Many-to-Many Pair Trading

Yingying Wang, Xiaodong Li, Pangjing Wu, Haoran Xie

Abstract: Pair trading is a market neutral strategy commonly used in hedge funds. There are two main phases of the pair trading: pair formation phase and trading phase. Finding profitable stock pairs during the pair formation phase is an important issue. Previous one-to-one pair trading mostly limits the pair formation phase to a small number of stocks. In this paper, we make an innovative improvement by extending one-to-one pair trading to many-to-many pair trading. We propose a framework that involves association rule algorithms and the OPTICS clustering algorithm, and uses the bipartite graph partition algorithm to form many-to-many pairs. Experimental results show that the framework proposed in this paper is effective in selecting many-to-many pairs for pair trading and more trading opportunities are obtained compared with traditional pair trading.