""" RaptorBT + Mt5Bridge 集成测试 从 Mt5Bridge API 拉取真实 K 线数据，用 RaptorBT 跑回测。 运行前确保： pip install raptorbt requests numpy pandas 运行： python test_with_mt5.py """ import os import requests import numpy as np import pandas as pd import raptorbt from datetime import datetime, timedelta, timezone BRIDGE = "http://61.164.252.86:13485" API_KEY = "UiHMqtaYLZzwBdcuS4RFmEGhgDO8N2eI" SYMBOL = "XAUUSD" TIMEFRAME = "TIMEFRAME_H1" BARS = 500 def api_get(path, params=None): resp = requests.get( f"{BRIDGE}{path}", params=params, headers={"X-API-Key": API_KEY}, timeout=15, ) resp.raise_for_status() return resp.json() def check_health(): print("=" * 60) print("1. 健康检查") print("=" * 60) data = api_get("/health") print(f" 状态: {data.get('status')}") print(f" MT5 连接: {data.get('mt5_connected')}") if not data.get("mt5_connected"): print(" ⚠️ MT5 未连接，后续可能失败") return data def fetch_account(): print("\n" + "=" * 60) print("2. 账户信息") print("=" * 60) data = api_get("/account")["data"][0] print(f" 余额: {data['balance']:.2f} {data['currency']}") print(f" 净值: {data['equity']:.2f}") print(f" 浮动盈亏: {data['profit']:.2f}") print(f" 杠杆: 1:{data['leverage']}") return data def fetch_klines(): print("\n" + "=" * 60) print(f"3. 拉取 K 线数据 ({SYMBOL}, {TIMEFRAME}, 最近 {BARS} 根)") print("=" * 60) date_to = datetime.now(timezone.utc).strftime("%Y-%m-%d") date_from = (datetime.now(timezone.utc) - timedelta(days=BARS // 24 + 30)).strftime("%Y-%m-%d") data = api_get("/rates/from-date", params={ "symbol": SYMBOL, "timeframe": TIMEFRAME, "date_from": date_from, "date_to": date_to, }) rows = data.get("data", []) if not rows: print(" ⚠️ 没有拉到数据，尝试 /rates/from-pos ...") data = api_get("/rates/from-pos", params={ "symbol": SYMBOL, "timeframe": TIMEFRAME, "start_pos": 0, "count": BARS, }) rows = data.get("data", []) if not rows: raise RuntimeError("两种方式都没拉到 K 线，检查品种名或 MT5 连接") df = pd.DataFrame(rows) df["time"] = pd.to_datetime(df["time"]) df = df.sort_values("time").reset_index(drop=True) print(f" 拉到 {len(df)} 根 K 线") print(f" 时间范围: {df['time'].iloc[0]} ~ {df['time'].iloc[-1]}") print(f" Close 范围: {df['close'].min():.2f} ~ {df['close'].max():.2f}") return df def run_backtest(df): print("\n" + "=" * 60) print("4. RaptorBT 回测 (SMA 交叉策略)") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) timestamps = df["time"].values.astype("int64") sma_fast = raptorbt.sma(close, period=10) sma_slow = raptorbt.sma(close, period=20) entries = (sma_fast > sma_slow) & np.roll(sma_fast <= sma_slow, 1) exits = (sma_fast < sma_slow) & np.roll(sma_fast >= sma_slow, 1) entries[:20] = False exits[:20] = False entries = entries.astype(bool) exits = exits.astype(bool) n_entries = int(entries.sum()) n_exits = int(exits.sum()) print(f" SMA(10)/SMA(20) 交叉信号: {n_entries} 次入场, {n_exits} 次出场") config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_fixed_stop(0.02) config.set_fixed_target(0.04) result = raptorbt.run_single_backtest( timestamps=timestamps, open=open_, high=high, low=low, close=close, volume=volume, entries=entries, exits=exits, direction=1, weight=1.0, symbol=SYMBOL, config=config, ) m = result.metrics print(f"\n {'─' * 40}") print(f" 回测结果") print(f" {'─' * 40}") print(f" 总收益率: {m.total_return_pct:>10.2f} %") print(f" 夏普比率: {m.sharpe_ratio:>10.2f}") print(f" 索提诺比率: {m.sortino_ratio:>10.2f}") print(f" 卡玛比率: {m.calmar_ratio:>10.2f}") print(f" 最大回撤: {m.max_drawdown_pct:>10.2f} %") print(f" 总交易数: {m.total_trades:>10d}") print(f" 胜率: {m.win_rate_pct:>10.1f} %") print(f" 盈利因子: {m.profit_factor:>10.2f}") print(f" 平均交易收益: {m.avg_trade_return_pct:>10.2f} %") print(f" 平均盈利: {m.avg_win_pct:>10.2f} %") print(f" 平均亏损: {m.avg_loss_pct:>10.2f} %") print(f" 最佳交易: {m.best_trade_pct:>10.2f} %") print(f" 最差交易: {m.worst_trade_pct:>10.2f} %") print(f" 期望值: {m.expectancy:>10.2f}") print(f" SQN: {m.sqn:>10.2f}") print(f" 总手续费: {m.total_fees_paid:>10.2f}") print(f" 市场暴露: {m.exposure_pct:>10.1f} %") trades = result.trades() if trades: print(f"\n 最近 5 笔交易:") print(f" {'ID':>6} {'方向':>4} {'入场价':>12} {'出场价':>12} {'收益%':>10} {'出场原因':>12}") for t in trades[-5:]: direction = "多" if t.direction == 1 else "空" print(f" {t.id:>6} {direction:>4} {t.entry_price:>12.5f} {t.exit_price:>12.5f} {t.return_pct:>10.2f} {t.exit_reason:>12}") return result def run_multi_indicator_backtest(df): print("\n" + "=" * 60) print("5. 多策略回测 (SMA交叉 + RSI均值回归)") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) timestamps = df["time"].values.astype("int64") sma_fast = raptorbt.sma(close, period=10) sma_slow = raptorbt.sma(close, period=20) rsi = raptorbt.rsi(close, period=14) entries_sma = ((sma_fast > sma_slow) & np.roll(sma_fast <= sma_slow, 1)).astype(bool) exits_sma = ((sma_fast < sma_slow) & np.roll(sma_fast >= sma_slow, 1)).astype(bool) entries_rsi = (rsi < 30).astype(bool) exits_rsi = (rsi > 70).astype(bool) entries_sma[:20] = False exits_sma[:20] = False entries_rsi[:14] = False exits_rsi[:14] = False strategies = [ (entries_sma, exits_sma, 1, 0.6, "SMA_Cross"), (entries_rsi, exits_rsi, 1, 0.4, "RSI_MeanRev"), ] config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_trailing_stop(0.03) result = raptorbt.run_multi_backtest( timestamps=timestamps, open=open_, high=high, low=low, close=close, volume=volume, strategies=strategies, config=config, combine_mode="any", ) m = result.metrics print(f" 总收益率: {m.total_return_pct:.2f}% 夏普: {m.sharpe_ratio:.2f} " f"最大回撤: {m.max_drawdown_pct:.2f}% 交易数: {m.total_trades} 胜率: {m.win_rate_pct:.1f}%") return result def show_indicators(df): print("\n" + "=" * 60) print("6. 内置指标演示") print("=" * 60) close = df["close"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) indicators = { "SMA(20)": raptorbt.sma(close, period=20), "EMA(20)": raptorbt.ema(close, period=20), "RSI(14)": raptorbt.rsi(close, period=14), "ATR(14)": raptorbt.atr(high, low, close, period=14), "ADX(14)": raptorbt.adx(high, low, close, period=14), "VWAP": raptorbt.vwap(high, low, close, volume), } macd_line, signal_line, hist = raptorbt.macd(close, 12, 26, 9) indicators["MACD_Line"] = macd_line indicators["MACD_Signal"] = signal_line stoch_k, stoch_d = raptorbt.stochastic(high, low, close, k_period=14, d_period=3) indicators["Stoch_K"] = stoch_k indicators["Stoch_D"] = stoch_d upper, middle, lower = raptorbt.bollinger_bands(close, period=20, std_dev=2.0) indicators["BB_Upper"] = upper indicators["BB_Lower"] = lower supertrend_val, supertrend_dir = raptorbt.supertrend(high, low, close, period=10, multiplier=3.0) indicators["Supertrend"] = supertrend_val indicators["Rolling_Min(20)"] = raptorbt.rolling_min(low, period=20) indicators["Rolling_Max(20)"] = raptorbt.rolling_max(high, period=20) last_idx = -1 print(f"\n 最新一根 K 线的指标值:") print(f" {'指标':<20} {'值':>15}") print(f" {'─' * 40}") for name, arr in indicators.items(): val = arr[last_idx] if np.isnan(val): print(f" {name:<20} {'NaN':>15}") else: print(f" {name:<20} {val:>15.4f}") def run_atr_stop_backtest(df): print("\n" + "=" * 60) print("7. ATR 动态止损 + 风险回报比止盈") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) timestamps = df["time"].values.astype("int64") sma_fast = raptorbt.sma(close, period=10) sma_slow = raptorbt.sma(close, period=20) entries = ((sma_fast > sma_slow) & np.roll(sma_fast <= sma_slow, 1)).astype(bool) exits = ((sma_fast < sma_slow) & np.roll(sma_fast >= sma_slow, 1)).astype(bool) entries[:20] = False exits[:20] = False config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_atr_stop(multiplier=2.0, period=14) config.set_risk_reward_target(ratio=2.0) result = raptorbt.run_single_backtest( timestamps=timestamps, open=open_, high=high, low=low, close=close, volume=volume, entries=entries, exits=exits, direction=1, weight=1.0, symbol=SYMBOL, config=config, ) m = result.metrics print(f" 策略: SMA 交叉 + 2×ATR 止损 + 2:1 风险回报止盈") print(f" 总收益率: {m.total_return_pct:.2f}% 夏普: {m.sharpe_ratio:.2f} " f"最大回撤: {m.max_drawdown_pct:.2f}% 交易数: {m.total_trades} 胜率: {m.win_rate_pct:.1f}%") trades = result.trades() exit_reasons = {} for t in trades: r = t.exit_reason exit_reasons[r] = exit_reasons.get(r, 0) + 1 print(f" 出场原因分布: {exit_reasons}") return result def run_basket_backtest(df): print("\n" + "=" * 60) print("8. 篮子回测 (多标的同步信号)") print("=" * 60) other_symbols = ["XAUUSD", "EURUSD", "GBPUSD"] dfs = {} for sym in other_symbols: try: date_to = datetime.now(timezone.utc).strftime("%Y-%m-%d") date_from = (datetime.now(timezone.utc) - timedelta(days=BARS // 24 + 30)).strftime("%Y-%m-%d") data = api_get("/rates/from-date", params={ "symbol": sym, "timeframe": TIMEFRAME, "date_from": date_from, "date_to": date_to, }) rows = data.get("data", []) if not rows: print(f" ⚠️ {sym} 无数据，跳过") continue sdf = pd.DataFrame(rows) sdf["time"] = pd.to_datetime(sdf["time"]) sdf = sdf.sort_values("time").reset_index(drop=True) dfs[sym] = sdf print(f" {sym}: 拉到 {len(sdf)} 根 K 线") except Exception as e: print(f" ⚠️ {sym} 拉取失败: {e}") if len(dfs) < 2: print(" ⚠️ 可用品种不足 2 个，跳过篮子回测") return None min_len = min(len(d) for d in dfs.values()) instruments = [] instrument_configs = {} per_capital = 100000.0 / len(dfs) for i, (sym, sdf) in enumerate(dfs.items()): c = sdf["close"].values.astype(np.float64)[:min_len] o = sdf["open"].values.astype(np.float64)[:min_len] h = sdf["high"].values.astype(np.float64)[:min_len] l = sdf["low"].values.astype(np.float64)[:min_len] v = sdf["tick_volume"].values.astype(np.float64)[:min_len] ts = sdf["time"].values.astype("int64")[:min_len] sma_f = raptorbt.sma(c, period=10) sma_s = raptorbt.sma(c, period=20) ent = ((sma_f > sma_s) & np.roll(sma_f <= sma_s, 1)).astype(bool) ext = ((sma_f < sma_s) & np.roll(sma_f >= sma_s, 1)).astype(bool) ent[:20] = False ext[:20] = False instruments.append((ts, o, h, l, c, v, ent, ext, 1, 1.0 / len(dfs), sym)) inst_cfg = raptorbt.PyInstrumentConfig(lot_size=0.01, alloted_capital=per_capital) instrument_configs[sym] = inst_cfg config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_trailing_stop(0.03) for mode in ["any", "all"]: result = raptorbt.run_basket_backtest( instruments=instruments, config=config, sync_mode=mode, instrument_configs=instrument_configs, ) m = result.metrics print(f" sync_mode={mode:<8} → 收益: {m.total_return_pct:>7.2f}% " f"夏普: {m.sharpe_ratio:>6.2f} 回撤: {m.max_drawdown_pct:>6.2f}% " f"交易: {m.total_trades:>3d} 胜率: {m.win_rate_pct:>5.1f}%") return result def run_pairs_backtest(df): print("\n" + "=" * 60) print("9. 配对交易 (EURUSD vs GBPUSD)") print("=" * 60) pairs = [("EURUSD", "GBPUSD"), ("EURUSD", "USDJPY")] for leg1_sym, leg2_sym in pairs: try: date_to = datetime.now(timezone.utc).strftime("%Y-%m-%d") date_from = (datetime.now(timezone.utc) - timedelta(days=BARS // 24 + 30)).strftime("%Y-%m-%d") data1 = api_get("/rates/from-date", params={ "symbol": leg1_sym, "timeframe": TIMEFRAME, "date_from": date_from, "date_to": date_to, }) data2 = api_get("/rates/from-date", params={ "symbol": leg2_sym, "timeframe": TIMEFRAME, "date_from": date_from, "date_to": date_to, }) rows1 = data1.get("data", []) rows2 = data2.get("data", []) if not rows1 or not rows2: print(f" ⚠️ {leg1_sym}/{leg2_sym} 数据不足，跳过") continue df1 = pd.DataFrame(rows1) df2 = pd.DataFrame(rows2) df1["time"] = pd.to_datetime(df1["time"]) df2["time"] = pd.to_datetime(df2["time"]) df1 = df1.sort_values("time").reset_index(drop=True) df2 = df2.sort_values("time").reset_index(drop=True) min_len = min(len(df1), len(df2)) c1 = df1["close"].values.astype(np.float64)[:min_len] c2 = df2["close"].values.astype(np.float64)[:min_len] spread = c1 - c2 spread_sma = pd.Series(spread).rolling(20).mean().values spread_std = pd.Series(spread).rolling(20).std().values zscore = (spread - spread_sma) / np.where(spread_std > 0, spread_std, 1e-10) entries = (zscore < -2.0).astype(bool) exits = (np.abs(zscore) < 0.5).astype(bool) entries[:20] = False exits[:20] = False config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_fixed_stop(0.03) result = raptorbt.run_pairs_backtest( leg1_timestamps=df1["time"].values.astype("int64")[:min_len], leg1_open=df1["open"].values.astype(np.float64)[:min_len], leg1_high=df1["high"].values.astype(np.float64)[:min_len], leg1_low=df1["low"].values.astype(np.float64)[:min_len], leg1_close=c1, leg1_volume=df1["tick_volume"].values.astype(np.float64)[:min_len], leg2_timestamps=df2["time"].values.astype("int64")[:min_len], leg2_open=df2["open"].values.astype(np.float64)[:min_len], leg2_high=df2["high"].values.astype(np.float64)[:min_len], leg2_low=df2["low"].values.astype(np.float64)[:min_len], leg2_close=c2, leg2_volume=df2["tick_volume"].values.astype(np.float64)[:min_len], entries=entries, exits=exits, direction=1, symbol=f"{leg1_sym}_{leg2_sym}", config=config, hedge_ratio=1.0, ) m = result.metrics print(f" {leg1_sym}/{leg2_sym} 配对 (Z-score 均值回归):") print(f" 收益: {m.total_return_pct:.2f}% 夏普: {m.sharpe_ratio:.2f} " f"回撤: {m.max_drawdown_pct:.2f}% 交易: {m.total_trades} 胜率: {m.win_rate_pct:.1f}%") except Exception as e: print(f" ⚠️ {leg1_sym}/{leg2_sym} 配对失败: {e}") return None def run_monte_carlo(result1, result2): print("\n" + "=" * 60) print("10. 蒙特卡洛组合模拟") print("=" * 60) returns1 = result1.returns() returns2 = result2.returns() min_len = min(len(returns1), len(returns2)) r1 = returns1[:min_len] r2 = returns2[:min_len] mask = ~(np.isnan(r1) | np.isnan(r2)) r1 = r1[mask] r2 = r2[mask] if len(r1) < 50: print(" ⚠️ 有效收益率数据不足，跳过蒙特卡洛模拟") return corr = np.corrcoef(r1, r2)[0, 1] correlation_matrix = [ np.array([1.0, corr]), np.array([corr, 1.0]), ] print(f" 策略 1 (SMA交叉) vs 策略 2 (多策略) 收益率相关系数: {corr:.3f}") for n_sim in [1000, 10000]: mc_result = raptorbt.simulate_portfolio_mc( returns=[r1, r2], weights=np.array([0.6, 0.4]), correlation_matrix=correlation_matrix, initial_value=100000.0, n_simulations=n_sim, horizon_days=252, seed=42, ) print(f"\n 模拟次数: {n_sim:,}") print(f" {'─' * 40}") print(f" 预期收益: {mc_result['expected_return']:>10.2f} %") print(f" 亏损概率: {mc_result['probability_of_loss']:>10.2%}") print(f" VaR (95%): {mc_result['var_95']:>10.2f} %") print(f" CVaR (95%): {mc_result['cvar_95']:>10.2f} %") print(f" 分位数路径终值:") for pct, path in mc_result["percentile_paths"]: print(f" P{pct:>2.0f}: {path[-1]:>12,.2f}") def export_results(result, df, strategy_name="sma_cross"): print("\n" + "=" * 60) print("11. 导出回测结果") print("=" * 60) output_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "backtest_output") os.makedirs(output_dir, exist_ok=True) trades = result.trades() if trades: rows = [] for t in trades: rows.append({ "trade_id": t.id, "symbol": t.symbol, "direction": "Long" if t.direction == 1 else "Short", "entry_idx": t.entry_idx, "exit_idx": t.exit_idx, "entry_time": df["time"].iloc[t.entry_idx] if t.entry_idx < len(df) else "", "exit_time": df["time"].iloc[t.exit_idx] if t.exit_idx < len(df) else "", "entry_price": t.entry_price, "exit_price": t.exit_price, "size": t.size, "pnl": t.pnl, "return_pct": t.return_pct, "fees": t.fees, "exit_reason": t.exit_reason, }) trades_df = pd.DataFrame(rows) trades_path = os.path.join(output_dir, f"{strategy_name}_trades.csv") trades_df.to_csv(trades_path, index=False, encoding="utf-8-sig") print(f" 交易记录 → {trades_path} ({len(trades_df)} 笔)") equity = result.equity_curve() drawdown = result.drawdown_curve() returns = result.returns() curve_df = pd.DataFrame({ "time": df["time"].values[:len(equity)], "equity": equity, "drawdown": drawdown, "returns": returns, }) curve_path = os.path.join(output_dir, f"{strategy_name}_curves.csv") curve_df.to_csv(curve_path, index=False, encoding="utf-8-sig") print(f" 权益曲线 → {curve_path} ({len(curve_df)} 根)") m = result.metrics metrics_dict = m.to_dict() metrics_dict["total_trades"] = m.total_trades metrics_dict["total_closed_trades"] = m.total_closed_trades metrics_dict["total_open_trades"] = m.total_open_trades metrics_dict["winning_trades"] = m.winning_trades metrics_dict["losing_trades"] = m.losing_trades metrics_dict["max_consecutive_wins"] = m.max_consecutive_wins metrics_dict["max_consecutive_losses"] = m.max_consecutive_losses metrics_dict["avg_holding_period"] = m.avg_holding_period metrics_dict["avg_winning_duration"] = m.avg_winning_duration metrics_dict["avg_losing_duration"] = m.avg_losing_duration metrics_dict["start_value"] = m.start_value metrics_dict["end_value"] = m.end_value metrics_dict["total_fees_paid"] = m.total_fees_paid metrics_dict["open_trade_pnl"] = m.open_trade_pnl metrics_dict["exposure_pct"] = m.exposure_pct metrics_dict["payoff_ratio"] = m.payoff_ratio metrics_dict["recovery_factor"] = m.recovery_factor metrics_dict["omega_ratio"] = m.omega_ratio metrics_df = pd.DataFrame(list(metrics_dict.items()), columns=["metric", "value"]) metrics_path = os.path.join(output_dir, f"{strategy_name}_metrics.csv") metrics_df.to_csv(metrics_path, index=False, encoding="utf-8-sig") print(f" 绩效指标 → {metrics_path} ({len(metrics_df)} 项)") print(f"\n 所有文件保存在: {output_dir}") def test_ferro_indicators(df): print("\n" + "=" * 60) print("7. ferro-ta 新指标全量测试 (45+ 指标)") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) n = len(close) errors = [] tested = 0 def check(name, arr, expect_len=None, expect_finite=True, value_range=None, min_valid=1): nonlocal tested tested += 1 if expect_len is not None and len(arr) != expect_len: errors.append(f"{name}: 长度 {len(arr)} != 期望 {expect_len}") return valid = arr[~np.isnan(arr)] if len(valid) < min_valid: errors.append(f"{name}: 有效值仅 {len(valid)} 个 (最少需要 {min_valid})") return if expect_finite: inf_count = np.isinf(valid).sum() if inf_count > 0: errors.append(f"{name}: 有 {inf_count} 个 inf 值") if value_range and len(valid) > 0: lo, hi = value_range out_of_range = ((valid < lo) | (valid > hi)).sum() if out_of_range > 0: errors.append(f"{name}: {out_of_range} 个值超出 [{lo}, {hi}]") last = arr[-1] status = f"{last:.4f}" if not np.isnan(last) else "NaN" print(f" {name:<25} 长度={len(arr):>4} 有效值={len(valid):>4} 最新={status}") # ── Overlap 均线类 ── print("\n ── Overlap 均线类 ──") check("WMA(20)", raptorbt.wma(close, period=20), n, min_valid=10) check("DEMA(20)", raptorbt.dema(close, period=20), n, min_valid=10) check("TEMA(20)", raptorbt.tema(close, period=20), n, min_valid=10) check("KAMA(10)", raptorbt.kama(close, period=10), n, min_valid=10) check("T3(20)", raptorbt.t3(close, period=20, vfactor=0.7), n, min_valid=10) check("TRIMA(20)", raptorbt.trima(close, period=20), n, min_valid=10) check("Midpoint(20)", raptorbt.midpoint(close, period=20), n, min_valid=10) check("Midprice(20)", raptorbt.midprice(high, low, period=20), n, min_valid=10) check("SAR(0.02,0.2)", raptorbt.sar(high, low, acceleration=0.02, maximum=0.2), n, min_valid=10) # ── Momentum 动量类 ── print("\n ── Momentum 动量类 ──") check("CCI(20)", raptorbt.cci(high, low, close, period=20), n, min_valid=10) check("WillR(14)", raptorbt.willr(high, low, close, period=14), n, value_range=(-101, 1), min_valid=10) check("ROC(10)", raptorbt.roc(close, period=10), n, min_valid=10) check("MOM(10)", raptorbt.mom(close, period=10), n, min_valid=10) check("CMO(14)", raptorbt.cmo(close, period=14), n, min_valid=10) check("TRIX(12)", raptorbt.trix(close, period=12), n, min_valid=10) check("UltOSC(7,14,28)", raptorbt.ultosc(high, low, close, period1=7, period2=14, period3=28), n, value_range=(-1, 101), min_valid=10) check("AroonOsc(14)", raptorbt.aroonosc(high, low, period=14), n, value_range=(-101, 101), min_valid=10) check("BOP", raptorbt.bop(open_, high, low, close), n, value_range=(-1.01, 1.01), min_valid=10) check("Plus_DI(14)", raptorbt.plus_di(high, low, close, period=14), n, value_range=(-1, 101), min_valid=10) check("Minus_DI(14)", raptorbt.minus_di(high, low, close, period=14), n, value_range=(-1, 101), min_valid=10) # ── 多输出动量指标 ── print("\n ── 多输出动量指标 ──") try: stochrsi_k, stochrsi_d = raptorbt.stochrsi(close, timeperiod=14, fastk_period=5, fastd_period=3) check("StochRSI_K", stochrsi_k, n, value_range=(-5, 105), min_valid=10) check("StochRSI_D", stochrsi_d, n, value_range=(-5, 105), min_valid=10) except Exception as e: errors.append(f"StochRSI: {e}") tested += 2 try: adx_val, plus_di, minus_di = raptorbt.adx_all(high, low, close, period=14) check("ADX_all(14)", adx_val, n, value_range=(-1, 101), min_valid=10) check("ADX_all +DI", plus_di, n, value_range=(-1, 101), min_valid=10) check("ADX_all -DI", minus_di, n, value_range=(-1, 101), min_valid=10) except Exception as e: errors.append(f"ADX_all: {e}") tested += 3 try: aroon_up, aroon_down = raptorbt.aroon(high, low, period=14) check("Aroon_Up", aroon_up, n, value_range=(-1, 101), min_valid=10) check("Aroon_Down", aroon_down, n, value_range=(-1, 101), min_valid=10) except Exception as e: errors.append(f"Aroon: {e}") tested += 2 try: ppo_line, ppo_signal, ppo_hist = raptorbt.ppo(close, fastperiod=12, slowperiod=26, signalperiod=9) check("PPO_Line", ppo_line, n, min_valid=10) check("PPO_Signal", ppo_signal, n, min_valid=10) check("PPO_Hist", ppo_hist, n, min_valid=10) except Exception as e: errors.append(f"PPO: {e}") tested += 3 # ── Volatility 波动率类 ── print("\n ── Volatility 波动率类 ──") check("NATR(14)", raptorbt.natr(high, low, close, period=14), n, min_valid=10) check("TRange", raptorbt.trange(high, low, close), n, min_valid=10) check("StdDev(20)", raptorbt.stddev(close, period=20, nbdev=1.0), n, min_valid=10) check("VAR(20)", raptorbt.var(close, period=20, nbdev=1.0), n, min_valid=10) # ── Volume 成交量类 ── print("\n ── Volume 成交量类 ──") check("AD", raptorbt.ad(high, low, close, volume), n, min_valid=10) check("ADOSC(3,10)", raptorbt.adosc(high, low, close, volume, fastperiod=3, slowperiod=10), n, min_valid=10) check("OBV", raptorbt.obv(close, volume), n, min_valid=10) check("MFI(14)", raptorbt.mfi(high, low, close, volume, period=14), n, value_range=(-1, 101), min_valid=10) # ── Price Transform 价格变换类 ── print("\n ── Price Transform 价格变换类 ──") check("TypPrice", raptorbt.typprice(high, low, close), n, min_valid=10) check("MedPrice", raptorbt.medprice(high, low), n, min_valid=10) check("AvgPrice", raptorbt.avgprice(open_, high, low, close), n, min_valid=10) check("WclPrice", raptorbt.wclprice(high, low, close), n, min_valid=10) # ── Statistic 统计类 ── print("\n ── Statistic 统计类 ──") check("LinearReg(20)", raptorbt.linearreg(close, period=20), n, min_valid=10) check("LinReg_Slope(20)", raptorbt.linearreg_slope(close, period=20), n, min_valid=10) check("LinReg_Angle(20)", raptorbt.linearreg_angle(close, period=20), n, min_valid=10) check("LinReg_Intercept(20)", raptorbt.linearreg_intercept(close, period=20), n, min_valid=10) check("TSF(20)", raptorbt.tsf(close, period=20), n, min_valid=10) check("Beta(20)", raptorbt.beta(close, close, period=20), n, min_valid=10) check("Correl(20)", raptorbt.correl(close, close, period=20), n, min_valid=10) # ── ADXr ── print("\n ── ADXr ──") check("ADXr(14)", raptorbt.adxr(high, low, close, period=14), n, value_range=(-1, 101), min_valid=10) # ── APO ── print("\n ── APO ──") check("APO(12,26)", raptorbt.apo(close, fastperiod=12, slowperiod=26), n, min_valid=10) # ── 汇总 ── print("\n" + "─" * 60) if errors: print(f" ❌ 测试 {tested} 个指标，发现 {len(errors)} 个错误:") for e in errors: print(f" - {e}") else: print(f" ✅ 全部 {tested} 个指标测试通过，无错误！") print("─" * 60) return len(errors) == 0 def run_ferro_strategy_backtest(df): print("\n" + "=" * 60) print("8. ferro-ta 指标策略回测 (SAR + ADX + CCI 组合)") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) timestamps = df["time"].values.astype("int64") sar_val = raptorbt.sar(high, low, acceleration=0.02, maximum=0.2) adx_val, plus_di, minus_di = raptorbt.adx_all(high, low, close, period=14) cci_val = raptorbt.cci(high, low, close, period=20) sar_above = sar_val > close sar_below = sar_val < close adx_strong = adx_val > 25 bullish_di = plus_di > minus_di bearish_di = minus_di > plus_di cci_oversold = cci_val < -100 cci_overbought = cci_val > 100 entries_long = sar_below & adx_strong & bullish_di & cci_oversold entries_short = sar_above & adx_strong & bearish_di & cci_overbought entries = entries_long | entries_short exits = (sar_below & bearish_di & adx_strong) | (sar_above & bullish_di & adx_strong) warmup = 40 entries[:warmup] = False exits[:warmup] = False n_long = int(entries_long.sum()) n_short = int(entries_short.sum()) n_exits = int(exits.sum()) print(f" SAR+ADX+CCI 组合信号: {n_long} 次做多, {n_short} 次做空, {n_exits} 次出场") config = raptorbt.PyBacktestConfig( initial_capital=100000.0, fees=0.001, slippage=0.0005, ) config.set_atr_stop(multiplier=2.5, period=14) config.set_fixed_target(0.04) result = raptorbt.run_single_backtest( timestamps=timestamps, open=open_, high=high, low=low, close=close, volume=volume, entries=entries, exits=exits, direction=1, weight=1.0, symbol=SYMBOL, config=config, ) m = result.metrics print(f"\n {'─' * 40}") print(f" SAR+ADX+CCI 策略回测结果") print(f" {'─' * 40}") print(f" 总收益率: {m.total_return_pct:>10.2f} %") print(f" 夏普比率: {m.sharpe_ratio:>10.2f}") print(f" 最大回撤: {m.max_drawdown_pct:>10.2f} %") print(f" 总交易数: {m.total_trades:>10d}") print(f" 胜率: {m.win_rate_pct:>10.1f} %") print(f" 盈利因子: {m.profit_factor:>10.2f}") trades = result.trades() if trades: exit_reasons = {} for t in trades: r = t.exit_reason exit_reasons[r] = exit_reasons.get(r, 0) + 1 print(f" 出场原因分布: {exit_reasons}") return result def test_extended_indicators(df): """P0 批 + Hilbert + 市场状态 + 投资组合工具 全量测试""" print("\n" + "=" * 60) print("12. P0 扩展指标全量测试") print("=" * 60) close = df["close"].values.astype(np.float64) open_ = df["open"].values.astype(np.float64) high = df["high"].values.astype(np.float64) low = df["low"].values.astype(np.float64) volume = df["tick_volume"].values.astype(np.float64) n = len(close) errors = [] def check(name, arr, expect_len=None, value_range=None, min_valid=1, dtype=None): if expect_len is not None and len(arr) != expect_len: errors.append(f"{name}: 长度 {len(arr)} != 期望 {expect_len}") return valid = arr[~np.isnan(arr)] if len(valid) < min_valid: errors.append(f"{name}: 有效值仅 {len(valid)} 个 (最少需要 {min_valid})") return if value_range and len(valid) > 0: lo, hi = value_range out = ((valid < lo) | (valid > hi)).sum() if out > 0: errors.append(f"{name}: {out} 个值超出 [{lo}, {hi}]") if dtype is not None and not isinstance(arr, dtype): errors.append(f"{name}: dtype 应为 {dtype.__name__} 而非 {type(arr).__name__}") last = arr[-1] status = f"{last:.4f}" if not np.isnan(last) else "NaN" print(f" {name:<30} 长度={len(arr):>4} 有效值={len(valid):>4} 最新={status}") def check_tuple(name, arrs, expect_len=None, value_ranges=None, min_valids=None): """测试多返回值的指标""" if not min_valids: min_valids = [1] * len(arrs) if not value_ranges: value_ranges = [None] * len(arrs) for i, (a, vmin, vr) in enumerate(zip(arrs, min_valids, value_ranges)): sub = check(f"{name}[{i}]", a, expect_len, vr, vmin) # ── P0 扩展指标 ── print("\n ── P0 扩展指标 ──") check("VWMA(20)", raptorbt.vwma(close, volume, period=20), n, min_valid=10) donch = raptorbt.donchian(high, low, period=20) check("Donchian(20) upper", donch[0], n, min_valid=10) check("Donchian(20) middle", donch[1], n, min_valid=10) check("Donchian(20) lower", donch[2], n, min_valid=10) # Donchian 的上下边界应当包含收盘价 donch_upper = donch[0] donch_lower = donch[2] valid_idx = ~np.isnan(donch_upper) & ~np.isnan(donch_lower) if valid_idx.sum() > 0: bad = ((donch_upper[valid_idx] < close[valid_idx]) | (donch_lower[valid_idx] > close[valid_idx])).sum() if bad > 0: errors.append(f"Donchian: {bad} 根 K 线收盘价落在通道之外") ci_val = raptorbt.choppiness_index(high, low, close, period=14) check("Choppiness(14)", ci_val, n, value_range=(-1, 101), min_valid=10) check("HullMA(14)", raptorbt.hull_ma(close, period=14), n, min_valid=10) cl, cs = raptorbt.chandelier_exit(high, low, close, period=22, multiplier=3.0) check("Chandelier long_exit", cl, n, min_valid=10) check("Chandelier short_exit", cs, n, min_valid=10) # Note: long_exit can be < short_exit in sharp trending markets — this is valid. # The real check is that both are finite and reasonably close to price. tenkan, kijun, senkou_a, senkou_b, chikou = raptorbt.ichimoku( high, low, close, tenkan_period=9, kijun_period=26, senkou_b_period=52, displacement=26) check("Ichimoku tenkan", tenkan, n, min_valid=10) check("Ichimoku kijun", kijun, n, min_valid=10) check("Ichimoku senkou_a", senkou_a, n, min_valid=10) check("Ichimoku senkou_b", senkou_b, n, min_valid=10) check("Ichimoku chikou", chikou, n, min_valid=10) pivot, r1, s1, r2, s2 = raptorbt.pivot_points(high, low, close, method="classic") check("Pivot pivot", pivot, n, min_valid=5) check("Pivot R1", r1, n, min_valid=5) check("Pivot S1", s1, n, min_valid=5) check("Pivot R2", r2, n, min_valid=5) check("Pivot S2", s2, n, min_valid=5) # R2 > R1 > pivot > S1 > S2 对于有效值 valid_pp = ~np.isnan(pivot) & ~np.isnan(r1) & ~np.isnan(s1) & ~np.isnan(r2) & ~np.isnan(s2) if valid_pp.sum() > 0: order = (r2[valid_pp] > r1[valid_pp]) & (r1[valid_pp] > pivot[valid_pp]) & \ (pivot[valid_pp] > s1[valid_pp]) & (s1[valid_pp] > s2[valid_pp]) bad = (~order).sum() if bad > 0: errors.append(f"Pivot: {bad} 处 R2>R1>PIVOT>S1>S2 排序异常") # ── Hilbert Transform ── print("\n ── Hilbert Transform (周期变换) ──") check("HT_Trendline", raptorbt.ht_trendline(close), n, min_valid=10) check("HT_DCPeriod", raptorbt.ht_dcperiod(close), n, value_range=(-1, 100), min_valid=10) check("HT_DCPhase", raptorbt.ht_dcphase(close), n, value_range=(-361, 361), min_valid=10) ip, quad = raptorbt.ht_phasor(close) check("HT_Phasor in_phase", ip, n, min_valid=10) check("HT_Phasor quadrature", quad, n, min_valid=10) sin, lead = raptorbt.ht_sine(close) check("HT_Sine", sin, n, value_range=(-1.05, 1.05), min_valid=10) check("HT_LeadSine", lead, n, value_range=(-1.05, 1.05), min_valid=10) mode = raptorbt.ht_trendmode(close) # mode 是 i32 数组 check("HT_TrendMode", mode, n, value_range=(-2, 2), min_valid=10) # 验证 TrendMode 只有 0/1 mode_valid = mode[~np.isnan(mode)] if mode.dtype.kind == 'f' else mode unique = np.unique(mode_valid) bad = np.isin(unique, [0, 1]).all() == False and len(unique) > 0 if len(unique) > 0 and not set(unique) <= {0, 1}: errors.append(f"HT_TrendMode: 含有非 0/1 值: {unique}") # ── 市场状态检测 ── print("\n ── 市场状态检测 ──") adx_raw = raptorbt.adx(high, low, close, period=14) r_adx = raptorbt.regime_adx(adx_raw, threshold=25.0) check("Regime_ADX", r_adx, n, value_range=(-2, 2), min_valid=5) # 验证只有 -1/0/1 unique_r = np.unique(r_adx) if not set(unique_r) <= {-1, 0, 1}: errors.append(f"Regime_ADX: 含有非 -1/0/1 值: {unique_r}") atr_raw = raptorbt.atr(high, low, close, period=14) r_combo = raptorbt.regime_combined(adx_raw, atr_raw, close, 25.0, 2.0) check("Regime_Combined", r_combo, n, value_range=(-2, 2), min_valid=5) unique_c = np.unique(r_combo) if not set(unique_c) <= {-1, 0, 1}: errors.append(f"Regime_Combined: 含有非 -1/0/1 值: {unique_c}") breaks = raptorbt.detect_breaks_cusum(close, window=30, threshold=1.5, slack=0.5) check("Breaks_CUSUM", breaks, n, value_range=(-2, 2), min_valid=5) unique_b = np.unique(breaks) if not set(unique_b) <= {-1, 0, 1}: errors.append(f"Breaks_CUSUM: 含有非 -1/0/1 值: {unique_b}") vol_breaks = raptorbt.rolling_variance_break(close, short_window=10, long_window=30, threshold=2.0) check("Breaks_VarRatio", vol_breaks, n, value_range=(-2, 2), min_valid=5) unique_v = np.unique(vol_breaks) if not set(unique_v) <= {-1, 0, 1}: errors.append(f"Breaks_VarRatio: 含有非 -1/0/1 值: {unique_v}") # ── 投资组合工具 ── print("\n ── 投资组合工具 ──") # 创建第二组数据（随机偏移模拟） close2 = close * (1 + np.random.uniform(-0.01, 0.01, n)) rb = raptorbt.rolling_beta(close, close2, window=20) check("Rolling_Beta", rb, n, min_valid=10) dd_series, max_dd = raptorbt.drawdown_series(close) check("DrawdownSeries", dd_series, n, value_range=(-1.01, 0.01), min_valid=5) if max_dd > 0: errors.append(f"DrawdownSeries max_dd 应为非正值: {max_dd}") check("ZScore(20)", raptorbt.zscore_series(close, window=20), n, min_valid=10) returns = np.diff(close) / close[:-1] returns2 = np.diff(close2) / close2[:-1] rs = raptorbt.relative_strength(returns, returns2) check("RelativeStrength", rs, None, min_valid=10) hedge = 1.0 spr = raptorbt.spread(close, close2, hedge) check("Spread", spr, n, min_valid=10) # 验证 spread ≈ close - close2 valid_sp = ~np.isnan(spr) & ~np.isnan(close - close2) if valid_sp.sum() > 0: diff = np.abs(spr[valid_sp] - (close[valid_sp] - close2[valid_sp])) bad = (diff > 1e-6).sum() if bad > 0: errors.append(f"Spread: {bad} 处与 close - close2 偏差超过 1e-6") r = raptorbt.ratio(close, close2) check("Ratio", r, n, min_valid=10) # 验证 ratio ≈ close / close2 valid_r = ~np.isnan(r) & ~np.isnan(close / close2) if valid_r.sum() > 0: diff_r = np.abs(r[valid_r] - close[valid_r] / close2[valid_r]) bad_r = (diff_r > 1e-6).sum() if bad_r > 0: errors.append(f"Ratio: {bad_r} 处与 close / close2 偏差超过 1e-6") # ── 汇总 ── print("\n" + "─" * 60) if errors: print(f" ❌ 测试发现 {len(errors)} 个错误:") for e in errors: print(f" - {e}") else: print(f" ✅ 全部扩展指标测试通过，无错误！") print("─" * 60) return len(errors) == 0 def main(): print("╔══════════════════════════════════════════════════════════╗") print("║ RaptorBT + Mt5Bridge 集成测试 ║") print("║ 从 MT5 拉取真实数据 → RaptorBT 回测 ║") print("╚══════════════════════════════════════════════════════════╝\n") check_health() fetch_account() df = fetch_klines() result1 = run_backtest(df) result2 = run_multi_indicator_backtest(df) show_indicators(df) test_ferro_indicators(df) test_extended_indicators(df) result_ferro = run_ferro_strategy_backtest(df) result3 = run_atr_stop_backtest(df) run_basket_backtest(df) run_pairs_backtest(df) run_monte_carlo(result1, result2) export_results(result1, df, strategy_name="sma_cross") export_results(result2, df, strategy_name="multi_strategy") if result3: export_results(result3, df, strategy_name="atr_stop_rr") if result_ferro: export_results(result_ferro, df, strategy_name="ferro_sar_adx_cci") print("\n" + "=" * 60) print("✅ 全部测试完成！") print("=" * 60) if __name__ == "__main__": main()