ABSTRACT

Quantitative trading in oil-based markets is investigated over 2003-2010, with a focus on WTI, Brent, heating oil and gas oil. A total of 861 spreads are considered. A novel optimal statistical arbitrage trading model is applied, with generalised stepwise procedures controlling for data snooping bias. Aggregating upward and downward mean-reversion, protable strategies are identied with Sharpe ratios greater than 2 in many instances. For the top categories, average daily returns range from 0.07 to 0.55%, with trade lengths of 9-55 days. A collapse in the number of protable trading strategies is seen in 2008. Robustness to varying transactions costs is examined.