Add classifier to the code

ChanKLC.py

@@ -32,7 +32,7 @@ class ChanKLC():
         self.distance = 0
         self.klc_fx_type = Chan_KLC_FX.UNKNOWN
     def set_klc_fx_type(self, klc_fx_type):
-        print(self.start_time, klc_fx_type, self.get_feature_data()['klu_macd'], self.get_feature_data()['klu_macdhist'], self.get_feature_data()['klu_rsi'])
+        #print(self.start_time, klc_fx_type, self.get_feature_data()['klu_macd'], self.get_feature_data()['klu_macdhist'], self.get_feature_data()['klu_rsi'])
         if self.check_klc_fx_type(klc_fx_type):
             self.klc_fx_type = klc_fx_type
     def check_klc_fx_type(self, klc_fx_type):

ChanLun_Classifier.py

@@ -6,7 +6,7 @@ sys.path.append(os.path.abspath("/Users/jack/Project/freqtrade/user_data/Chan"))
 import numpy as np
 from datetime import timedelta
 from pandas import DataFrame
-from ChanEnum import Chan_FX_TYPE, Chan_KLINE_DIR, Chan_BI_DIR, Chan_SEG_DIR, Chan_ZS_DIR, Chan_BSP_DIR, Chan_BSP_TYPE
+from ChanEnum import Chan_FX_TYPE, Chan_KLINE_DIR, Chan_BI_DIR, Chan_SEG_DIR, Chan_ZS_DIR, Chan_BSP_DIR, Chan_BSP_TYPE, Chan_KLC_FX
 from ChanKLU import ChanKLU
 from ChanKLC import ChanKLC
 from ChanBI import ChanBI
@@ -279,31 +279,24 @@ class ChanLunClassifier:
         :return: 特征矩阵X和标签y
         """
         # 使用ChanLun获取bi_list
-        bi_list = self.chan.cal_bi_list(self.chan.get_klc_list(dataframe))
         klc_list = self.chan.get_klc_list(dataframe)
+        bi_list = self.chan.cal_bi_list(klc_list)
         seg_list = self.chan.get_seg_list(bi_list)
         # 筛选方向为UP的bi的起始klc
         feature_data = []
         labels = []
         
-        bi_index = 0
+        bi_index = 1
         sample_list = []
         for klc in klc_list:
-            if klc.pre and klc.next:
+            if klc.klc_fx_type != Chan_KLC_FX.UNKNOWN:
-                if klc.high > klc.pre.high and klc.high > klc.next.high:
-                    klc.set_fx(Chan_FX_TYPE.TOP)
-                elif klc.low < klc.pre.low and klc.low < klc.next.low:
-                    klc.set_fx(Chan_FX_TYPE.BOTTOM)
-                else:
-                    klc.set_fx(Chan_FX_TYPE.UNKNOWN)
-            if klc.fx != Chan_FX_TYPE.UNKNOWN:
                 sample_list.append(klc)
         for klc in sample_list:
             if bi_index >= len(bi_list):
                 bi_index = len(bi_list) - 1
-            bi = bi_list[bi_index]
+            #bi = bi_list[bi_index]
-            if klc.end_klu and bi.end_klc and klc.start_klu.index >= bi.start_klc.start_klu.index and klc.end_klu.index <= bi.end_klc.end_klu.index:
+            #if klc.end_klu and bi.end_klc and klc.start_klu.index >= bi.start_klc.start_klu.index and klc.end_klu.index <= bi.end_klc.end_klu.index:
-                klc.set_bi(bi)
+                #klc.set_bi(bi)
             
             # 提取特征
             features = klc.get_feature_data()
@@ -318,17 +311,21 @@ class ChanLunClassifier:
             
             # 判断这个bi是否赚钱（这里简单定义为：如果bi的结束价格高于起始价格，则标记为1，否则为0）
             # 这个标签定义可以根据实际需求修改
-            bi = bi_list[bi_index]
+            matched = False
-            if bi.start_klc.index == klc.index:
+            for bi in bi_list:
-            #if klc.index == seg.start_bi.start_klc.index and seg.dir == Chan_SEG_DIR.UP:
+                if bi.end_klc and bi.end_klc.index == klc.index:
                     label = 1
-                bi_index += 2
+                    matched = True
-            else:
+                    break
+            if not matched:
                 label = 0
             
             feature_data.append(feature_vec)
             labels.append(label)
-        print("Trainning data: ", klc_list[-1].start_time, klc_list[-1].fx)
+        # 在return前添加
+        positive_count = np.sum(labels)
+        print(f"正样本数量: {positive_count}, 负样本数量: {len(labels) - positive_count}")
+        print("Trainning data: ", len(feature_data), klc_list[-1].start_time, klc_list[-1].klc_fx_type , "---------------------")
         return np.array(feature_data), np.array(labels)
     def get_validate_feature_data(self, dataframe):
         """
@@ -337,23 +334,16 @@ class ChanLunClassifier:
         :return: 特征矩阵X和标签y
         """
         # 使用ChanLun获取bi_list
-        bi_list = self.chan.cal_bi_list(self.chan.get_klc_list(dataframe))
         klc_list = self.chan.get_klc_list(dataframe)
+        bi_list = self.chan.cal_bi_list(klc_list)
         seg_list = self.chan.get_seg_list(bi_list)
         # 筛选方向为UP的bi的起始klc
         feature_data = []
         labels = []
-        bi_index = 0
+        bi_index = 1
         sample_list = []
         for klc in klc_list:
-            if klc.pre and klc.next:
+            if klc.klc_fx_type != Chan_KLC_FX.UNKNOWN:
-                if klc.high > klc.pre.high and klc.high > klc.next.high:
-                    klc.set_fx(Chan_FX_TYPE.TOP)
-                elif klc.low < klc.pre.low and klc.low < klc.next.low:
-                    klc.set_fx(Chan_FX_TYPE.BOTTOM)
-                else:
-                    klc.set_fx(Chan_FX_TYPE.UNKNOWN)
-            if klc.fx != Chan_FX_TYPE.UNKNOWN:
                 sample_list.append(klc)
         for klc in sample_list:
             if bi_index >= len(bi_list):
@@ -371,16 +361,18 @@ class ChanLunClassifier:
                 else:
                     feature_vec.append(0)
             seg = seg_list[bi_index]
-            if bi.start_klc.index == klc.index:
+            matched = False
-            #if klc.index == seg.start_bi.start_klc.index and seg.dir == Chan_SEG_DIR.UP:
+            for bi in bi_list:
+                if bi.end_klc and bi.end_klc.index == klc.index:
                     label = 1
-                bi_index += 2
+                    matched = True
-            else:
+                    break
+            if not matched:
                 label = 0
             
             feature_data.append(feature_vec)
             labels.append(label)
-        
+        print("Validating data: ", len(feature_data), klc_list[-1].start_time, klc_list[-1].klc_fx_type , "---------------------")
         return np.array(feature_data), np.array(labels)
     def validate_model(self, dataframe=None):
         """

config/ChanLun_SOL.json

@@ -43,7 +43,7 @@
         "ccxt_config": {},
         "ccxt_async_config": {},
         "pair_whitelist": [
-            "BTC/USDT:USDT",
+            "SOL/USDT:USDT",
         ],
         "pair_blacklist": [
             "BNB/.*"

strategies/ChanLun_SOL_5.py

@@ -21,10 +21,10 @@ logger = logging.getLogger(__name__)
 ### Now you can use logger.info('asfd') to log
 # freqtrade plot-dataframe --strategy ChanLun_SOL_5 --datadir user_data/data/binance -c ./user_data/ChanLun_SOL.json --timerange=20250309-
 
-# freqtrade trade -c ./user_data/ChanLun_SOL.json --strategy ChanLun_SOL_5 --strategy-path ./user_data/strategies
+# freqtrade trade -c ./user_data/Chan/config/ChanLun_SOL.json --strategy ChanLun_SOL_5 --strategy-path ./user_data/Chan/strategies
-# freqtrade backtesting -c ./user_data/ChanLun_SOL.json --strategy ChanLun_SOL_5 --strategy-path ./user_data/strategies --timerange=20250416-
+# freqtrade backtesting -c ./user_data/Chan/config/ChanLun_SOL.json --strategy ChanLun_SOL_5 --strategy-path ./user_data/Chan/strategies --timerange=20250416-
-# freqtrade download-data -c ./user_data/ChanLun_SOL.json -t 1m --pairs SOL/USDT:USDT --timerange=20250405-
+# freqtrade download-data -c ./user_data/Chan/config/ChanLun_SOL.json -t 1m --pairs SOL/USDT:USDT --timerange=20250405-
-# freqtrade hyperopt --hyperopt-loss SharpeHyperOptLossDaily --spaces roi stoploss --strategy ChanLun_SOL_5 --strategy-path ./user_data/strategies -c ./user_data/ChanLun_SOL.json -e 200 --timerange=20250201-20250401
+# freqtrade hyperopt --hyperopt-loss SharpeHyperOptLossDaily --spaces roi stoploss --strategy ChanLun_SOL_5 --strategy-path ./user	_data/Chan/strategies -c ./user_data/Chan/config/ChanLun_SOL.json -e 200 --timerange=20250201-20250401
 
 # sudo docker compose run --rm chan_btc backtesting -c ./user_data/ChanLun_SOL.json --strategy ChanLun_SOL --strategy-path ./user_data/strategies --timerange=20250101-
 # sudo docker compose run --rm chan_btc download-data -c ./user_data/ChanLun_SOL.json --pairs SOL/USDT:USDT -t 1m --timerange 20240101-
@@ -118,9 +118,9 @@ class ChanLun_SOL_5(IStrategy):
 		self.classifier.train_model(dataframe_1d, model_name="1d_model")
 		"""
 
-		"""
+		
 		if self.classifier.model is None:
-			self.classifier.train_model(dataframe_30, model_name="30m_model")
+			#self.classifier.train_model(dataframe_30, model_name="30m_model")
 			self.classifier.load_model(model_name="30m_model")
 			klc_list = self.chan.get_klc_list(dataframe_30)
 			top_avg = 0
@@ -129,12 +129,12 @@ class ChanLun_SOL_5(IStrategy):
 			bottom_count = 0
 			for index in range(int(len(klc_list) * 0.8), len(klc_list)):
 				klc = klc_list[index]
-				if self.classifier.predict(klc) > 0.01 and klc.fx == Chan_FX_TYPE.BOTTOM:
+				if self.classifier.predict(klc) > 0.45 and klc.fx == Chan_FX_TYPE.BOTTOM:
 					features = klc.get_feature_data()
 					print(klc.end_time, klc.fx, self.classifier.predict(klc), features['klc_macd'], features['klc_macd_hist'], features['klc_rsi'], features['klc_macd_signal'])
 					bottom_avg += self.classifier.predict(klc)
 					bottom_count += 1
-				if self.classifier.predict(klc) > 0.05 and klc.fx == Chan_FX_TYPE.TOP:
+				if self.classifier.predict(klc) > 0.44 and klc.fx == Chan_FX_TYPE.TOP:
 					features = klc.get_feature_data()
 					print(klc.end_time, klc.fx, self.classifier.predict(klc), features['klc_macd'], features['klc_macd_hist'], features['klc_rsi'], features['klc_macd_signal'])
 					top_avg += self.classifier.predict(klc)
@@ -143,7 +143,7 @@ class ChanLun_SOL_5(IStrategy):
 			top_avg /= top_count
 			print(bottom_avg, top_avg)
 		print("-------------------------------------------------------------------------------")
-		"""
+		
 		"""
 		self.print_xgb(dataframe, "1m_model")
 		self.print_xgb(dataframe_5, "5m_model")
@@ -163,11 +163,11 @@ class ChanLun_SOL_5(IStrategy):
 		#dataframe_60['state'] = self.chan.cal_klu_state(dataframe_60)
 		#dataframe_4h['state'] = self.chan.resample_klc_list(dataframe_4h)
 
-		self.chan.plot_dual(dataframe_30, dataframe_60)
+		#self.chan.plot_dual(dataframe_30, dataframe_60)
 		dataframe['rsi'] = ta.RSI(dataframe, timeperiod=14)
 		#self.print_macd_div_list(dataframe)
 		#self.print_resample_df(dataframe, 1, 50)
-		self.chan.get_bi_list(dataframe_30)
+		#self.chan.get_bi_list(dataframe_30)
 		if self.last_time + timedelta(minutes=1) < datetime.now():
 			#print(informative.iloc[-1])
 			#self.log_macd_div_list(dataframe)