Selectivity, Information And The Return To Futures Trading

Abstract

One of the more controversial issues in modern financial economics and in futures trading in particular is whether traders have the ability to earn returns above what they could with a buy-and-hold strategy.  The weight of evidence in support of Martingale price movements generally has been considered to be evidence that the expected value of “trading returns” is zero.  This paper shows, however, that, when the contingent claims in a futures contract are taken into account in defining return, expected trading returns may not be zero, even if prices follow the Martingale pattern.  We also point out that, if a sample of trades is representative of some trading strategy with its corresponding trading information, the impact of that strategy on a trader’s expected return can be represented by a  probability model of the strategy’s success.  This results because, to be successful, a trading strategy must select trades nonrandomly.  Using these results, the paper specifies a model of the expected return of an arbitrary trading strategy.  As an illustration, the model is estimated for an artificially constructed strategy in gold futures that imitates what the industry claims is the epitome of futures trading performance – many small losses more than offset by a few big gains.  Statistical tests based on the estimation of this model support the characterization of returns being due to the trading strategy using information to nonrandomly select trades.