Multi-filter Optimization for Pair Trading(中文)
配對交易(Pair Trading)，或稱作統計套利(Statistical Arbitrage)，最早是1980年代由摩根士丹利的研究人員 Nunzio Tartaglia所提出的一種交易策略，該方法主要是利用統計模型量化市場中的資料，找出具有相同趨勢或高度相關的兩個標的進行對沖，並在偏離均衡時進場以獲取中間的價差(spread)，因為該策略對沖銷掉系統性風險，可以確保無論在多頭或空頭時都能穩定獲利，是一種市場中立策略(Market Neutral Strategy)。
Pair trading is a market neutral strategy, its concept is using multiple high connected symbol which have high linkage between prices, by long and short portfolio in the same time to eliminate systematic risk. In this article, we refer to previous research, include Bertram(2010), Engle and Granger (1987) , Sarmento and Horta (2020), and Vidyamurthy (2004), using pair trading to backtest market ETF in America from 2008 to 2022.
We apply Engle and Granger (1987) two step cointegration method for pair trading, moreover, we improve some concept and backtest method through past literature including OU process, Momentum, Machine Learning, Hurst exponent etc. Our data is comes from U.S. ETF Channel, we choose the top 20 largest trading volume symbol in each segment as our symbol pool, using three years rolling window for training data and one year for testing data.
Figure 1 is the performance of cumulated return, we can see that, from 2010 to 2022, the cumulated return is very well, only have slightly drawn down in 2019.
Figure 2 is the distribution of return, we can observe that the return distribution is a negative skewness, which means the mass of the distribution is concentrated on the right of the figure.
Figure 3 is the Statistics of the strategies in each year, we can see that, there are significant positive return from 2010 to 2018, only have negative return in 2019. Also, considering 2020 and 2022 are surfing international crisis(Covid-19), this strategy still have positive return and prevent the potential lose when market have panic selling.
In conclude, although pair trading has been invented in 1980s, through the optimization method from recent research, it still have potential profit.
Contact us at info@gammaparadigm.com for the full report.
David Wang(Tomo)
Quantitative Research Intern, Quantitative Finance, Gamma Paradigm
