关键词: 配对交易, 最优阈值, 部分协整, 遗传算法

Abstract: Pairs trading is a kind of statistical arbitrage strategy through spread arbitrage. Its main research content is to find the paired risk assets and the optimal threshold. Partial cointegration method can effectively improve the number of paired risk assets and transaction frequency. This paper integrates genetic algorithm into partial cointegration pairs trading method, uses genetic algorithm to obtain optimal threshold, introduces sliding window to detect partial cointegration of paired stocks, and adopts two-way trading mechanism to grasp more trading opportunities. This trading method overcomes some problems of partial cointegration methods in terms of threshold selection roughness, parameter failure, and loss of trading opportunities. Both methods are sufficiently tested under S&P 500 and various market conditions of Chinese stock market. Results show that the performance of the method presented in this paper is obviously better than that of the original method in various markets and the returns are robust.

Key words: Pairs Trading, Optimal Threshold, Partial Cointegration, Genetic Algorithm