Exchange-traded fund pair-trading strategies using autocorrelation-based mean reversion

A simple pair-trading strategy based on autocorrelation can be profitable if applied over non-standard time frames. They observe that, for such a strategy, exchange-traded funds are more suitable than stocks and allow higher information ratios to be achieved

In this article, a simple mean reversion strategy is applied to both constituent shares of the S&P 100 index and the 100 most liquid exchange-traded funds (ETFs). In the first step, we form the pairs from two ETFs or two shares with the conditional correlation above the threshold of 0.8. This leaves us with pairs consisting of two instruments with similar recent behaviour. In the second step, we eliminate from the selection the pairs with the previous day’s normalised return smaller than one