# HighFrequency_Hybrid_Fixed.py from freqtrade.strategy import IStrategy, DecimalParameter, IntParameter from pandas import DataFrame import numpy as np import pandas as pd class HighFrequency_Hybrid_Fixed(IStrategy): """ 混合型策略（修复版） - 盘整时做市（高频小利） - 趋势时低频趋势跟随（回调入场） - 使用 ROI（entry-based）做 TP，使用 custom_stoploss 做出场保护（基于 entry_price + ATR） - 使用 confirm_trade_entry 过滤大 spread / 突发 ATR - dynamic stake（波动大时降低仓位） """ INTERFACE_VERSION = 3 timeframe = "1m" startup_candle_count = 30 # ROI (take-profit) 以 entry_price 为基准（避免用 mid） minimal_roi = { "0": 0.0015, # 0.15% 立即可做目标（可按需调） "30": 0.001, "60": 0 } # 基本 stoploss（作为最后后备） stoploss = -0.06 # 若 custom_stoploss 返回 None，将使用此值作为保险 # 下单类型（完整字段） order_types = { "entry": "limit", "exit": "limit", "stoploss": "limit", "stoploss_on_exchange": False, "force_entry": "limit", "force_exit": "limit", } # ============= 超参数（可 hyperopt） ============= stake_unit = DecimalParameter(0.006, 0.015, default=0.01, decimals=3, space="buy") # 基础 stake（1%） num_levels = IntParameter(1, 2, default=1, space="buy") price_offset_pct = DecimalParameter(0.08, 0.2, default=0.12, decimals=3, space="buy") # 0.12% 默认 take_profit_pct = DecimalParameter(0.12, 0.30, default=0.18, decimals=3, space="sell") # ROI 主目标 0.18% atr_threshold_pct = DecimalParameter(0.10, 0.35, default=0.20, decimals=3, space="buy") # ATR阈值（% of mid） atr_multiplier = DecimalParameter(1.0, 3.0, default=1.6, decimals=2, space="sell") # custom stoploss基于ATR倍数 max_spread = DecimalParameter(0.003, 0.02, default=0.006, decimals=4, space="buy") # 最大允许spread（比例） ema_fast = IntParameter(5, 8, default=5, space="buy") ema_slow = IntParameter(12, 20, default=12, space="buy") trend_ema_period = IntParameter(34, 60, default=34, space="buy") trend_strength_threshold = DecimalParameter(0.005, 0.015, default=0.008, decimals=4, space="buy") # safety caps max_total_risk = 0.08 # 总仓位上限 8% # ================================================ def populate_indicators(self, df: DataFrame, metadata: dict) -> DataFrame: df = df.copy() # mid price df["mid"] = (df["high"] + df["low"]) / 2.0 # TR & ATR (短周期) df["tr"] = np.maximum(df["high"] - df["low"], np.maximum(abs(df["high"] - df["close"].shift(1)), abs(df["low"] - df["close"].shift(1)))) # 用 7 或 9 更短的 ATR 对 1m 更敏感（不使用过长会错判） df["atr"] = df["tr"].rolling(7).mean().bfill() # EMAs df["ema_fast"] = df["close"].ewm(span=self.ema_fast.value, adjust=False).mean() df["ema_slow"] = df["close"].ewm(span=self.ema_slow.value, adjust=False).mean() df["ema_trend"] = df["close"].ewm(span=self.trend_ema_period.value, adjust=False).mean() # trend strength df["trend_strength"] = abs(df["close"] / df["ema_trend"] - 1) # spread proxy - if bid/ask available use it, else proxy with high-low but keep threshold loose if "bid" in df.columns and "ask" in df.columns: mid_askbid = (df["ask"] + df["bid"]) / 2.0 df["spread"] = (df["ask"] - df["bid"]) / mid_askbid else: df["spread"] = (df["high"] - df["low"]) / df["mid"] # proxy (candle range) # in_range (盘整判断)：ATR 相对较小并且趋势强度低 df["in_range"] = (df["atr"] < (df["mid"] * (self.atr_threshold_pct.value / 100.0))) & \ (df["trend_strength"] < self.trend_strength_threshold.value) return df def is_trending(self, row) -> bool: # 快慢 EMA 明显分离并且趋势强度较高 return ((row["ema_fast"] > row["ema_slow"] and (row["ema_fast"] / row["ema_slow"] - 1) > 0.005) or (row["ema_fast"] < row["ema_slow"] and (row["ema_slow"] / row["ema_fast"] - 1) > 0.005)) and \ row["trend_strength"] > self.trend_strength_threshold.value def populate_entry_trend(self, df: DataFrame, metadata: dict) -> DataFrame: """ 入场逻辑： - 盘整模式：当 close <= mid*(1-offset) 且 in_range 且 spread 小，标记 buy（limit） - 趋势模式：当正在趋势中且发生小幅回调到 ema_fast * (1 - pullback_pct) 时买入（低频） """ df = df.copy() df["buy"] = 0 levels = int(self.num_levels.value) base_offset = float(self.price_offset_pct.value) / 100.0 pullback_pct = 0.005 # 趋势回调阈值 0.5% cond_common = (df["spread"] <= float(self.max_spread.value)) for i in range(1, levels + 1): offset = base_offset * i buy_price = df["mid"] * (1 - offset) # 盘整模式买点 df.loc[(df["in_range"]) & (df["close"] <= buy_price) & cond_common, "buy"] = 1 # 趋势模式回调入场（低频） df.loc[(df.apply(self.is_trending, axis=1)) & (df["close"] < df["ema_fast"] * (1 - pullback_pct)) & cond_common, "buy"] = 1 return df def populate_exit_trend(self, df: DataFrame, metadata: dict) -> DataFrame: """ exit 使用 ROI（entry based）与 custom_stoploss；populate_exit_trend 仍提供保底退出（例如极端情况） 为兼容回测/可视化，保留 conservative mid-based exits as fallback but main logic uses ROI/custom_stoploss. """ df = df.copy() df["sell"] = 0 # fallback conservative exits to avoid infinite holds (not the main exit) tp_fallback = float(self.take_profit_pct.value) / 100.0 sl_fallback = 0.05 # 5%极端回退保护（仅作最后保障） df.loc[df["close"] >= df["mid"] * (1 + tp_fallback), "sell"] = 1 df.loc[df["close"] <= df["mid"] * (1 - sl_fallback), "sell"] = 1 return df def custom_stoploss(self, pair: str, trade, current_time, current_rate, current_profit, **kwargs): """ 动态止损基于 entry_rate 与实时 ATR： - stop_pct = max(min_stop, atr_pct * atr_multiplier) - 返回 stoploss as negative fraction (e.g. -0.003) 如果返回 None 则使用默认 self.stoploss。 """ try: # 获取最新df df = self.dp.get_pair_dataframe(pair=pair, timeframe=self.timeframe, limit=5) last = df.iloc[-1] entry_price = trade.open_rate if hasattr(trade, "open_rate") else trade.entry_price atr = float(last["atr"]) atr_pct = atr / entry_price if entry_price and entry_price > 0 else 0.001 mult = float(self.atr_multiplier.value) # dynamic stoploss: 根据 atr_pct 放大；至少 min_stop min_stop = 0.003 # 最小止损 0.3% computed = max(min_stop, atr_pct * mult) stoploss_val = -computed # 保证不超过全局 stoploss cap # self.stoploss 是 -0.06（-6%），不放开更大 if stoploss_val < self.stoploss: # 不小于全局 stoploss（即不要更严） return self.stoploss return stoploss_val except Exception: return None def custom_entry(self, pair: str, current_time, current_rate, current_profit, **kwargs): """ 根据当前波动动态调整 stake：波动越大，stake越小。 返回 fraction(0..1) """ try: df = self.dp.get_pair_dataframe(pair=pair, timeframe=self.timeframe, limit=5) last = df.iloc[-1] atr = float(last["atr"]) atr_pct = atr / last["close"] if last["close"] and last["close"] > 0 else 0.001 base = float(self.stake_unit.value) # simple scale: if atr_pct > 0.5% -> reduce stake by half; else keep base if atr_pct > 0.005: stake = max(0.002, base * 0.5) # 最小 0.2% elif atr_pct > 0.003: stake = max(0.003, base * 0.7) else: stake = base # ensure not to exceed max_total_risk minus existing position position = self.dp.get_position(pair, lookback=1) current_amount = position.amount if position else 0 # current_amount is in base asset units — we can't easily translate to fraction here reliably, # so we just return stake fraction as designed return float(stake) except Exception: return float(self.stake_unit.value) def confirm_trade_entry(self, pair: str, order_type: str, amount: float, rate: float, time_in_force: str, **kwargs) -> bool: """ 下单前再次检查最新盘口/波动，避免在 spread/ATR 突增时下单 """ try: df = self.dp.get_pair_dataframe(pair=pair, timeframe=self.timeframe, limit=2) last = df.iloc[-1] # 若实际spread列存在，优先使用 spread = float(last.get("spread", 0.0)) atr = float(last.get("atr", 0.0)) mid = float(last.get("mid", last["close"])) # 如果 spread 超过阈值，不下单 if spread > float(self.max_spread.value): return False # 如果 ATR 突增（> mid * 1%），不下单 if atr > mid * 0.01: return False return True except Exception: return True