import yfinance as yf import pandas as pd import numpy as np import matplotlib.pyplot as plt import seaborn as sns import core.logger as logging from config import OKX_MONITOR_CONFIG, MYSQL_CONFIG, WINDOW_SIZE from core.db.db_market_data import DBMarketData from core.db.db_huge_volume_data import DBHugeVolumeData from core.utils import timestamp_to_datetime, transform_date_time_to_timestamp # seaborn支持中文 plt.rcParams["font.family"] = ["SimHei"] logger = logging.logger class ORBStrategy: def __init__( self, initial_capital=25000, max_leverage=4, risk_per_trade=0.01, commission_per_share=0.0005, ): """ 初始化ORB策略参数 :param initial_capital: 初始账户资金（美元） :param max_leverage: 最大杠杆倍数（默认4倍，符合FINRA规定） :param risk_per_trade: 单次交易风险比例（默认1%） :param commission_per_share: 每股交易佣金（美元，默认0.0005） """ logger.info(f"初始化ORB策略参数：初始账户资金={initial_capital}，最大杠杆倍数={max_leverage}，单次交易风险比例={risk_per_trade}，每股交易佣金={commission_per_share}") self.initial_capital = initial_capital self.max_leverage = max_leverage self.risk_per_trade = risk_per_trade self.commission_per_share = commission_per_share self.data = None # 存储K线数据 self.trades = [] # 存储交易记录 self.equity_curve = None # 存储账户净值曲线 mysql_user = MYSQL_CONFIG.get("user", "xch") mysql_password = MYSQL_CONFIG.get("password", "") if not mysql_password: raise ValueError("MySQL password is not set") mysql_host = MYSQL_CONFIG.get("host", "localhost") mysql_port = MYSQL_CONFIG.get("port", 3306) mysql_database = MYSQL_CONFIG.get("database", "okx") self.db_url = f"mysql+pymysql://{mysql_user}:{mysql_password}@{mysql_host}:{mysql_port}/{mysql_database}" self.db_market_data = DBMarketData(self.db_url) def fetch_intraday_data(self, symbol, start_date, end_date, interval="5m"): """ 获取日内5分钟K线数据（需yfinance支持，部分数据可能有延迟） :param ticker: 股票代码（如QQQ、TQQQ） :param start_date: 起始日期（格式：YYYY-MM-DD） :param end_date: 结束日期（格式：YYYY-MM-DD） :param interval: K线周期（默认5分钟） """ logger.info(f"开始获取{symbol}数据：{start_date}至{end_date}，间隔{interval}") # data = yf.download( # symbol, start=start_date, end=end_date, interval=interval, progress=False # ) data = self.db_market_data.query_market_data_by_symbol_bar( symbol, interval, start=start_date, end=end_date ) data = pd.DataFrame(data) data.sort_values(by="date_time", inplace=True) # 保留核心列：开盘价、最高价、最低价、收盘价、成交量 data["Open"] = data["open"] data["High"] = data["high"] data["Low"] = data["low"] data["Close"] = data["close"] data["Volume"] = data["volume"] # 将data["date_time"]从字符串类型转换为日期 data["date_time"] = pd.to_datetime(data["date_time"]) # data["Date"]为日期，不包括时分秒，即date_time如果是2025-01-01 10:00:00，则Date为2025-01-01 data["Date"] = data["date_time"].dt.date # 将Date转换为datetime64[ns]类型以确保类型一致 data["Date"] = pd.to_datetime(data["Date"]) self.data = data[["Date", "date_time", "Open", "High", "Low", "Close", "Volume"]].copy() logger.info(f"成功获取{symbol}数据：{len(self.data)}根{interval}K线") def calculate_shares(self, account_value, entry_price, stop_price): """ 根据ORB公式计算交易股数 :param account_value: 当前账户价值（美元） :param entry_price: 交易entry价格（第二根5分钟K线开盘价） :param stop_price: 止损价格（多头=第一根K线最低价，空头=第一根K线最高价） :return: 整数股数（Shares） """ logger.info(f"开始计算交易股数：账户价值={account_value}，entry价格={entry_price}，止损价格={stop_price}") # 计算单交易风险金额（$R） risk_per_trade_dollar = abs(entry_price - stop_price) # 风险金额取绝对值 if risk_per_trade_dollar <= 0: return 0 # 无风险时不交易 # 公式1：基于风险预算的最大股数（风险控制优先） shares_risk = (account_value * self.risk_per_trade) / risk_per_trade_dollar # 公式2：基于杠杆限制的最大股数（杠杆约束） shares_leverage = (self.max_leverage * account_value) / entry_price # 取两者最小值（满足风险和杠杆双重约束） max_shares = min(shares_risk, shares_leverage) # 扣除佣金影响（简化计算：假设佣金从可用资金中扣除） commission_cost = max_shares * self.commission_per_share if (account_value - commission_cost) < 0: return 0 # 扣除佣金后资金不足，不交易 return int(max_shares) # 股数取整 def generate_orb_signals(self): """ 生成ORB策略信号（每日仅1次交易机会） - 第一根5分钟K线：确定开盘区间（High1, Low1） - 第二根5分钟K线：根据第一根K线方向生成多空信号 """ logger.info("开始生成ORB策略信号") if self.data is None: raise ValueError("请先调用fetch_intraday_data获取数据") signals = [] # 按日期分组处理每日数据 for date, daily_data in self.data.groupby("Date"): daily_data = daily_data.sort_index() # 按时间排序 if len(daily_data) < 2: continue # 当日K线不足2根，跳过 # 第一根5分钟K线（开盘区间） first_candle = daily_data.iloc[0] high1 = first_candle["High"] low1 = first_candle["Low"] open1 = first_candle["Open"] close1 = first_candle["Close"] # 第二根5分钟K线（entry信号） second_candle = daily_data.iloc[1] entry_price = second_candle["Open"] # entry价格=第二根K线开盘价 entry_time = second_candle.date_time # entry时间 # 生成信号：第一根K线方向决定多空（排除十字星：open1 == close1） if open1 < close1: # 第一根K线收涨→多头信号 signal = "Long" stop_price = low1 # 多头止损=第一根K线最低价 elif open1 > close1: # 第一根K线收跌→空头信号 signal = "Short" stop_price = high1 # 空头止损=第一根K线最高价 else: # 十字星→无信号 signal = "None" stop_price = None signals.append( { "Date": date, "EntryTime": entry_time, "Signal": signal, "EntryPrice": entry_price, "StopPrice": stop_price, "High1": high1, "Low1": low1, } ) # 将信号合并到原始数据 signals_df = pd.DataFrame(signals) # 确保Date列类型一致，将Date转换为datetime64[ns]类型 signals_df['Date'] = pd.to_datetime(signals_df['Date']) # 使用merge而不是join来合并数据,根据signals_df的EntryTime与self.data的date_time进行匹配 # TODO: 这里需要优化 self.data = self.data.merge(signals_df, left_on="date_time", right_on="EntryTime", how="left") # 将Date_x和Date_y合并为Date self.data["Date"] = self.data["Date_x"].combine_first(self.data["Date_y"]) # 删除Date_x和Date_y self.data.drop(columns=["Date_x", "Date_y"], inplace=True) logger.info( f"生成信号完成：共{len(signals_df)}个交易日，其中多头{sum(signals_df['Signal']=='Long')}次，空头{sum(signals_df['Signal']=='Short')}次" ) def backtest(self, profit_target_multiple=10): """ 回测ORB策略 :param profit_target_multiple: 盈利目标倍数（默认10倍$R，即10R） """ logger.info(f"开始回测ORB策略：盈利目标倍数={profit_target_multiple}") if "Signal" not in self.data.columns: raise ValueError("请先调用generate_orb_signals生成策略信号") account_value = self.initial_capital # 初始账户价值 current_position = None # 当前持仓（None=空仓，Long/Short=持仓） equity_history = [account_value] # 净值历史 trade_id = 0 # 交易ID # 按时间遍历数据（每日仅处理第二根K线后的信号） for date, daily_data in self.data.groupby("Date"): daily_data = daily_data.sort_index() if len(daily_data) < 2: continue # 获取当日信号（第二根K线的信号） signal_row = ( daily_data[~pd.isna(daily_data["Signal"])].iloc[0] if sum(~pd.isna(daily_data["Signal"])) > 0 else None ) if signal_row is None: # 无信号→当日不交易，净值保持不变 equity_history.append(account_value) continue # 提取信号参数 signal = signal_row["Signal"] if pd.isna(signal): continue entry_price = signal_row["EntryPrice"] stop_price = signal_row["StopPrice"] high1 = signal_row["High1"] low1 = signal_row["Low1"] risk_assumed = abs(entry_price - stop_price) # 计算$R profit_target = ( entry_price + (risk_assumed * profit_target_multiple) if signal == "Long" else entry_price - (risk_assumed * profit_target_multiple) ) # 计算交易股数 shares = self.calculate_shares(account_value, entry_price, stop_price) if shares == 0: # 股数为0→不交易 equity_history.append(account_value) continue # 计算佣金（买入/卖出各收一次） total_commission = shares * self.commission_per_share * 2 # 往返佣金 # 模拟日内持仓：寻找止损/止盈触发点，或当日收盘平仓 daily_prices = daily_data[ daily_data.date_time > signal_row.date_time ] # 从entry时间开始遍历 exit_price = None exit_time = None exit_reason = None for idx, (time, row) in enumerate(daily_prices.iterrows()): high = row["High"] low = row["Low"] close = row["Close"] # 检查止损/止盈条件 if signal == "Long": # 多头：跌破止损→止损；突破止盈→止盈 if low <= stop_price: exit_price = stop_price exit_reason = "Stop Loss" exit_time = time break elif high >= profit_target: exit_price = profit_target exit_reason = "Profit Target (10R)" exit_time = time break elif signal == "Short": # 空头：突破止损→止损；跌破止盈→止盈 if high >= stop_price: exit_price = stop_price exit_reason = "Stop Loss" exit_time = time break elif low <= profit_target: exit_price = profit_target exit_reason = "Profit Target (10R)" exit_time = time break # 若未触发止损/止盈，当日收盘平仓 if exit_price is None: exit_price = daily_prices.iloc[-1]["Close"] exit_reason = "End of Day (EoD)" exit_time = daily_prices.iloc[-1].date_time # 计算盈亏 if signal == "Long": profit_loss = (exit_price - entry_price) * shares - total_commission else: # Short profit_loss = (entry_price - exit_price) * shares - total_commission # 更新账户价值 account_value += profit_loss account_value = max(account_value, 0) # 账户价值不能为负 # 记录交易 self.trades.append( { "TradeID": trade_id, "Date": date, "Signal": signal, "EntryTime": signal_row.date_time, "EntryPrice": entry_price, "ExitTime": exit_time, "ExitPrice": exit_price, "Shares": shares, "RiskAssumed": risk_assumed, "ProfitLoss": profit_loss, "ExitReason": exit_reason, "AccountValueAfter": account_value, } ) # 记录净值 equity_history.append(account_value) trade_id += 1 # 生成净值曲线 self.equity_curve = pd.Series( equity_history, index=pd.date_range( start=self.data.index[0].date(), periods=len(equity_history), freq="D" ), ) # 输出回测结果 trades_df = pd.DataFrame(self.trades) total_return = ( (account_value - self.initial_capital) / self.initial_capital * 100 ) win_rate = ( (trades_df["ProfitLoss"] > 0).sum() / len(trades_df) * 100 if len(trades_df) > 0 else 0 ) logger.info("\n" + "=" * 50) logger.info("ORB策略回测结果") logger.info("=" * 50) logger.info(f"初始资金：${self.initial_capital:,.2f}") logger.info(f"最终资金：${account_value:,.2f}") logger.info(f"总收益率：{total_return:.2f}%") logger.info(f"总交易次数：{len(trades_df)}") logger.info(f"胜率：{win_rate:.2f}%") if len(trades_df) > 0: logger.info(f"平均每笔盈亏：${trades_df['ProfitLoss'].mean():.2f}") logger.info(f"最大单笔盈利：${trades_df['ProfitLoss'].max():.2f}") logger.info(f"最大单笔亏损：${trades_df['ProfitLoss'].min():.2f}") def plot_equity_curve(self): """ 绘制账户净值曲线 """ logger.info("开始绘制账户净值曲线") if self.equity_curve is None: raise ValueError("请先调用backtest进行回测") # seaborn风格设置 sns.set_theme(style="whitegrid") # plt.rcParams['font.family'] = "SimHei" plt.rcParams["font.sans-serif"] = ["SimHei"] # 也可直接用字体名 plt.rcParams["font.size"] = 11 # 设置字体大小 plt.rcParams["axes.unicode_minus"] = False # 解决负号显示问题 plt.figure(figsize=(12, 6)) plt.plot( self.equity_curve.index, self.equity_curve.values, label="ORB策略净值", color="blue", ) plt.axhline( y=self.initial_capital, color="red", linestyle="--", label="初始资金" ) plt.title("ORB策略账户净值曲线", fontsize=14) plt.xlabel("日期", fontsize=12) plt.ylabel("账户价值（美元）", fontsize=12) plt.legend() plt.grid(True, alpha=0.3) plt.show() # ------------------- 策略示例：回测QQQ的ORB策略（2016-2023） ------------------- if __name__ == "__main__": # 初始化ORB策略 orb_strategy = ORBStrategy( initial_capital=25000, max_leverage=4, risk_per_trade=0.01, commission_per_share=0.0005, ) # 1. 获取QQQ的5分钟日内数据（2024-2025，注意：yfinance免费版可能限制历史日内数据，建议用专业数据源） orb_strategy.fetch_intraday_data( symbol="ETH-USDT", start_date="2025-05-15", end_date="2025-08-20", interval="5m" ) # 2. 生成ORB策略信号 orb_strategy.generate_orb_signals() # 3. 回测策略（盈利目标10R） orb_strategy.backtest(profit_target_multiple=10) # 4. 绘制净值曲线 orb_strategy.plot_equity_curve()