freqai同时优化回归和分类参数

freqtrade/templates/freqaiprimer.py

@@ -265,7 +265,13 @@ class FreqaiPrimer(IStrategy):
         ]
         choices = [0, 1, 2, 3]
         
-        dataframe["&*-optimal_first_length"] = np.select(conditions, choices, default=4)
+        # 确保分类目标列是整数类型
+        optimal_length_class = np.select(conditions, choices, default=4)
+        dataframe["&*-optimal_first_length"] = optimal_length_class.astype(int)
+        
+        # 添加调试信息
+        logger.info(f"[{pair}] 📈 分类目标统计: {dict(pd.Series(optimal_length_class).value_counts())}")
+        logger.info(f"[{pair}] 🎯 最新分类值: {optimal_length_class[-5:]}")
         
         return dataframe
     def is_stochrsi_overbought(self, dataframe: DataFrame, period=10, threshold=85) -> bool:
@@ -1343,14 +1349,22 @@ class FreqaiPrimer(IStrategy):
             # 使用分类模型预测最优 first_length
             if "&*-optimal_first_length" in dataframe.columns:
                 # 获取最新预测值
-                optimal_length_class = int(dataframe["&*-optimal_first_length"].iloc[-1])
-                length_mapping = {0: 2, 1: 4, 2: 6, 3: 8, 4: 10}
-                first_length = length_mapping.get(optimal_length_class, 2)
-                logger.info(f"[{pair}] 使用分类模型优化 first_length: {first_length} (类别: {optimal_length_class})")
+                predicted_value = dataframe["&*-optimal_first_length"].iloc[-1]
+                if pd.notna(predicted_value):
+                    optimal_length_class = int(predicted_value)
+                    length_mapping = {0: 2, 1: 4, 2: 6, 3: 8, 4: 10}
+                    first_length = length_mapping.get(optimal_length_class, 2)
+                    logger.info(f"[{pair}] ✅ 分类模型预测成功: first_length={first_length} (类别: {optimal_length_class})")
+                else:
+                    first_length = 2  # 保持默认值为2
+                    logger.warning(f"[{pair}] ⚠️ 分类模型预测为NaN，使用默认值: {first_length}")
             else:
-                # 回退到默认值
-                first_length = 2
-                logger.warning(f"[{pair}] 分类模型未预测，使用默认 first_length: {first_length}")
+                first_length = 2  # 保持默认值为2
+                logger.warning(f"[{pair}] ❌ 分类模型列不存在，使用默认值: {first_length}")
+                
+            # 调试信息：显示当前dataframe列
+            ai_columns = [col for col in dataframe.columns if col.startswith('&') or col.startswith('&*')]
+            logger.info(f"[{pair}] 📊 当前AI预测列: {ai_columns}")
             
             # 根据 first_length 动态调整其他段长
             second_length = first_length * 3

test_mixed_models_config.json

@@ -0,0 +1,102 @@
+{
+  "strategy": "freqaiprimer",
+  "max_open_trades": 3,
+  "stake_currency": "USDT",
+  "stake_amount": 25,
+  "tradable_balance_ratio": 0.99,
+  "fiat_display_currency": "USD",
+  "dry_run": true,
+  "cancel_open_orders_on_exit": false,
+  "trading_mode": "spot",
+  "exchange": {
+    "name": "binance",
+    "key": "",
+    "secret": "",
+    "ccxt_config": {"enableRateLimit": true},
+    "ccxt_async_config": {"enableRateLimit": true},
+    "pair_blacklist": [
+      "BNB/.*"
+    ]
+  },
+  "pairlists": [
+    {
+      "method": "StaticPairList",
+      "pairs": ["SOL/USDT", "WCT/USDT"]
+    }
+  ],
+  "freqai": {
+    "enabled": true,
+    "identifier": "test_mixed_models",
+    "purge_old_models": 2,
+    "train_period_days": 15,
+    "backtest_period_days": 7,
+    "live_retrain_candles": 100,
+    "fit_live_predictions_candles": 100,
+    "feature_parameters": {
+      "include_timeframes": ["3m", "15m", "1h"],
+      "label_period_candles": 12,
+      "include_shifted_candles": 3,
+      "indicator_periods_candles": [10, 20],
+      "DI_threshold": 30,
+      "weight_factor": 0.9
+    },
+    "data_split_parameters": {
+      "test_size": 0.2,
+      "shuffle": false
+    },
+    "model_training_parameters": {
+      "price_value_divergence": {
+        "model": "LightGBMRegressor",
+        "model_params": {
+          "n_estimators": 200,
+          "learning_rate": 0.05,
+          "num_leaves": 31,
+          "verbose": -1
+        }
+      },
+      "optimal_first_length": {
+        "model": "LightGBMClassifier",
+        "model_params": {
+          "n_estimators": 150,
+          "learning_rate": 0.1,
+          "num_leaves": 15,
+          "max_depth": 8,
+          "min_child_samples": 10,
+          "class_weight": "balanced",
+          "verbose": -1
+        }
+      }
+    }
+  },
+  "timeframe": "3m",
+  "dry_run_wallet": 1000,
+  "unfilledtimeout": {
+    "entry": 10,
+    "exit": 30
+  },
+  "entry_pricing": {
+    "price_side": "same",
+    "use_order_book": true,
+    "order_book_top": 1,
+    "check_depth_of_market": {
+      "enabled": false,
+      "bids_to_ask_delta": 1
+    }
+  },
+  "exit_pricing": {
+    "price_side": "same",
+    "use_order_book": true,
+    "order_book_top": 1
+  },
+  "order_types": {
+    "entry": "limit",
+    "exit": "limit",
+    "emergency_exit": "market",
+    "force_exit": "market",
+    "stoploss": "market",
+    "stoploss_on_exchange": false
+  },
+  "edge": {
+    "enabled": false
+  }
+}