Statistical Arbitrage and Market Efficiency: Enhanced Theory, Robust Tests and Further Applications

Abstract

Statistical arbitrage enables tests of market efficiency which circumvent the joint-hypotheses dilemma. This paper makes several contributions to the statistical arbitrage framework. First, we enlarge the set of statistical arbitrage opportunities in Hogan, Jarrow, Teo, and Warachka (2004) to avoid penalizing incremental trading profits with positive deviations from their expected value. Second, we provide a statistical methodology to remedy the lack of consistency and statistical power in their Bonferroni approach. In addition, this procedure allows for autocorrelation and non-normality in trading profits. Third, we apply our tests to a wide range of trading strategies based on stock momentum, stock value, stock liquidity, and industry momentum. Over 50% of these strategies are found to violate market efficiency. We also identify dominant trading strategies which converge to arbitrage most rapidly.