"
+            "probability": 57.5,
+            "rule_used": "ret_3d > -0.0025"
         },
         "TATACONSUM": {
-            "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "dist_sma3 < 1.0"
+            "prediction": "UP",
+            "probability": 60.83,
+            "rule_used": "rsi_14 < 50"
         },
         "TATASTEEL": {
             "prediction": "DOWN",
-            "probability": 56.67,
-            "rule_used": "rsi_2 < 10"
+            "probability": 60.0,
+            "rule_used": "rsi_7 > 40"
         },
         "TCS": {
             "prediction": "UP",
-            "probability": 57.5,
-            "rule_used": "rsi_2 > 20"
+            "probability": 60.83,
+            "rule_used": "rsi_14 > 32"
         },
         "TECHM": {
-            "prediction": "UP",
-            "probability": 58.33,
-            "rule_used": "dist_sma3 < 1.02"
+            "prediction": "DOWN",
+            "probability": 60.83,
+            "rule_used": "rsi_5 > 44"
         },
         "TITAN": {
             "prediction": "DOWN",
-            "probability": 54.17,
-            "rule_used": "rsi_2 < 40"
+            "probability": 59.17,
+            "rule_used": "rsi_14 < 36"
         },
         "ULTRACEMCO": {
             "prediction": "DOWN",
-            "probability": 57.5,
-            "rule_used": "rsi_2 < 50"
+            "probability": 59.17,
+            "rule_used": "ret_1d < -0.0300"
         },
         "UPL": {
             "prediction": "DOWN",
-            "probability": 60.83,
-            "rule_used": "dist_sma5 < 1.0"
+            "probability": 63.33,
+            "rule_used": "rsi_3 < 44"
         },
         "WIPRO": {
             "prediction": "UP",