fix training on prev inferences,

fix sim PnL calculations

core/orchestrator.py

@@ -2808,8 +2808,16 @@ class TradingOrchestrator:
                 "inference_price": current_price,  # Store price at inference time
             }
 
+            # Store previous inference for training before overwriting
+            previous_inference = self.last_inference.get(model_name)
+            
             # Store only the last inference per model (for immediate training)
             self.last_inference[model_name] = inference_record
+            
+            # If we have a previous inference, trigger training on it immediately
+            if previous_inference and not previous_inference.get("outcome_evaluated", False):
+                logger.debug(f"Triggering immediate training on previous inference for {model_name}")
+                asyncio.create_task(self._trigger_immediate_training_for_previous_inference(model_name, previous_inference, current_price))
 
             # Also save to database using database manager for future training and analysis
             asyncio.create_task(
@@ -3167,6 +3175,18 @@ class TradingOrchestrator:
                 logger.debug(f"Skipping {model_name} - already evaluated")
                 return
 
+            # Check if enough time has passed since inference (minimum 30 seconds for meaningful price movement)
+            timestamp = inference_record.get("timestamp")
+            if isinstance(timestamp, str):
+                timestamp = datetime.fromisoformat(timestamp)
+            
+            time_since_inference = (datetime.now() - timestamp).total_seconds()
+            min_training_delay = 30  # Minimum 30 seconds before training
+            
+            if time_since_inference < min_training_delay:
+                logger.debug(f"Skipping {model_name} - only {time_since_inference:.1f}s since inference (minimum {min_training_delay}s required)")
+                return
+
             # Get current price for outcome evaluation
             current_price = self._get_current_price(symbol)
             if current_price is None:
@@ -3379,6 +3399,20 @@ class TradingOrchestrator:
         except Exception as e:
             logger.error(f"Error in immediate training for {model_name}: {e}")
 
+    async def _trigger_immediate_training_for_previous_inference(self, model_name: str, previous_inference: Dict, current_price: float):
+        """Trigger immediate training for a previous inference with current price"""
+        try:
+            logger.info(f"Training {model_name} on previous inference with current price: {current_price}")
+            
+            # Evaluate the previous prediction and train the model immediately
+            await self._evaluate_and_train_on_record(previous_inference, current_price)
+            
+            # Mark as evaluated
+            previous_inference["outcome_evaluated"] = True
+            
+        except Exception as e:
+            logger.error(f"Error in immediate training for previous inference {model_name}: {e}")
+
     async def _evaluate_and_train_on_record(self, record: Dict, current_price: float):
         """Evaluate prediction outcome and train model"""
         try:
@@ -4534,6 +4568,7 @@ class TradingOrchestrator:
                         price_direction_pred,
                         features_refined,
                         advanced_pred,
+                        multi_timeframe_pred,
                     ) = self.cnn_model(features_tensor)
 
                     # Calculate primary Q-value loss

core/trading_executor.py

@@ -1441,10 +1441,13 @@ class TradingExecutor:
         
         if self.simulation_mode:
             logger.info(f"SIMULATION MODE ({self.trading_mode.upper()}) - Short close logged but not executed")
-            # Calculate simulated fees in simulation mode
+            # Calculate simulated fees in simulation mode - FIXED to include both entry and exit fees
             trading_fees = self.exchange_config.get('trading_fees', {})
             taker_fee_rate = trading_fees.get('taker_fee', trading_fees.get('default_fee', 0.0006))
-            simulated_fees = position.quantity * current_price * taker_fee_rate
+            # Calculate both entry and exit fees
+            entry_fee = position.quantity * position.entry_price * taker_fee_rate
+            exit_fee = position.quantity * current_price * taker_fee_rate
+            simulated_fees = entry_fee + exit_fee
 
             # Get current leverage setting
             leverage = self.get_leverage()
@@ -1452,8 +1455,8 @@ class TradingExecutor:
             # Calculate position size in USD
             position_size_usd = position.quantity * position.entry_price
             
-            # Calculate gross PnL (before fees) with leverage
+            # Calculate gross PnL (before fees) with leverage - FIXED for SHORT positions
-            gross_pnl = (current_price - position.entry_price) * position.quantity * leverage
+            gross_pnl = (position.entry_price - current_price) * position.quantity * leverage
             
             # Calculate net PnL (after fees)
             net_pnl = gross_pnl - simulated_fees
@@ -1543,8 +1546,8 @@ class TradingExecutor:
                 # Calculate position size in USD
                 position_size_usd = position.quantity * position.entry_price
                 
-                # Calculate gross PnL (before fees) with leverage
+                # Calculate gross PnL (before fees) with leverage - FIXED for SHORT positions
-                gross_pnl = (current_price - position.entry_price) * position.quantity * leverage
+                gross_pnl = (position.entry_price - current_price) * position.quantity * leverage
                 
                 # Calculate net PnL (after fees)
                 net_pnl = gross_pnl - fees
@@ -1619,10 +1622,13 @@ class TradingExecutor:
         
         if self.simulation_mode:
             logger.info(f"SIMULATION MODE ({self.trading_mode.upper()}) - Long close logged but not executed")
-            # Calculate simulated fees in simulation mode
+            # Calculate simulated fees in simulation mode - FIXED to include both entry and exit fees
             trading_fees = self.exchange_config.get('trading_fees', {})
             taker_fee_rate = trading_fees.get('taker_fee', trading_fees.get('default_fee', 0.0006))
-            simulated_fees = position.quantity * current_price * taker_fee_rate
+            # Calculate both entry and exit fees
+            entry_fee = position.quantity * position.entry_price * taker_fee_rate
+            exit_fee = position.quantity * current_price * taker_fee_rate
+            simulated_fees = entry_fee + exit_fee
 
             # Get current leverage setting
             leverage = self.get_leverage()

data/ui_state.json

@@ -21,9 +21,9 @@
             "training_enabled": true
         },
         "dqn_agent": {
-            "inference_enabled": false,
+            "inference_enabled": true,
             "training_enabled": true
         }
     },
-    "timestamp": "2025-07-30T00:17:57.738273"
+    "timestamp": "2025-07-30T00:41:19.241862"
 }