Online Structural Break Detection for Pairs Trading Using Wavelet Transform and Hybrid Deep Learning Model

Shen-Hang Huang, Wen-Yueh Shih, Jing-You Lu, Hao-Han Chang, Chao-Hsien Chu, Jun-Zhe Wang, Jiun-Long Huang, Tian-Shyr Dai

Abstract: Pairs trading is a statistical arbitrage strategy which first monitors two stocks whose prices are cointegrated, and then makes arbitrage when the prices of these two stocks get non-conintegrated. The phenomenon that the cointegration relationship between two stocks does not exist any longer is called structural break, and detecting structural breaks is important for pairs trading. To detect structural breaks as soon as possible, we propose in this paper a hybrid deep learning model using both frequency-domain and time-domain features to detect structural breaks. To evaluate the performance of traditional methods and our model, we collect the historical tick data of the top 150 companies from Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) for experiments. Experimental results show that our proposed method is able to detect structural breaks more accurately than tradition methods.