""" 多时间框架趋势跟随策略 使用多个时间框架分析确认趋势方向的策略 """ from freqtrade.strategy import IStrategy, IntParameter, DecimalParameter from pandas import DataFrame import talib.abstract as ta from technical import qtpylib import numpy as np import logging logger = logging.getLogger(__name__) class MultiTimeframeTrendStrategy(IStrategy): """ 多时间框架趋势跟随策略 策略逻辑: 1. 使用日线确定主趋势方向 2. 使用4小时线确定中期趋势 3. 使用1小时线寻找入场时机 4. 所有时间框架趋势一致时才入场 5. 使用层级止损管理风险 """ INTERFACE_VERSION = 3 # 基础配置 minimal_roi = { "0": 0.25, # 25%收益立即止盈 "60": 0.15, # 1小时后15%收益 "120": 0.10, # 2小时后10%收益 "240": 0.05, # 4小时后5%收益 "480": 0.02 # 8小时后2%收益 } stoploss = -0.08 # 8%止损 timeframe = '1h' # 主时间框架1小时 # 策略控制 can_short = False startup_candle_count = 100 process_only_new_candles = True use_exit_signal = True use_custom_stoploss = True # 信息时间框架 informative_pairs = [] # 可优化参数 # 主趋势参数（日线） daily_ema_fast = IntParameter(8, 16, default=12, space="buy") daily_ema_slow = IntParameter(24, 40, default=30, space="buy") daily_rsi_period = IntParameter(10, 20, default=14, space="buy") # 中期趋势参数（4小时） h4_ema_fast = IntParameter(12, 24, default=18, space="buy") h4_ema_slow = IntParameter(36, 60, default=48, space="buy") h4_macd_fast = IntParameter(8, 16, default=12, space="buy") h4_macd_slow = IntParameter(21, 35, default=26, space="buy") # 短期入场参数（1小时） h1_ema_fast = IntParameter(8, 20, default=14, space="buy") h1_ema_slow = IntParameter(24, 40, default=32, space="buy") h1_rsi_period = IntParameter(10, 20, default=14, space="buy") h1_rsi_threshold = IntParameter(45, 60, default=50, space="buy") # ADX趋势强度参数（各时间框架） adx_daily_threshold = IntParameter(20, 35, default=25, space="buy") adx_h4_threshold = IntParameter(20, 35, default=25, space="buy") adx_h1_threshold = IntParameter(20, 35, default=25, space="buy") # 成交量参数 volume_factor = DecimalParameter(1.2, 2.5, default=1.8, space="buy") # 止损参数 atr_period = IntParameter(10, 20, default=14, space="sell") atr_multiplier = DecimalParameter(2.0, 4.0, default=3.0, space="sell") def informative_pairs(self): """ 定义信息时间框架 """ pairs = self.dp.current_whitelist() informative_pairs = [] for pair in pairs: # 添加4小时和日线时间框架 informative_pairs.append((pair, '4h')) informative_pairs.append((pair, '1d')) return informative_pairs def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """ 计算主时间框架（1小时）指标 """ # 1小时时间框架指标 # EMA dataframe['ema_fast_h1'] = ta.EMA(dataframe, timeperiod=self.h1_ema_fast.value) dataframe['ema_slow_h1'] = ta.EMA(dataframe, timeperiod=self.h1_ema_slow.value) # RSI dataframe['rsi_h1'] = ta.RSI(dataframe, timeperiod=self.h1_rsi_period.value) # MACD macd_h1 = ta.MACD(dataframe) dataframe['macd_h1'] = macd_h1['macd'] dataframe['macd_signal_h1'] = macd_h1['macdsignal'] dataframe['macd_hist_h1'] = macd_h1['macdhist'] # ADX dataframe['adx_h1'] = ta.ADX(dataframe, timeperiod=14) dataframe['di_plus_h1'] = ta.PLUS_DI(dataframe, timeperiod=14) dataframe['di_minus_h1'] = ta.MINUS_DI(dataframe, timeperiod=14) # ATR dataframe['atr_h1'] = ta.ATR(dataframe, timeperiod=self.atr_period.value) # 成交量 dataframe['volume_sma_h1'] = dataframe['volume'].rolling(window=20).mean() dataframe['volume_ratio_h1'] = dataframe['volume'] / dataframe['volume_sma_h1'] # 1小时趋势判断 dataframe['trend_h1'] = dataframe['ema_fast_h1'] > dataframe['ema_slow_h1'] dataframe['price_above_ema_fast_h1'] = dataframe['close'] > dataframe['ema_fast_h1'] # 获取4小时和日线数据 if self.dp: # 4小时数据 dataframe_4h = self.dp.get_pair_dataframe( pair=metadata['pair'], timeframe='4h' ) dataframe = self.populate_indicators_4h(dataframe, dataframe_4h) # 日线数据 dataframe_1d = self.dp.get_pair_dataframe( pair=metadata['pair'], timeframe='1d' ) dataframe = self.populate_indicators_1d(dataframe, dataframe_1d) return dataframe def populate_indicators_4h(self, dataframe: DataFrame, dataframe_4h: DataFrame) -> DataFrame: """ 计算4小时时间框架指标并合并到1小时数据 """ # 4小时EMA dataframe_4h['ema_fast_4h'] = ta.EMA(dataframe_4h, timeperiod=self.h4_ema_fast.value) dataframe_4h['ema_slow_4h'] = ta.EMA(dataframe_4h, timeperiod=self.h4_ema_slow.value) # 4小时MACD macd_4h = ta.MACD(dataframe_4h, fastperiod=self.h4_macd_fast.value, slowperiod=self.h4_macd_slow.value) dataframe_4h['macd_4h'] = macd_4h['macd'] dataframe_4h['macd_signal_4h'] = macd_4h['macdsignal'] dataframe_4h['macd_hist_4h'] = macd_4h['macdhist'] # 4小时ADX dataframe_4h['adx_4h'] = ta.ADX(dataframe_4h, timeperiod=14) dataframe_4h['di_plus_4h'] = ta.PLUS_DI(dataframe_4h, timeperiod=14) dataframe_4h['di_minus_4h'] = ta.MINUS_DI(dataframe_4h, timeperiod=14) # 4小时RSI dataframe_4h['rsi_4h'] = ta.RSI(dataframe_4h, timeperiod=14) # 4小时趋势判断 dataframe_4h['trend_4h'] = dataframe_4h['ema_fast_4h'] > dataframe_4h['ema_slow_4h'] dataframe_4h['macd_bullish_4h'] = dataframe_4h['macd_4h'] > dataframe_4h['macd_signal_4h'] # 合并4小时数据到1小时数据 dataframe = dataframe.merge( dataframe_4h[['date', 'ema_fast_4h', 'ema_slow_4h', 'trend_4h', 'macd_4h', 'macd_signal_4h', 'macd_hist_4h', 'macd_bullish_4h', 'adx_4h', 'di_plus_4h', 'di_minus_4h', 'rsi_4h']].set_index('date'), left_index=True, right_index=True, how='left' ) # 前向填充4小时数据 columns_4h = ['ema_fast_4h', 'ema_slow_4h', 'trend_4h', 'macd_4h', 'macd_signal_4h', 'macd_hist_4h', 'macd_bullish_4h', 'adx_4h', 'di_plus_4h', 'di_minus_4h', 'rsi_4h'] dataframe[columns_4h] = dataframe[columns_4h].fillna(method='ffill') return dataframe def populate_indicators_1d(self, dataframe: DataFrame, dataframe_1d: DataFrame) -> DataFrame: """ 计算日线时间框架指标并合并到1小时数据 """ # 日线EMA dataframe_1d['ema_fast_1d'] = ta.EMA(dataframe_1d, timeperiod=self.daily_ema_fast.value) dataframe_1d['ema_slow_1d'] = ta.EMA(dataframe_1d, timeperiod=self.daily_ema_slow.value) # 日线RSI dataframe_1d['rsi_1d'] = ta.RSI(dataframe_1d, timeperiod=self.daily_rsi_period.value) # 日线MACD macd_1d = ta.MACD(dataframe_1d) dataframe_1d['macd_1d'] = macd_1d['macd'] dataframe_1d['macd_signal_1d'] = macd_1d['macdsignal'] dataframe_1d['macd_hist_1d'] = macd_1d['macdhist'] # 日线ADX dataframe_1d['adx_1d'] = ta.ADX(dataframe_1d, timeperiod=14) dataframe_1d['di_plus_1d'] = ta.PLUS_DI(dataframe_1d, timeperiod=14) dataframe_1d['di_minus_1d'] = ta.MINUS_DI(dataframe_1d, timeperiod=14) # 日线趋势判断 dataframe_1d['trend_1d'] = dataframe_1d['ema_fast_1d'] > dataframe_1d['ema_slow_1d'] dataframe_1d['macd_bullish_1d'] = dataframe_1d['macd_1d'] > dataframe_1d['macd_signal_1d'] dataframe_1d['rsi_bullish_1d'] = dataframe_1d['rsi_1d'] > 50 # 合并日线数据到1小时数据 dataframe = dataframe.merge( dataframe_1d[['date', 'ema_fast_1d', 'ema_slow_1d', 'trend_1d', 'macd_1d', 'macd_signal_1d', 'macd_hist_1d', 'macd_bullish_1d', 'rsi_1d', 'rsi_bullish_1d', 'adx_1d', 'di_plus_1d', 'di_minus_1d']].set_index('date'), left_index=True, right_index=True, how='left' ) # 前向填充日线数据 columns_1d = ['ema_fast_1d', 'ema_slow_1d', 'trend_1d', 'macd_1d', 'macd_signal_1d', 'macd_hist_1d', 'macd_bullish_1d', 'rsi_1d', 'rsi_bullish_1d', 'adx_1d', 'di_plus_1d', 'di_minus_1d'] dataframe[columns_1d] = dataframe[columns_1d].fillna(method='ffill') return dataframe def populate_entry_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """ 定义买入条件 - 多时间框架趋势一致 """ conditions = [ # 日线趋势确认（主趋势） dataframe['trend_1d'].fillna(False), dataframe['macd_bullish_1d'].fillna(False), dataframe['rsi_bullish_1d'].fillna(False), dataframe['adx_1d'].fillna(0) > self.adx_daily_threshold.value, dataframe['di_plus_1d'].fillna(0) > dataframe['di_minus_1d'].fillna(0), # 4小时趋势确认（中期趋势） dataframe['trend_4h'].fillna(False), dataframe['macd_bullish_4h'].fillna(False), dataframe['rsi_4h'].fillna(50) > 50, dataframe['adx_4h'].fillna(0) > self.adx_h4_threshold.value, dataframe['di_plus_4h'].fillna(0) > dataframe['di_minus_4h'].fillna(0), # 1小时入场信号（短期时机） dataframe['trend_h1'], dataframe['price_above_ema_fast_h1'], dataframe['rsi_h1'] > self.h1_rsi_threshold.value, dataframe['rsi_h1'] < 80, # 避免极度超买 # 1小时MACD确认 dataframe['macd_h1'] > dataframe['macd_signal_h1'], dataframe['macd_hist_h1'] > 0, dataframe['macd_hist_h1'] > dataframe['macd_hist_h1'].shift(1), # 1小时ADX确认 dataframe['adx_h1'] > self.adx_h1_threshold.value, dataframe['di_plus_h1'] > dataframe['di_minus_h1'], # 成交量确认 dataframe['volume_ratio_h1'] > self.volume_factor.value, # 价格相对位置 dataframe['close'] > dataframe['ema_slow_h1'], # 多时间框架EMA排列确认 dataframe['ema_fast_1d'].fillna(0) > dataframe['ema_slow_1d'].fillna(0), # 日线EMA向上 dataframe['ema_fast_4h'].fillna(0) > dataframe['ema_slow_4h'].fillna(0), # 4小时EMA向上 # 当前价格高于所有关键EMA dataframe['close'] > dataframe['ema_fast_1d'].fillna(dataframe['close']), dataframe['close'] > dataframe['ema_fast_4h'].fillna(dataframe['close']), ] # 组合条件 dataframe.loc[ ( conditions[0] & # 日线趋势 conditions[1] & # 日线MACD conditions[2] & # 日线RSI conditions[3] & # 日线ADX conditions[4] & # 日线DI conditions[5] & # 4小时趋势 conditions[6] & # 4小时MACD conditions[7] & # 4小时RSI conditions[8] & # 4小时ADX conditions[9] & # 4小时DI conditions[10] & # 1小时趋势 conditions[11] & # 1小时价格位置 conditions[12] & # 1小时RSI下限 conditions[13] & # 1小时RSI上限 conditions[14] & # 1小时MACD conditions[15] & # 1小时MACD柱状图 conditions[16] & # 1小时MACD增强 conditions[17] & # 1小时ADX conditions[18] & # 1小时DI conditions[19] & # 成交量 conditions[20] & # 价格位置 conditions[21] & # 日线EMA排列 conditions[22] & # 4小时EMA排列 conditions[23] & # 价格vs日线EMA conditions[24] # 价格vs4小时EMA ), 'enter_long' ] = 1 return dataframe def populate_exit_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """ 定义卖出条件 - 任一时间框架趋势转向 """ conditions = [ # 日线趋势转向 ~dataframe['trend_1d'].fillna(True), ~dataframe['macd_bullish_1d'].fillna(True), dataframe['di_minus_1d'].fillna(0) > dataframe['di_plus_1d'].fillna(0), # 4小时趋势转向 ~dataframe['trend_4h'].fillna(True), ~dataframe['macd_bullish_4h'].fillna(True), dataframe['rsi_4h'].fillna(50) < 45, # 1小时趋势转向 ~dataframe['trend_h1'], dataframe['close'] < dataframe['ema_fast_h1'], dataframe['rsi_h1'] < 40, # 1小时MACD转向 dataframe['macd_h1'] < dataframe['macd_signal_h1'], dataframe['macd_hist_h1'] < 0, # ADX弱化 dataframe['adx_h1'] < 20, ] dataframe.loc[ ( conditions[0] | # 日线趋势转向 conditions[1] | # 日线MACD转向 conditions[2] | # 日线DI转向 conditions[3] | # 4小时趋势转向 conditions[4] | # 4小时MACD转向 conditions[5] | # 4小时RSI转弱 conditions[6] | # 1小时趋势转向 conditions[7] | # 1小时价格跌破EMA conditions[8] | # 1小时RSI转弱 conditions[9] | # 1小时MACD转向 conditions[10] | # 1小时MACD柱状图 conditions[11] # ADX弱化 ), 'exit_long' ] = 1 return dataframe def custom_stoploss(self, pair: str, trade, current_time, current_rate: float, current_profit: float, **kwargs) -> float: """ 多时间框架动态止损 """ dataframe, _ = self.dp.get_analyzed_dataframe(pair, self.timeframe) last_candle = dataframe.iloc[-1].squeeze() # ATR止损 atr_distance = last_candle['atr_h1'] * self.atr_multiplier.value atr_stop_distance = atr_distance / current_rate # 多层级止损 if current_profit > 0.12: # 盈利超过12% # 跟踪4小时EMA ema_4h_distance = abs(current_rate - last_candle.get('ema_fast_4h', current_rate)) / current_rate return max(-ema_4h_distance * 1.2, -0.02) elif current_profit > 0.08: # 盈利超过8% # 跟踪1小时快EMA ema_h1_distance = abs(current_rate - last_candle['ema_fast_h1']) / current_rate return max(-ema_h1_distance * 1.5, -0.025) elif current_profit > 0.04: # 盈利超过4% return max(-atr_stop_distance * 0.8, -0.03) elif current_profit > 0.02: # 盈利超过2% return max(-atr_stop_distance * 0.9, -0.04) else: # 初始宽松止损，多时间框架策略需要更多空间 return max(-atr_stop_distance * 1.3, self.stoploss) def custom_exit(self, pair: str, trade, current_time, current_rate: float, current_profit: float, **kwargs) -> str: """ 自定义退出逻辑 """ dataframe, _ = self.dp.get_analyzed_dataframe(pair, self.timeframe) last_candle = dataframe.iloc[-1].squeeze() # 主趋势（日线）转向 if (not last_candle.get('trend_1d', True) and current_profit > 0.03): return "daily_trend_reversal" # 中期趋势（4小时）转向 if (not last_candle.get('trend_4h', True) and current_profit > 0.02): return "h4_trend_reversal" # 多时间框架RSI全部转弱 if (last_candle.get('rsi_1d', 50) < 45 and last_candle.get('rsi_4h', 50) < 45 and last_candle['rsi_h1'] < 40 and current_profit > 0.02): return "multi_timeframe_rsi_weak" # 时间止损（多时间框架策略允许更长持仓） trade_duration = (current_time - trade.open_date_utc).total_seconds() / 3600 if trade_duration > 120 and current_profit < 0.01: # 120小时（5天）无显著盈利 return "time_exit" return None def confirm_trade_entry(self, pair: str, order_type: str, amount: float, rate: float, time_in_force: str, current_time, entry_tag: str, side: str, **kwargs) -> bool: """ 交易确认 """ dataframe, _ = self.dp.get_analyzed_dataframe(pair, self.timeframe) if len(dataframe) < 1: return False last_candle = dataframe.iloc[-1].squeeze() # 确保所有时间框架趋势一致 daily_trend = last_candle.get('trend_1d', False) h4_trend = last_candle.get('trend_4h', False) h1_trend = last_candle['trend_h1'] if not (daily_trend and h4_trend and h1_trend): return False # 确保所有时间框架ADX足够强 daily_adx = last_candle.get('adx_1d', 0) h4_adx = last_candle.get('adx_4h', 0) h1_adx = last_candle['adx_h1'] if not (daily_adx > 20 and h4_adx > 20 and h1_adx > 20): return False # 确保价格位置合理 if last_candle['rsi_h1'] > 85: # 避免极度超买 return False # 确保MACD信号明确 if last_candle['macd_hist_h1'] <= 0: return False return True