【期货量化策略】期货量化交易策略跨品种套利（Python量化）-深圳市維司達科技有限公司

一、前言

跨品种套利基本原理
相关性分析
价差计算
套利策略实现
风险控制

二、为什么选择天勤量化（TqSdk）

功能	说明
多品种数据	支持同时获取多个品种数据
实时行情	支持实时行情数据
数据同步	支持多品种数据同步
快速执行	支持快速下单

安装方法：

pipinstalltqsdk pandas numpy

三、跨品种套利基础

3.1 套利原理

原理	说明
相关性	相关品种价格相关
价差波动	价差在合理范围波动
回归均值	价差会回归均值
套利机会	价差偏离时套利

3.2 相关品种

类型	说明	示例
产业链	上下游关系	螺纹钢-铁矿石
替代品	可替代品种	豆粕-菜粕
互补品	互补关系	焦炭-焦煤
同板块	同板块品种	有色金属

四、相关性分析

4.1 价格相关性

#!/usr/bin/env python# -*- coding: utf-8 -*-""" 功能：跨品种套利 说明：本代码仅供学习参考 """fromtqsdkimportTqApi,TqAuthimportpandasaspdimportnumpyasnpdefcalculate_correlation(klines1,klines2,window=60):"""计算价格相关性"""returns1=klines1['close'].pct_change()returns2=klines2['close'].pct_change()correlation=returns1.rolling(window).corr(returns2)returncorrelation# 使用示例api=TqApi(auth=TqAuth("快期账户","快期密码"))klines1=api.get_kline_serial("SHFE.rb2510",3600,500)klines2=api.get_kline_serial("DCE.i2510",3600,500)api.wait_update()correlation=calculate_correlation(klines1,klines2)print(f"当前相关性:{correlation.iloc[-1]:.4f}")api.close()

4.2 价差相关性

defcalculate_spread_correlation(klines1,klines2):"""计算价差相关性"""price1=klines1['close']price2=klines2['close']# 计算价差spread=price1/price2# 计算价差变化的相关性spread_change=spread.pct_change()returnspread_change

五、价差计算

5.1 价差比率

defcalculate_spread_ratio(klines1,klines2,window=20):"""计算价差比率"""price1=klines1['close']price2=klines2['close']# 价差比率spread_ratio=price1/price2# 历史均值mean_ratio=spread_ratio.rolling(window).mean()std_ratio=spread_ratio.rolling(window).std()# 当前偏离current_ratio=spread_ratio.iloc[-1]z_score=(current_ratio-mean_ratio.iloc[-1])/std_ratio.iloc[-1]return{'spread_ratio':current_ratio,'mean_ratio':mean_ratio.iloc[-1],'z_score':z_score}

5.2 价差波动

defcalculate_spread_volatility(klines1,klines2,window=20):"""计算价差波动"""price1=klines1['close']price2=klines2['close']spread=price1/price2 spread_returns=spread.pct_change()volatility=spread_returns.rolling(window).std()returnvolatility

六、套利策略实现

6.1 价差套利策略

defspread_arbitrage_strategy(api,symbol1,symbol2,klines1,klines2,threshold=2):"""价差套利策略"""spread_info=calculate_spread_ratio(klines1,klines2)z_score=spread_info['z_score']# Z-score过大，做空价差ifz_score>threshold:api.insert_order(symbol1,"SELL","OPEN",1)api.insert_order(symbol2,"BUY","OPEN",1)return1# Z-score过小，做多价差elifz_score<-threshold:api.insert_order(symbol1,"BUY","OPEN",1)api.insert_order(symbol2,"SELL","OPEN",1)return-1return0

6.2 对冲比率套利

fromsklearn.linear_modelimportLinearRegressiondefhedge_ratio_arbitrage(api,symbol1,symbol2,klines1,klines2,threshold=2):"""对冲比率套利"""price1=klines1['close']price2=klines2['close']# 估计对冲比率X=price1.values.reshape(-1,1)y=price2.values model=LinearRegression()model.fit(X,y)hedge_ratio=model.coef_[0]# 计算价差spread=price2-hedge_ratio*price1 spread_mean=spread.rolling(20).mean()spread_std=spread.rolling(20).std()current_spread=spread.iloc[-1]z_score=(current_spread-spread_mean.iloc[-1])/spread_std.iloc[-1]# 执行套利ifz_score>threshold:api.insert_order(symbol1,"SELL","OPEN",1)api.insert_order(symbol2,"BUY","OPEN",int(hedge_ratio))return1elifz_score<-threshold:api.insert_order(symbol1,"BUY","OPEN",1)api.insert_order(symbol2,"SELL","OPEN",int(hedge_ratio))return-1return0

七、风险控制

7.1 相关性监控

defmonitor_correlation(klines1,klines2,min_correlation=0.7):"""监控相关性"""correlation=calculate_correlation(klines1,klines2)current_corr=correlation.iloc[-1]ifcurrent_corr<min_correlation:returnFalse,"相关性过低"returnTrue,"相关性正常"

7.2 价差止损

defset_spread_stop_loss(entry_spread,current_spread,stop_loss_std=3):"""设置价差止损"""spread_change=abs(current_spread-entry_spread)spread_std=abs(entry_spread)*0.1# 简化处理ifspread_change>stop_loss_std*spread_std:returnTrue# 触发止损returnFalse