import os import pandas as pd import numpy as np import matplotlib.pyplot as plt from tqdm import tqdm from concurrent.futures import ProcessPoolExecutor, as_completed import logging from numba import jit from datetime import datetime # ========== 环境配置 ========== plt.rcParams['font.sans-serif'] = ['SimHei'] plt.rcParams['axes.unicode_minus'] = False logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s') # ========== 策略参数配置 ========== # ========== 全局参数 ========== ATR_WINDOW = 5 VOLATILITY_WINDOW = 10 BULL_THRESHOLD = 0.83 BEAR_THRESHOLD = 0.77 NEUTRAL_THRESHOLD = 0.81 HOLDING_DAYS_MAP = { 'bull': 4, 'bear': 2, 'neutral': 3 } # ========== 核心策略逻辑 ========== @jit(nopython=True) def calculate_technical_indicators(close, high, low, volume, atr_window, volatility_window): """支持Numba的参数传递""" n = len(close) macd = np.zeros(n) signal = np.zeros(n) atr = np.zeros(n) # MACD计算 ema12, ema26 = np.zeros(n), np.zeros(n) for i in range(1, n): ema12[i] = ema12[i - 1] * 11 / 13 + close[i] * 2 / 13 ema26[i] = ema26[i - 1] * 25 / 27 + close[i] * 2 / 27 macd[i] = ema12[i] - ema26[i] signal[i] = signal[i - 1] * 0.8 + macd[i] * 0.2 # ATR计算 for i in range(1, n): tr = max(high[i] - low[i], abs(high[i] - close[i - 1]), abs(low[i] - close[i - 1])) atr[i] = atr[i-1] * (atr_window-1)/atr_window + tr/atr_window return macd, signal, atr @jit(nopython=True) def generate_trading_signals(close, open_, high, low, volume, macd, signal, atr, threshold, volatility_window): """生成交易信号""" n = len(close) signals = np.zeros(n, dtype=np.bool_) for i in range(3, n): # 基础K线形态条件 is_red = close[i] > open_[i] upper_shadow = high[i] - max(close[i], open_[i]) lower_shadow = min(close[i], open_[i]) - low[i] body_size = abs(close[i] - open_[i]) cond1 = is_red and (high[i] / close[i - 1] > 1.005) cond2 = (body_size > upper_shadow) and (body_size > lower_shadow) cond3 = (high[i] / low[i] < 1.12) and (high[i] / open_[i] > 1.036) cond4 = close[i] < close[i - 1] * 1.10 # 排除涨停 # 技术指标条件 cond5 = atr[i] > np.mean(atr[i - 4:i + 1]) * 0.8 cond6 = (macd[i] - signal[i]) > (macd[i - 1] - signal[i - 1]) * 1.2 # 波动率条件 llv = np.min(low[max(0, i - volatility_window + 1):i + 1]) hhv = np.max(high[max(0, i - volatility_window + 1):i + 1]) cond7 = (llv / hhv) < threshold # 量能条件 vol_cond1 = volume[i] < np.mean(volume[max(0, i - 10):i]) vol_cond2 = volume[i] < np.min(volume[max(0, i - 20):i - 1]) * 3.5 signals[i] = cond1 & cond2 & cond3 & cond4 & cond5 & cond6 & cond7 & vol_cond1 & vol_cond2 return signals # ========== 市场状态判断 ========== def get_market_condition(index_data): """动态判断市场状态""" if len(index_data) < 60: return 'neutral' ma20 = index_data['close'].rolling(20).mean().iloc[-1] ma60 = index_data['close'].rolling(60).mean().iloc[-1] if pd.isna(ma20) or pd.isna(ma60): return 'neutral' if ma20 > ma60 * 1.05: return 'bull' elif ma20 < ma60 * 0.95: return 'bear' else: return 'neutral' # ========== 数据加载处理 ========== def load_index_data(index_path): """加载并预处理指数数据""" try: # 自动检测日期列名 df = pd.read_csv(index_path, sep='\t', nrows=0) date_col = 'date' if 'date' in df.columns else 'trade_date' index_data = pd.read_csv( index_path, sep='\t', usecols=[date_col, 'open', 'high', 'low', 'close', 'volume'], parse_dates=[date_col], date_parser=lambda x: pd.to_datetime(x, format='%Y%m%d') ) # index_data.rename(columns={date_col: 'trade_date'}, inplace=True) index_data.sort_values(date_col, inplace=True) logging.info( f"指数数据加载成功，时间范围: {index_data['trade_date'].min().date()} 至 {index_data['trade_date'].max().date()}") return index_data except Exception as e: logging.error(f"指数数据加载失败: {str(e)}") return None def process_stock_file(file_path, index_data): """处理单个股票文件""" try: # 加载并预处理数据 df = pd.read_csv(file_path, sep='\t', usecols=['trade_date', 'open', 'high', 'low', 'close', 'vol']) df = df.rename(columns={'vol': 'volume'}) df['trade_date'] = pd.to_datetime(df['trade_date'], format='%Y%m%d', errors='coerce') df = df.dropna(subset=['trade_date']).sort_values('trade_date') # 对齐指数时间范围 start_date = index_data['trade_date'].min() end_date = index_data['trade_date'].max() df = df[(df['trade_date'] >= start_date) & (df['trade_date'] <= end_date)] # if len(df) < StrategyConfig.MIN_TRADE_DAYS: # return None # 计算技术指标 close = df['close'].values.astype(np.float64) high = df['high'].values.astype(np.float64) low = df['low'].values.astype(np.float64) volume = df['volume'].values.astype(np.float64) macd, signal, atr = calculate_technical_indicators( close, high, low, volume, atr_window=ATR_WINDOW, volatility_window=VOLATILITY_WINDOW ) # # 获取市场状态 # market_condition = get_market_condition(index_data) # threshold = StrategyConfig.THRESHOLDS[market_condition] # 生成信号 signals = generate_trading_signals( close, df['open'].values, high, low, volume, macd, signal, atr, threshold=BULL_THRESHOLD, volatility_window=VOLATILITY_WINDOW ) df['signal'] = signals return os.path.basename(file_path).split('_')[0], df except Exception as e: logging.error(f"处理文件 {os.path.basename(file_path)} 失败: {str(e)}") return None # ========== 回测分析模块 ========== def backtest_strategy(all_data, index_data): """执行动态持仓周期回测""" results = [] for stock_code, data in all_data.items(): if data is None or 'signal' not in data.columns: continue signals = data[data['signal']] for idx in signals.index: # 动态获取市场状态 current_date = data.iloc[idx]['trade_date'] market_condition = get_market_condition(index_data) holding_days = HOLDING_DAYS_MAP.get(market_condition, 2) # 计算退出时间 exit_idx = idx + holding_days + 1 # 包含买入当天 if exit_idx >= len(data): continue # 计算收益 entry_price = data.loc[idx, 'close'] exit_prices = data.iloc[idx + 1:exit_idx]['close'] max_profit = (exit_prices.max() - entry_price) / entry_price max_loss = (exit_prices.min() - entry_price) / entry_price final_return = (exit_prices.iloc[-1] - entry_price) / entry_price results.append({ 'code': stock_code, 'date': current_date.strftime('%Y-%m-%d'), 'market': market_condition, 'holding_days': holding_days, 'return': final_return, 'max_profit': max_profit, 'max_loss': max_loss }) return pd.DataFrame(results) if results else pd.DataFrame() def analyze_results(results_df): """分析回测结果""" if results_df.empty: logging.warning("无有效交易记录") return # 基础统计 total_trades = len(results_df) annual_return = results_df['return'].mean() * 252 win_rate = len(results_df[results_df['return'] > 0]) / total_trades profit_factor = results_df[results_df['return'] > 0]['return'].mean() / \ abs(results_df[results_df['return'] < 0]['return'].mean()) print(f"\n策略表现汇总:") print(f"总交易次数: {total_trades}") print(f"年化收益率: {annual_return:.2%}") print(f"胜率: {win_rate:.2%}") print(f"盈亏比: {profit_factor:.2f}") # 分市场状态分析 if 'market' in results_df.columns: market_stats = results_df.groupby('market').agg({ 'return': ['mean', 'count'], 'holding_days': 'mean' }) print("\n分市场状态表现:") print(market_stats) # 可视化 plt.figure(figsize=(12, 5)) plt.subplot(121) results_df['return'].hist(bins=20, alpha=0.7) plt.title('收益率分布') plt.xlabel('收益率') plt.ylabel('频次') plt.subplot(122) if 'market' in results_df.columns: for condition, group in results_df.groupby('market'): plt.scatter(group['holding_days'], group['return'], alpha=0.5, label=condition) plt.legend() plt.axhline(0, color='red', linestyle='--') plt.title('持仓周期 vs 收益率') plt.xlabel('持仓天数') plt.ylabel('收益率') plt.tight_layout() plt.show() # ========== 主程序 ========== if __name__ == "__main__": # 配置路径 STOCK_DIR = '/day/' INDEX_PATH = '/index/000001.SH.txt' # 加载指数数据 logging.info("正在加载指数数据...") index_data = load_index_data(INDEX_PATH) if index_data is None: exit() # 并行处理个股数据 logging.info("正在加载个股数据...") stock_files = [os.path.join(STOCK_DIR, f) for f in os.listdir(STOCK_DIR) if f.endswith('.txt') and not any(kw in f for kw in ['ST', '*ST', '688'])] all_data = {} with ProcessPoolExecutor(max_workers=os.cpu_count()) as executor: futures = {executor.submit(process_stock_file, f, index_data): f for f in stock_files} for future in tqdm(as_completed(futures), total=len(futures)): result = future.result() if result: code, data, _ = result all_data[code] = data if not all_data: logging.error("没有加载到有效股票数据") exit() # 执行回测 logging.info("开始回测...") results_df = backtest_strategy(all_data, index_data) # 分析结果 analyze_results(results_df) # 保存结果 if not results_df.empty: results_df.to_csv('strategy_backtest_results.csv', index=False) logging.info("回测结果已保存至 strategy_backtest_results.csv")