Update app.py

app.py

@@ -373,7 +373,8 @@ def mutate_agent(symbol, timeframe, success=True):
     except Exception as e: print(f"🧬 Erreur Mutation : {e}")
 
 # --- 🧠 TRAINING ENGINE ---
-def trigger_training(symbol="SOL/USD"):
+# --- 🧠 TRAINING ENGINE ---
+def trigger_training(symbol="BTC/USD"):
     try:
         print(f"⚙️ Tentative d'entraînement pour {symbol}...")
         memory_guard()
@@ -384,15 +385,51 @@ def trigger_training(symbol="SOL/USD"):
         df_final = prepare_features_sync(symbol, '1h', limit_bars=1000)
         if df_final.empty or len(df_final) < 100: return f"❌ Données vides après indicateurs."
 
-        from ml_model import train_model as train_ml
-        from time_model import train_time_model as train_time
-        train_ml(df_final); train_time(df_final)
-        
+        # 1. Entraînement des vieux modèles (Fallback de sécurité)
+        try:
+            from ml_model import train_model as train_ml
+            from time_model import train_time_model as train_time
+            train_ml(df_final); train_time(df_final)
+        except Exception as e: print(f"⚠️ Erreur Classique: {e}")
+
+        # 2. 🦖 ENTRAÎNEMENT DU DINOSAURE (LightGBM)
+        if LGBM_AVAILABLE:
+            print("🦖 [V30] Forgeage du Cerveau LightGBM en cours...")
+            ml_cols = ["RSI", "Dist_High_24h", "Dist_Low_24h", "EMA_dist", "EMA_slope", "ATR_ratio", "VOL_ratio"]
+            
+            # On crée la cible (la prochaine bougie sera-t-elle verte ?)
+            df_final['Target'] = (df_final['close'].shift(-1) > df_final['close']).astype(int)
+            df_train = df_final.dropna(subset=ml_cols + ['Target'])
+            
+            X = df_train[ml_cols]
+            y = df_train['Target']
+            
+            params = {
+                'objective': 'binary', 
+                'metric': 'binary_logloss', 
+                'boosting_type': 'gbdt', 
+                'learning_rate': 0.05, 
+                'num_leaves': 31, 
+                'verbose': -1
+            }
+            dtrain = lgb.Dataset(X, label=y)
+            
+            # Entraînement ultra-rapide
+            model = lgb.train(params, dtrain, 100)
+            model.save_model('dino_lgbm_model.txt')
+            
+            global dino_brain
+            dino_brain = lgb.Booster(model_file='dino_lgbm_model.txt')
+            print("✅ [V30] Cerveau LightGBM créé, sauvegardé et injecté !")
+
+        # Rechargement des vieux modèles en mémoire globale
         global ml_model, time_model
         ml_model, time_model = joblib.load("ml_model_v9.pkl"), joblib.load("time_model.pkl")
         gc.collect()
-        return f"✅ IA ré-entraînée avec succès."
-    except Exception as e: return f"❌ Erreur Training : {e}"
+        
+        return f"✅ IA ré-entraînée avec succès (T-Rex en ligne !)."
+    except Exception as e: 
+        return f"❌ Erreur Training : {e}"
 
 # --- 🚀 MOTEUR AUTO-PILOTE & DREAM MODE ---
 AUTO_SYMBOLS = ["BTC/USD", "ETH/USD"]