Does technical analysis work for crude oil futures trading? In their August 2016 paper entitled “Performance of Technical Trading Rules: Evidence from the Crude Oil Market”, Ioannis Psaradellis, Jason Laws, Athanasios Pantelous and Georgios Sermpinis investigate the profitability of a wide range technical trading rules applied to West Texas Intermediate (WTI) light sweet crude oil futures and the United States Oil (USO) fund, which holds front-month futures contracts. They consider 7,846 trading rules grouped into five families (filter rules, moving averages, support and resistance rules, channel breakouts and on-balance volume averages) used on daily prices. They assume these rules earn the risk-free rate when neutral. They measure performance during four subperiods (April 2007-May 2009; June 2009-March 2011; April 2011-July 2013; August 2013-December 2015), reflecting bullish or bearish and contango or backwardation subperiods. They focus on average return, Sharpe ratio and Calmar ratio as key performance statistics. They begin with in-sample tests and progressively account for trading frictions (one-way 0.033% for futures and 0.05% for USO), data snooping bias (via two methods) and out-of-sample rule identification. Using daily prices for the specified assets during April 2006 through December 2015, they find that:
- For gross performance of rules tested in-sample:
- Percentages of rules outperforming passive continuous holdings in-sample on a gross basis varies considerably by subperiod:
- For average return and Sharpe ratio, percentages range from just 10% for futures during the third subperiod to 65% for futures during the fourth. Outperformance concentrates in the first and fourth subperiods.
- For Calmar ratio, percentages are comparatively low, ranging from 1% for futures during the third subperiod to 25% for USO during the first. These modest frequencies of outperformance also concentrate in the first and fourth subperiods.
- Annualized Sharpe ratios for the best rules are in the range 1.7 to 2.3 across subperiods. No rule family dominates the best rules.
- Percentages of rules outperforming passive continuous holdings in-sample on a gross basis varies considerably by subperiod:
- Applying the assumed levels of trading frictions to in-sample tests shaves percentages of outperforming rules by a few points. Annualized net Sharpe ratios for the best rules are in the range 1.6 to 2.2 across subperiods.
- For net performance of portfolios reformed every six months using the best rules selected out-of-sample based on data from the past six months and corrected for data snooping bias:
- Annualized Sharpe ratios range from -0.93 for futures during the third subperiod to 0.37 for USO during the first.
- Across the two asset types, four subperiods and two methods of correcting for snooping bias, the average percentage of portfolios beating the median net Sharpe ratio of all trading rules is less than 2%.
In summary, evidence indicates that technical trading rules do not work for crude oil futures or USO on a net out-of-sample basis.
Cautions regarding findings include:
- A sample period of less than a decade is short in terms of variety of economic conditions, and six months may be short as a rolling training interval.
- Some investors may bear trading frictions higher than assumed values, making profitability of technical trading more elusive than indicated.
- Higher or lower data frequencies may work better.
See other tests of the same set of rules in “Technical Analysis: ‘Anathema to the Academic World’?”, “Does Technical Trading Work with Commodity Futures?” and “Technical Analysis Tested on Long-run DJIA Data”.
