15 related strategies (⧉ identical code, ≈ similar name)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 | """ AgentSignalStrategy - 读取 Agent 团队生成的信号,由 Freqtrade 执行交易 工作模式: 1. Live/Dry-run: 读取 signal.json 中的 Agent 信号 → 执行交易 2. 回测: 使用简单技术指标规则生成信号 (模拟 Agent 决策) 核心回调: - populate_indicators: 计算基础指标 - populate_entry_trend: 入场信号 (Agent 信号 或 回测规则) - populate_exit_trend: 出场信号 - custom_stoploss: 动态止损 + 移动止盈 - custom_exit: 最终止盈级别全仓平仓 - adjust_trade_position: 分批止盈减仓 - custom_stake_amount: Agent 仓位控制 - leverage: 杠杆控制 """ import json import logging from datetime import datetime, timezone from pathlib import Path import pandas_ta # noqa: F401 from freqtrade.strategy import IStrategy, DecimalParameter from pandas import DataFrame logger = logging.getLogger(__name__) SIGNAL_FILE = Path("data/output/signals/signal.json") def _ensure_utc(dt: datetime) -> datetime: """确保 datetime 带 UTC 时区信息""" if dt.tzinfo is None: return dt.replace(tzinfo=timezone.utc) return dt class AgentSignalStrategy(IStrategy): """读取 Agent 团队生成的信号,由 Freqtrade 执行交易""" INTERFACE_VERSION = 3 can_short = True # ROI (回测用,实盘由 Agent 信号控制) minimal_roi = {"0": 0.20, "120": 0.10, "240": 0.05, "480": 0.02} stoploss = -0.05 trailing_stop = False timeframe = "5m" process_only_new_candles = True startup_candle_count = 100 # 需要 100 根 K 线热身 # 移动止盈参数 (可 Hyperopt 优化) ts_profit_1 = DecimalParameter(0.03, 0.08, default=0.05, space="sell") ts_profit_2 = DecimalParameter(0.08, 0.15, default=0.10, space="sell") ts_profit_3 = DecimalParameter(0.15, 0.30, default=0.20, space="sell") def __init__(self, config: dict | None = None) -> None: if config is not None: super().__init__(config) # 信号缓存(基于文件 mtime,避免每根 K 线重复读磁盘) self._signal_cache: list[dict] = [] self._signal_mtime: float = 0.0 def _read_signals(self) -> list[dict]: """读取并过滤有效信号(带 mtime 缓存)""" if not SIGNAL_FILE.exists(): return [] mtime = SIGNAL_FILE.stat().st_mtime if mtime == self._signal_mtime and self._signal_cache is not None: return self._signal_cache try: data = json.loads(SIGNAL_FILE.read_text()) now = datetime.now(timezone.utc) self._signal_cache = [ s for s in data.get("signals", []) if _ensure_utc(datetime.fromisoformat(s["expires_at"])) > now ] self._signal_mtime = mtime return self._signal_cache except (json.JSONDecodeError, KeyError, ValueError) as e: logger.warning(f"读取信号失败: {e}") return [] def _get_signal_for_pair(self, pair: str) -> dict | None: """获取指定交易对的信号""" symbol = pair.replace("/", "").replace(":USDT", "") for s in self._read_signals(): if s["symbol"] == symbol: return s return None def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """计算指标 (回测和实盘通用)""" # EMA dataframe["ema_7"] = dataframe.ta.ema(length=7) dataframe["ema_25"] = dataframe.ta.ema(length=25) dataframe["ema_99"] = dataframe.ta.ema(length=99) # RSI dataframe["rsi"] = dataframe.ta.rsi(length=14) # MACD macd = dataframe.ta.macd(fast=12, slow=26, signal=9) if macd is not None: dataframe["macd"] = macd.iloc[:, 0] dataframe["macd_signal"] = macd.iloc[:, 1] dataframe["macd_hist"] = macd.iloc[:, 2] # Bollinger Bands bbands = dataframe.ta.bbands(length=20, std=2) if bbands is not None: dataframe["bb_lower"] = bbands.iloc[:, 0] dataframe["bb_mid"] = bbands.iloc[:, 1] dataframe["bb_upper"] = bbands.iloc[:, 2] # ATR dataframe["atr"] = dataframe.ta.atr(length=14) # Volume EMA dataframe["volume_ema"] = dataframe.ta.ema(close=dataframe["volume"], length=20) return dataframe def populate_entry_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """入场信号""" dataframe.loc[:, "enter_long"] = 0 dataframe.loc[:, "enter_short"] = 0 # --- 模式 1: 读取 Agent 信号 (实盘/模拟盘) --- signal = self._get_signal_for_pair(metadata["pair"]) if signal is not None: # P8: 入场价格范围检查 entry_range = signal.get("entry_price_range") if entry_range and len(entry_range) == 2: current_price = dataframe["close"].iloc[-1] low, high = entry_range tolerance = (high - low) * 0.15 # 15% 容差 if not (low - tolerance <= current_price <= high + tolerance): logger.info( "[Agent] %s 价格 %.2f 不在范围 [%.2f, %.2f]±15%%, 跳过", metadata["pair"], current_price, low, high, ) return dataframe if signal["action"] == "long": dataframe.loc[dataframe.index[-1], "enter_long"] = 1 logger.info( f"[Agent] {metadata['pair']} LONG " f"置信度={signal.get('confidence')} " f"杠杆={signal.get('leverage')}" ) elif signal["action"] == "short": dataframe.loc[dataframe.index[-1], "enter_short"] = 1 logger.info( f"[Agent] {metadata['pair']} SHORT " f"置信度={signal.get('confidence')} " f"杠杆={signal.get('leverage')}" ) return dataframe # --- 模式 2: 规则化信号 (回测用) --- # 做多条件: EMA 多头排列 + RSI 超卖回升 + MACD 金叉 dataframe.loc[ ( (dataframe["ema_7"] > dataframe["ema_25"]) & (dataframe["ema_25"] > dataframe["ema_99"]) & (dataframe["rsi"] > 30) & (dataframe["rsi"] < 70) & (dataframe["macd_hist"] > 0) & (dataframe["macd_hist"].shift(1) <= 0) & (dataframe["volume"] > dataframe["volume_ema"]) ), "enter_long", ] = 1 # 做空条件: EMA 空头排列 + RSI 超买回落 + MACD 死叉 dataframe.loc[ ( (dataframe["ema_7"] < dataframe["ema_25"]) & (dataframe["ema_25"] < dataframe["ema_99"]) & (dataframe["rsi"] < 70) & (dataframe["rsi"] > 30) & (dataframe["macd_hist"] < 0) & (dataframe["macd_hist"].shift(1) >= 0) & (dataframe["volume"] > dataframe["volume_ema"]) ), "enter_short", ] = 1 return dataframe def populate_exit_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame: """出场信号""" dataframe.loc[:, "exit_long"] = 0 dataframe.loc[:, "exit_short"] = 0 # Agent 平仓信号 signal = self._get_signal_for_pair(metadata["pair"]) if signal is not None: if signal["action"] == "close_long": dataframe.loc[dataframe.index[-1], "exit_long"] = 1 elif signal["action"] == "close_short": dataframe.loc[dataframe.index[-1], "exit_short"] = 1 return dataframe # 回测规则出场 # 多单出场: RSI 超买 或 EMA 死叉 dataframe.loc[ (dataframe["rsi"] > 75) | (dataframe["ema_7"] < dataframe["ema_25"]), "exit_long", ] = 1 # 空单出场: RSI 超卖 或 EMA 金叉 dataframe.loc[ (dataframe["rsi"] < 25) | (dataframe["ema_7"] > dataframe["ema_25"]), "exit_short", ] = 1 return dataframe def custom_stoploss( self, pair: str, trade, current_time: datetime, current_rate: float, current_profit: float, after_fill: bool, **kwargs, ) -> float: """动态止损 + 移动止盈 优先级: 1. Agent trailing_stop_pct (AI 推荐的尾随比例,盈利 > 2% 后激活) 2. Agent 固定止损价 3. 三档移动止盈 4. 默认止损 """ signal = self._get_signal_for_pair(pair) # 优先级 1: Agent 尾随止盈(盈利 > 2% 后激活) if signal and signal.get("trailing_stop_pct") and current_profit > 0.02: trail_pct = signal["trailing_stop_pct"] / 100 return -(current_profit - trail_pct) # 优先级 1.5: MFE 自适应尾随 (ATR-based) # 浮盈 >= 2×ATR 后保本,之后每 1×ATR 收紧止损 if signal and current_profit > 0 and trade: atr_val = None atr_pct = signal.get("atr_pct") if atr_pct: atr_val = atr_pct / 100 elif hasattr(trade, "pair"): # 从 dataframe 中获取 ATR (如有) try: dp = self.dp if dp: df = dp.get_pair_dataframe(trade.pair, self.timeframe) if df is not None and "atr" in df.columns and len(df) > 0: last_atr = df["atr"].iloc[-1] last_close = df["close"].iloc[-1] if last_close > 0 and last_atr > 0: atr_val = last_atr / last_close except Exception: pass if atr_val and atr_val > 0: trigger = atr_val * 2 if current_profit >= trigger: trail_steps = int((current_profit - trigger) / atr_val) # 保本 + 阶梯收紧 (每步收紧 0.5×ATR) tightened = trail_steps * atr_val * 0.5 new_sl = -(current_profit - 0.001 - tightened) return min(new_sl, self.stoploss) # 优先级 2: Agent 固定止损价 if signal and signal.get("stop_loss") and trade: sl_price = signal["stop_loss"] if trade.is_short: sl_pct = (sl_price - current_rate) / current_rate else: sl_pct = (current_rate - sl_price) / current_rate if sl_pct > 0: return -sl_pct else: # 价格已穿越止损价,立即触发止损 return self.stoploss # 优先级 3: 移动止盈(三档尾随) # 若有 AI take_profit 且未全部执行完,跳过固定尾随(避免与分批止盈冲突) if signal and signal.get("take_profit") and trade: tp_list = signal["take_profit"] filled_orders = trade.orders or [] all_tp_filled = all( any( o.ft_order_tag == f"tp_{i}_{tp.get('price')}" and o.ft_is_open is False for o in filled_orders ) for i, tp in enumerate(tp_list[:-1]) ) if len(tp_list) >= 2 else True if not all_tp_filled: return self.stoploss # 全部 TP 完成后,保护尾随 if current_profit > 0.03: return -(current_profit - 0.02) if current_profit > self.ts_profit_3.value: # > 20%: 尾随 3%(收紧保护) return -(current_profit - 0.03) elif current_profit > self.ts_profit_2.value: # > 10%: 尾随 5% return -(current_profit - 0.05) elif current_profit > self.ts_profit_1.value: # > 5%: 移至成本线 return -0.001 return self.stoploss def custom_exit( self, pair: str, trade, current_time: datetime, current_rate: float, current_profit: float, **kwargs, ) -> str | None: """最终止盈级别全仓平仓 分批止盈的中间级别由 adjust_trade_position 减仓处理, 最后一级在这里触发全部平仓。 """ signal = self._get_signal_for_pair(pair) if not signal or not signal.get("take_profit"): return None tp_list = signal["take_profit"] if not tp_list: return None # 最后一级止盈 → 全仓平仓 last_tp = tp_list[-1] tp_price = last_tp.get("price") if tp_price is None: return None hit = (not trade.is_short and current_rate >= tp_price) or \ (trade.is_short and current_rate <= tp_price) if hit: return f"take_profit_full_{tp_price}" return None def adjust_trade_position( self, trade, current_time: datetime, current_rate: float, current_profit: float, min_stake: float, max_stake: float, current_entry_rate: float, current_exit_rate: float, current_entry_profit: float, current_exit_profit: float, **kwargs, ) -> float | None: """分批止盈:按 AI 信号的 take_profit 列表逐级减仓 返回负数表示减仓金额。最后一级由 custom_exit 全平。 """ signal = self._get_signal_for_pair(trade.pair) if not signal or not signal.get("take_profit"): return None tp_list = signal["take_profit"] if len(tp_list) < 2: # 只有一级止盈,由 custom_exit 全平处理 return None filled_orders = trade.orders or [] # 遍历除最后一级外的所有止盈级别 for i, tp in enumerate(tp_list[:-1]): tp_price = tp.get("price") tp_pct = tp.get("pct", 50) if tp_price is None: continue hit = (not trade.is_short and current_rate >= tp_price) or \ (trade.is_short and current_rate <= tp_price) # 检查是否已对该级别执行过减仓 tag = f"tp_{i}_{tp_price}" already_filled = any( o.ft_order_tag == tag for o in filled_orders if o.ft_is_open is False ) if hit and not already_filled: # P4: 基于剩余仓位计算减仓量 total_tp_pct_filled = sum( tp_list[j].get("pct", 0) for j in range(i) if any( o.ft_order_tag == f"tp_{j}_{tp_list[j].get('price')}" and o.ft_is_open is False for o in filled_orders ) ) remaining_pct = 100 - total_tp_pct_filled reduce_pct = min(tp_pct, remaining_pct) reduce_amount = trade.amount * current_rate / trade.leverage * reduce_pct / 100 logger.info( "[Agent] %s 分批止盈 TP%d @ %.2f, 减仓 %.1f%% (剩余 %.1f%%)", trade.pair, i + 1, tp_price, reduce_pct, remaining_pct, ) return -reduce_amount return None def custom_stake_amount( self, pair: str, current_time: datetime, current_rate: float, proposed_stake: float, min_stake: float | None, max_stake: float, leverage: float, entry_tag: str | None, side: str, **kwargs, ) -> float: """控制初始仓位大小""" min_stake_amount = min_stake if min_stake else 0 signal = self._get_signal_for_pair(pair) if signal and signal.get("position_size_usdt"): stake = min(signal["position_size_usdt"], max_stake) return max(stake, min_stake_amount) return max(proposed_stake, min_stake_amount) def leverage( self, pair: str, current_time: datetime, current_rate: float, proposed_leverage: float, max_leverage: float, entry_tag: str | None, side: str, **kwargs, ) -> float: """杠杆控制 优先读取 Agent 信号杠杆,否则使用默认值。 硬上限从 signal.max_leverage → settings.yaml → 5x 依次 fallback。 """ signal = self._get_signal_for_pair(pair) # 确定杠杆硬上限: signal.max_leverage → settings → 5 hard_limit = 5.0 if signal and signal.get("max_leverage"): hard_limit = float(signal["max_leverage"]) else: try: import yaml with open("config/settings.yaml") as f: settings = yaml.safe_load(f) or {} hard_limit = float( settings.get("risk", {}).get("max_leverage", 5) ) except Exception: pass if signal and signal.get("leverage"): lev = float(signal["leverage"]) return min(lev, max_leverage, hard_limit) return min(3.0, max_leverage, hard_limit) |
Strategy League — fixed backtest that feeds the ranking
Export report Freqtrade logsRun finished · took 315.1s
ℹ️ This strategy uses custom_stoploss() — freqtrade only
re-checks these once per 5m candle by default, not against the price movement within it.
For a more accurate read, re-run this backtest locally with --timeframe-detail 1m. Freqle doesn't do this for every check here: multiplying every
League/sweep backtest by a finer detail timeframe is more compute than the sandbox can sustain
across every indexed strategy. why this matters →
- profit isn't statistically significant (p=0.98) — hard to tell apart from luck
- only 0% of resampled runs were profitable
- profitable in only 14% of rolling 3-month windows
- did not beat simply holding the market
- very deep drawdown (-95%)
Resampling the trade sequence 2,000× shows the spread of results this edge could plausibly produce — separating a dependable strategy from one that got lucky once.
Loading charts…
Monthly breakdown
| Month | Regime | Trades | Profit % | Avg % | Win | Loss | Win % | DD % | Avg dur |
|---|---|---|---|---|---|---|---|---|---|
| Sep 2021 | bearish trending high vol | 448 | -31.59 | -0.71 | 117 | 331 | 26.1 | -95.04 | 1h 10m |
| Aug 2021 | bullish trending high vol | 802 | -29.52 | -0.37 | 244 | 558 | 30.4 | -80.29 | 1h 18m |
| Jul 2021 | bullish trending high vol | 832 | -40.33 | -0.49 | 241 | 591 | 29.0 | -64.95 | 1h 15m |
| Jun 2021 | bearish trending high vol | 869 | -22.58 | -0.26 | 288 | 581 | 33.1 | -46.32 | 1h 19m |
| May 2021 | bearish trending high vol | 918 | +38.47 | 0.42 | 312 | 606 | 34.0 | -60.95 | 1h 09m |
| Apr 2021 | bearish choppy high vol | 971 | -57.20 | -0.60 | 285 | 686 | 29.4 | -52.88 | 1h 17m |
| Mar 2021 | bullish choppy high vol | 911 | -35.05 | -0.39 | 275 | 636 | 30.2 | -24.87 | 1h 22m |
| Feb 2021 | bullish trending high vol | 764 | +48.76 | 0.66 | 278 | 486 | 36.4 | -9.25 | 1h 11m |
| Jan 2021 | bullish trending high vol | 865 | +38.90 | 0.45 | 318 | 547 | 36.8 | -10.06 | 1h 12m |
Trade charts — best 2 and worst 2 performing pairs (full OHLC candles are expensive to render for every pair)
Backtests — over a market period
Backtest this strategy over a chosen crypto-cycle period. These don't affect the League ranking, and need that period's candle data downloaded.
Log in or sign up to run backtests.
| Period | Range | Total % | Win % | Max DD | Trades | |
|---|---|---|---|---|---|---|
| 2020 · DeFi Summer & Pre-Halving Rally | 20200101-20210101 | not run | ||||
| 2021 · Institutional Bull Market | 20210101-20220101 | not run | ||||
| 2022 · Post-Bull Crash & Macro Tightening | 20220101-20230101 | not run | ||||
| 2023–2024 · Recovery & ETF Anticipation | 20230101-20250101 | not run | ||||
| 2025–2026 · Current Cycle | 20250101-20260101 | not run | ||||
Walk forward
Out-of-sample backtest on recent data · 33 pairs · 20260101-20260701.
Backtest trust check
1 potential lookahead pattern(s) found · 2 to review
| Line | Pattern | Detail | |
|---|---|---|---|
| 137 | leak | iloc_last | iloc[-1] in populate_* applies the newest candle to all rows |
| 223 | review | dead_callback | custom_stoploss() is defined but use_custom_stoploss isn't True, and freqtrade only calls it when that flag is set -- the method never runs and every trade uses the static stoploss |
| 355 | review | dead_callback | adjust_trade_position() is defined but position_adjustment_enable isn't True -- freqtrade never calls it, so the DCA/pyramiding logic here does nothing and every trade stays at its initial stake |
ran by Ron · took s
Lookahead analysis
freqtrade lookahead-analysis: detects strategies peeking at future candles.