Futures Spread Mean Reversion

Article Strategy library · Author: Chaos

Summary

This example describes a mean-reversion approach to the price spread between near- and far-month index futures contracts. On 15-minute bars, it calculates the difference between the contracts’ closing prices over an 80-bar window, then estimates the spread’s mean and standard deviation. The current spread is compared with upper and lower bands set two standard deviations from that mean. A move above the upper band triggers a short spread by selling the near contract and buying the far one; a move below the lower band triggers the reverse position. Both legs are closed when the spread returns to the mean.

The code specifies a 20-lot position and a backtest interval from December 2020 to January 2021, but provides no performance results. It also omits explicit stop-loss, transaction-cost, and leg-risk handling. The position state is updated when target positions are submitted, so the example does not show checks that both orders filled as intended. Its fixed contract pair and short sample make it an implementation sketch rather than evidence that the spread is reliably stationary or profitable.

Key ideas

The strategy models the price difference between near- and far-month futures as a mean-reverting spread.
It enters a short spread above the rolling mean plus two standard deviations and a long spread below the corresponding lower band.
Each spread position is closed when the current spread returns to its rolling mean.
The example uses 15-minute bars, an 80-bar calculation window, and a fixed contract pair.
The supplied backtest dates have no accompanying performance results or detailed execution-risk analysis.

Tags

Futures Arbitrage Mean reversion Statistics Execution Backtesting

Full text

# stock-future-cross-period-spread


# stock-future-cross-period-spread









## Source (Apache-2.0)

```python
#!/usr/bin/env python
# -*- coding: utf-8 -*-
__author__ = "Chaos"

from datetime import date

import numpy as np
from tqsdk import TqApi, TqAuth, TqBacktest, TargetPosTask, BacktestFinished

# 参数设置
NEAR_CONTRACT = "CFFEX.IH2101"  # 近月合约
FAR_CONTRACT = "CFFEX.IH2102"  # 远月合约
K = 2  # 标准差倍数
WINDOW = 80  # 计算窗口
LOTS = 20  # 交易手数
START_DATE = date(2020, 12, 21)  # 回测开始日期
END_DATE = date(2021, 1, 15)  # 回测结束日期

# 创建API实例
api = TqApi(backtest=TqBacktest(start_dt=START_DATE, end_dt=END_DATE),
            auth=TqAuth("快期账户", "快期密码"))

# 订阅行情
near_quote = api.get_quote(NEAR_CONTRACT)
far_quote = api.get_quote(FAR_CONTRACT)
near_klines = api.get_kline_serial(NEAR_CONTRACT, 15 * 60, WINDOW * 2)  # 15分钟K线
far_klines = api.get_kline_serial(FAR_CONTRACT, 15 * 60, WINDOW * 2)

# 创建目标持仓任务
near_pos = TargetPosTask(api, NEAR_CONTRACT)
far_pos = TargetPosTask(api, FAR_CONTRACT)

# 持仓状态: 0-无持仓, 1-多价差(买近卖远), -1-空价差(卖近买远)
position_state = 0

print(f"策略启动: {NEAR_CONTRACT}-{FAR_CONTRACT} 跨期套利")
print(f"参数设置: K={K}倍标准差, 窗口={WINDOW}, 交易手数={LOTS}手")

try:
    while True:
        api.wait_update()

        # 检查K线是否更新
        if api.is_changing(near_klines) or api.is_changing(far_klines):
            # 确保有足够的数据
            if len(near_klines) < WINDOW or len(far_klines) < WINDOW:
                continue

            # 计算价差指标
            near_close = near_klines.close.iloc[-WINDOW:]
            far_close = far_klines.close.iloc[-WINDOW:]
            spread = near_close - far_close

            # 计算均值和标准差
            mean = np.mean(spread)
            std = np.std(spread)
            current_spread = near_quote.last_price - far_quote.last_price

            # 计算上下边界
            upper_bound = mean + K * std
            lower_bound = mean - K * std

            print(f"价差: {current_spread:.2f}, 均值: {mean:.2f}, "
                  f"上界: {upper_bound:.2f}, 下界: {lower_bound:.2f}")

            # 交易逻辑
            if position_state == 0:  # 无持仓状态
                if current_spread > upper_bound:  # 做空价差(卖近买远)
                    near_pos.set_target_volume(-LOTS)
                    far_pos.set_target_volume(LOTS)
                    position_state = -1
                    print(f"开仓: 卖出{LOTS}手{NEAR_CONTRACT}, 买入{LOTS}手{FAR_CONTRACT}")

                elif current_spread < lower_bound:  # 做多价差(买近卖远)
                    near_pos.set_target_volume(LOTS)
                    far_pos.set_target_volume(-LOTS)
                    position_state = 1
                    print(f"开仓: 买入{LOTS}手{NEAR_CONTRACT}, 卖出{LOTS}手{FAR_CONTRACT}")

            elif position_state == 1:  # 持有多价差
                if current_spread >= mean:  # 平仓获利
                    near_pos.set_target_volume(0)
                    far_pos.set_target_volume(0)
                    position_state = 0
                    print("平仓: 价差回到均值，平仓获利")

            elif position_state == -1:  # 持有空价差
                if current_spread <= mean:  # 平仓获利
                    near_pos.set_target_volume(0)
                    far_pos.set_target_volume(0)
                    position_state = 0
                    print("平仓: 价差回到均值，平仓获利")

except BacktestFinished as e:
    api.close()
    print("回测结束")


```

Shown in full with attribution under the source's licence. Licence: Apache-2.0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.

Cross-Expiry Futures Spread Mean Reversion with Standard Deviation Bands

Summary

Key ideas

Tags