Pair Trading: Random Weight Approach

Abstract

Pairs trading are standard approaches for statistical arbitrage detection. The logic behind pair trading approach is to construct a portfolio of two financial assets with special weights where this portfolio has zero value in time zero and creates positive value with a high probability in a certain future. In this paper, a random weight approach is introduced for construction of pairs trading strategies. Random weight approach is a method at which the weights of each asset comes from a statistical distribution. In this note, the beta distribution is selected. Under this circumstance, first, condition for market neutrality of portfolio is considered then, the probability of attaining the profit is maximized. The portfolio contains long position in random units of first asset and random units of another asset in short position. Strategies are given and parameters selections for maximizing the probability of profit are proposed.