# strategies/SimpleRSIStrategy_Fixed.py import talib.abstract as ta import pandas as pd import numpy as np from freqtrade.strategy import IStrategy from freqtrade.persistence import Trade from typing import Dict, List, Optional from functools import reduce import logging logger = logging.getLogger(__name__) class SimpleRSIStrategyFixed(IStrategy): """ 修复版简化 RSI 策略：结合 EMA 交叉 + RSI 超卖 + 上升趋势 修复内容： 1. 添加缺失的导入语句 2. 修复除零错误 3. 优化性能 4. 改进代码可读性 5. 添加数据验证 """ INTERFACE_VERSION = 3 # Can this strategy go short? can_short: bool = False # Minimal ROI designed for the strategy. # 基于x.py策略，使用更保守的ROI设置 # minimal_roi = { # "30": 0.3, # 20% 目标收益 # # "" # } # Optimal stoploss designed for the strategy. # 基于x.py策略的卖出条件：跌破MA60且跌幅超过10% stoploss = -0.05 # 超参数定义 minimal_roi = { "0": 100 } timeframe = '1h' # 策略参数 rsi_period_short = 6 rsi_period_long = 12 ema_period_short = 10 ema_period_long = 20 rsi_oversold_threshold = 20 rsi_oversold_tolerance = 1.05 ema_trend_threshold = 0.01 ema_breakout_threshold = 0.02 ema_separation_threshold = 0.02 def populate_indicators(self, dataframe: pd.DataFrame, metadata: Dict) -> pd.DataFrame: """ 添加技术指标到数据框 """ df = dataframe.copy() # 将date设置为索引用于resample df['date'] = df['date'].dt.tz_convert('Asia/Shanghai') df = df.set_index('date') df_resample = df.resample('24h', label='right').agg({ 'open': 'first', 'high': 'max', 'low': 'min', 'close': 'last', 'volume': 'sum' }) # RSI 指标 (6, 12) 及位移 df_resample['rsi_6'] = ta.RSI(df_resample['close'], timeperiod=6) df_resample['rsi_12'] = ta.RSI(df_resample['close'], timeperiod=12) for i in range(1, 6): df_resample[f'rsi_6_shift{i}'] = df_resample['rsi_6'].shift(i) df_resample[f'rsi_12_shift{i}'] = df_resample['rsi_12'].shift(i) # EMA 指标 (10, 20, 30) 及位移 df_resample['ema_10'] = ta.EMA(df_resample['close'], timeperiod=10) df_resample['ema_20'] = ta.EMA(df_resample['close'], timeperiod=20) df_resample['ema_30'] = ta.EMA(df_resample['close'], timeperiod=30) for i in range(1, 6): df_resample[f'ema_10_shift{i}'] = df_resample['ema_10'].shift(i) df_resample[f'ema_20_shift{i}'] = df_resample['ema_20'].shift(i) df_resample[f'ema_30_shift{i}'] = df_resample['ema_30'].shift(i) # 方便条件判断的差值与比率（带除零保护） df_resample['ema_10_20_diff'] = df_resample['ema_10'] - df_resample['ema_20'] df_resample['ema_10_20_ratio'] = np.where( df_resample['ema_20'] != 0, df_resample['ema_10_20_diff'] / df_resample['ema_20'], 0 ) for i in range(1, 6): df_resample[f'ema_10_20_diff_shift{i}'] = df_resample['ema_10_20_diff'].shift(i) df_resample[f'ema_10_20_ratio_shift{i}'] = np.where( df_resample[f'ema_20_shift{i}'] != 0, df_resample[f'ema_10_20_diff_shift{i}'] / df_resample[f'ema_20_shift{i}'], 0 ) df_resample = df_resample.drop(columns=['open', 'high', 'low', 'close', 'volume']) df_resample.index = df_resample.index.tz_convert('UTC') dataframe = dataframe.join(df_resample, on='date', how='left').ffill() dataframe.to_csv("rsi_eth2.csv", index=False, encoding='utf-8-sig') logger.debug(f"Indicators populated successfully for {metadata.get('pair', 'unknown')}") return dataframe def populate_entry_trend(self, dataframe: pd.DataFrame, metadata: Dict) -> pd.DataFrame: """ 生成买入信号 """ try: # 初始化信号列 dataframe.loc[:, 'enter_long'] = 0 # 数据验证 required_columns = [ 'ema_10', 'ema_20', 'ema_30', 'rsi_6', 'rsi_12', 'ema_10_shift1', 'ema_10_shift2', 'ema_10_shift3', 'ema_10_shift4', 'ema_10_shift5', 'ema_20_shift1', 'ema_20_shift2', 'ema_20_shift3', 'ema_20_shift4', 'ema_20_shift5', 'rsi_6_shift1', 'rsi_6_shift2', 'rsi_6_shift3', 'rsi_12_shift1', 'rsi_12_shift2', 'rsi_12_shift3', 'ema_10_20_ratio', 'ema_10_20_ratio_shift5', 'ema_10_20_diff', 'ema_10_20_diff_shift1', 'ema_10_20_diff_shift2', 'ema_10_20_diff_shift3', 'ema_10_20_diff_shift4', 'ema_10_20_diff_shift5', 'close' ] if not all(col in dataframe.columns for col in required_columns): logger.warning(f"Missing required columns: {required_columns}") return dataframe # 按给定逻辑实现买入条件 ema_10 = dataframe['ema_10'] ema_20 = dataframe['ema_20'] ema_10_s1 = dataframe['ema_10_shift1'] ema_10_s2 = dataframe['ema_10_shift2'] ema_10_s3 = dataframe['ema_10_shift3'] ema_10_s4 = dataframe['ema_10_shift4'] ema_10_s5 = dataframe['ema_10_shift5'] ema_20_s1 = dataframe['ema_20_shift1'] ema_20_s2 = dataframe['ema_20_shift2'] ema_20_s3 = dataframe['ema_20_shift3'] ema_20_s4 = dataframe['ema_20_shift4'] ema_20_s5 = dataframe['ema_20_shift5'] rsi_6 = dataframe['rsi_6'] rsi_12 = dataframe['rsi_12'] rsi_6_s1 = dataframe['rsi_6_shift1'] rsi_6_s2 = dataframe['rsi_6_shift2'] rsi_6_s3 = dataframe['rsi_6_shift3'] rsi_12_s1 = dataframe['rsi_12_shift1'] rsi_12_s2 = dataframe['rsi_12_shift2'] rsi_12_s3 = dataframe['rsi_12_shift3'] close = dataframe['close'] # 比率与安全除法 ratio_now = np.where(ema_20 != 0, (ema_10 - ema_20) / ema_20, 0) ratio_s5 = np.where(ema_20_s5 != 0, (ema_10_s5 - ema_20_s5) / ema_20_s5, 0) ratio_pos_and_small = (ratio_s5 != 0) & ((ratio_now / ratio_s5) > 0) & ((ratio_now / ratio_s5) < 0.2) cond_1 = ( ( (ema_10 > ema_20) & (ema_10_s1 > ema_20_s1) & (ema_10_s1 > ema_10_s2) & (ema_10 > ema_10_s1) & (ratio_now > 0.01) ) | ( (ema_10 > 1.02 * ema_10_s1) & (ema_10_s1 > ema_10_s2) & (ema_10 > ema_20) ) | ( ratio_pos_and_small & ((ema_20 - ema_10) < (ema_20_s1 - ema_10_s1)) & ((ema_20 - ema_10) < (ema_20_s2 - ema_10_s2)) & ((ema_20 - ema_10) < (ema_20_s3 - ema_10_s3)) & ((ema_20 - ema_10) < (ema_20_s4 - ema_10_s4)) & ((ema_20 - ema_10) < (ema_20_s5 - ema_10_s5)) & (ema_20_s5 > ema_10_s5) & (ema_10 < close) ) ) cond_rsi = ( (rsi_6 > 1.5 * rsi_6_s1) & (rsi_6 > 0.95 * rsi_12) & (rsi_6_s1 < 20 * 1.05) & (rsi_6_s2 < 20 * 1.05) & (rsi_6_s3 < 20 * 1.05) & (rsi_12_s1 < 25 * 1.05) & (rsi_12_s2 < 25 * 1.05) & (rsi_12_s3 < 25 * 1.05) ) buy_condition = cond_1 | cond_rsi dataframe.loc[buy_condition, 'enter_long'] = 1 # 记录信号统计 signal_count = buy_condition.sum() if signal_count > 0: logger.info(f"Generated {signal_count} buy signals for {metadata.get('pair', 'unknown')}") return dataframe except Exception as e: logger.error(f"Error in populate_entry_trend: {e}") return dataframe def populate_exit_trend(self, dataframe: pd.DataFrame, metadata: Dict) -> pd.DataFrame: """ 生成卖出信号 """ try: dataframe.loc[:, 'exit_long'] = 0 # 数据验证 required_columns = [ 'ema_10', 'ema_20', 'ema_10_shift1', 'ema_10_shift2', 'ema_10_shift3', 'ema_10_shift4', 'ema_10_shift5', 'ema_20_shift1', 'ema_20_shift2', 'ema_20_shift3', 'ema_20_shift4', 'ema_20_shift5', 'ema_10_20_ratio', 'ema_10_20_ratio_shift4', ] if not all(col in dataframe.columns for col in required_columns): logger.warning(f"Missing required columns: {required_columns}") return dataframe ema_10 = dataframe['ema_10'] ema_20 = dataframe['ema_20'] ema_10_s1 = dataframe['ema_10_shift1'] ema_10_s2 = dataframe['ema_10_shift2'] ema_10_s3 = dataframe['ema_10_shift3'] ema_10_s4 = dataframe['ema_10_shift4'] ema_10_s5 = dataframe['ema_10_shift5'] ema_20_s1 = dataframe['ema_20_shift1'] ema_20_s2 = dataframe['ema_20_shift2'] ema_20_s3 = dataframe['ema_20_shift3'] ema_20_s4 = dataframe['ema_20_shift4'] ema_20_s5 = dataframe['ema_20_shift5'] ratio_now = np.where(ema_20 != 0, (ema_10 - ema_20) / ema_20, 0) ratio_s4 = np.where(ema_20_s4 != 0, (ema_10_s4 - ema_20_s4) / ema_20_s4, 0) ratio_pos_and_lt03 = (ratio_s4 != 0) & ((ratio_now / ratio_s4) > 0) & ((ratio_now / ratio_s4) < 0.3) cond_block_1 = ( ( (ema_10 < ema_20) & (ema_10_s1 < ema_20_s1) & (ema_10_s2 < ema_20_s2) & (ema_10_s5 > ema_20_s5) ) | ( (ema_10 < 0.995 * ema_10_s1) & (ema_10_s1 < ema_10_s2) & (ema_10 < ema_20) ) ) cond_block_2 = ( (ratio_now < 0.02) & ratio_pos_and_lt03 & (ema_10_s4 > ema_20_s4) & (ema_10 < ema_10_s1) & (ema_10 < ema_10_s2) & (ema_10 < ema_10_s3) & (ema_10 < ema_10_s4) ) widening_now_vs_history = ( ((ema_20 - ema_10) > (ema_20_s1 - ema_10_s1)) & ((ema_20 - ema_10) > (ema_20_s2 - ema_10_s2)) & ((ema_20 - ema_10) > (ema_20_s3 - ema_10_s3)) & ((ema_20 - ema_10) > (ema_20_s4 - ema_10_s4)) & ((ema_20 - ema_10) > (ema_20_s5 - ema_10_s5)) ) sell_condition = ( ( cond_block_1 & (ema_10 < ema_10_s1) & (ema_10 < ema_10_s2) & (ema_10 < ema_10_s3) & (ema_10 < ema_10_s4) & (ema_10 < ema_10_s5) & widening_now_vs_history ) | ( cond_block_2 & ((ema_20 - ema_10) > (ema_20_s1 - ema_10_s1)) & ((ema_20 - ema_10) > (ema_20_s2 - ema_10_s2)) & ((ema_20 - ema_10) > (ema_20_s3 - ema_10_s3)) & ((ema_20 - ema_10) > (ema_20_s4 - ema_10_s4)) ) ) dataframe.loc[sell_condition, 'exit_long'] = 1 # 记录信号统计 signal_count = sell_condition.sum() if signal_count > 0: logger.info(f"Generated {signal_count} sell signals for {metadata.get('pair', 'unknown')}") return dataframe except Exception as e: logger.error(f"Error in populate_exit_trend: {e}") return dataframe # def custom_stoploss(self, pair: str, trade: Trade, current_time, current_rate, current_profit, **kwargs) -> float: # """ # 自定义止损逻辑 # """ # try: # # 动态止损：根据RSI调整止损点 # dataframe, _ = self.dp.get_analyzed_dataframe(pair, self.timeframe) # if len(dataframe) > 0: # last_candle = dataframe.iloc[-1] # rsi_short = last_candle.get('rsi_6', 50) # # RSI越高，止损越宽松 # if rsi_short > 70: # return -0.15 # 更宽松的止损 # elif rsi_short < 30: # return -0.05 # 更严格的止损 # else: # return self.stoploss # 默认止损 # else: # return self.stoploss # except Exception as e: # logger.error(f"Error in custom_stoploss: {e}") # return self.stoploss