松鼠宽客

专享策略22 | 跨品种套利策略：基于范围过滤器的量化交易方法

松鼠Quant

2025-06-03

量化策略开发，高质量社群，交易思路分享等相关内容

工具推荐

『正文』

策略概述

跨品种套利是一种常见的量化交易策略，它通过捕捉两个相关品种之间的价差变化来获利。本文介绍的vip22策略是一种基于范围过滤器(Range Filter)的跨品种套利策略。

该策略的核心思想是：

计算两个品种价格比率的动态范围
当比率突破上轨时做多价差(买焦炭/卖焦煤)
当比率突破下轨时做空价差(卖焦炭/买焦煤)
使用动态止损机制控制风险

数学原理

1. 平滑范围(Smooth Range)计算

平滑范围是策略的核心指标，用于确定价差的动态边界。它通过两层指数移动平均(EMA)计算得出：

第一层EMA：

第二层EMA：

最终平滑范围：

2. 范围过滤器(RanAge Filter)

范围过滤器用于生成动态的交易信号通道：

3. 趋势计数

策略还计算了向上和向下的连续趋势计数：

策略实现

核心代码结构

def vip22_strategy_optimized(api: StrategyAPI): """ vip22跨品种套利策略主函数（高度优化版本）基于价差的范围过滤器进行交易决策使用缓存和增量计算大幅提高性能 """ # 1. 获取参数和数据ifnot api.require_data_sources(2):return
# 2. 计算技术指标（增量方式）iflen(g_state['filt_history']) < 2:return
curr_filt = g_state['filt_history'][-1] curr_upward = g_state['upward_history'][-1] curr_downward = g_state['downward_history'][-1]curr_filt = g_state['filt_history'][-1] curr_upward = g_state['upward_history'][-1] curr_downward = g_state['downward_history'][-1]
# 3. 计算交易条件 long_cond = (curr_source > curr_filt and (curr_source != prev_source) and curr_upward > g_state['x_filter']) short_cond = (curr_source < curr_filt and (curr_source != prev_source) and curr_downward > g_state['x_filter'])
# 4. 执行交易逻辑 execute_trading_logic(api, pos_data0, pos_data1, long_cond, short_cond, curr_source, prev_source, chigh, clow, lots_0, lots_1, bar_idx)

增量计算优化

为了提高性能，策略采用了增量计算技术：

def smooth_range_incremental(price_diff, period, s_range, ema_state): """ 增量计算SmoothRange函数 """ alpha1 = 2.0 / (period + 1) alpha2 = 2.0 / (period * 2 - 1 + 1) # 第一层EMA if ema_state['avrng1'] is None: ema_state['avrng1'] = price_diff else: ema_state['avrng1'] = alpha1 * price_diff + (1 - alpha1) * ema_state['avrng1'] # 第二层EMA if ema_state['smooth_avrng1'] is None: ema_state['smooth_avrng1'] = ema_state['avrng1'] else: ema_state['smooth_avrng1'] = alpha2 * ema_state['avrng1'] + (1 - alpha2) * ema_state['smooth_avrng1'] return ema_state['smooth_avrng1'] * s_range

交易逻辑实现

def execute_trading_logic(api, pos_data0, pos_data1, long_cond, short_cond, curr_source, prev_source, chigh, clow, lots_0, lots_1, bar_idx): """ 执行交易逻辑的辅助函数 - 套利策略同开同平版本 """ # 检查是否有持仓 has_position = (pos_data0 != 0or pos_data1 != 0) ifnot has_position: # 无持仓时，同时开仓 if long_cond: # 价差向上突破，做多价差 api.buy(volume=lots_0, order_type='next_bar_open', index=0) api.sellshort(volume=lots_1, order_type='next_bar_open', index=1) else: # 有持仓时，检查平仓条件 should_close = check_exit_conditions(pos_data0, pos_data1, curr_source, prev_source, chigh, clow) if should_close: # 平仓两个品种 close_positions(api, pos_data0, pos_data1)

策略优势