Optimal Pairs Trading with Dynamic Mean-Variance

Pairs trading is one of convergence trades, which simultaneously buy relatively underpriced assets and sell relatively overpriced assets to exploit temporary mispricing. This paper studies time consistency optimal pairs trading strategies under both "symmetric'' and "non-symmetric'' trading constraints. We provide fully analytical trading strategies under Mean-Variance (MV) framework by assuming the price spread of a pair of related securities follows an Ornstein-Uhlenbeck (OU) process. Model calibration is discussed and real-data experiments based on pairs of stocks and futures traded on China security markets are presented to illustrate the practical implementation of the proposed approach.