Recent work has looked at whether using simple mean reversion technical trading strategies can be employed profitably for portfolios for Crude Oil, Natural Gas, Gasoline and Heating Oil futures. Considering twenty two years of data – through the use of mean-reverting calendar spread portfolios, dynamic hedge ratios and trading signals based on Bollinger Bands – it has been shown that most combinations of front-month and second month futures are significantly profitable for all commodities with the best results for WTI Crude and Natural Gas. A number of papers over the last 5 years have raised serious doubts whether energy futures markets can be considered weakly efficient in the short-term.

This post gives an overview of the recent literature and the various trading strategies employed.


Over the past decade, energy commodity prices have exhibited dramatic rises and falls to an extent not observed since the energy crisis of the 1970s. At the same time, energy futures have experienced an impressive increase in financial investor demand after the US futures markets have been transformed radically due to the US Commodity Futures Modernization Act in 2000 which introduced more flexibility, allowing financial agents such as commodity index funds to enter them. On the one hand, energy commodities are regarded as low-cost diversification instruments that widen the opportunity set for portfolio optimization. Furthermore, markets with a high liquidity attract short-term investors who intend to go long or short the asset for only a few days or even on an intra day basis. These developments have triggered a controversial discussion in literature aiming to establish whether supply and demand fundamentals or speculative trading effects prevail in the price building process of energy futures contracts.

Given the ever increasing interest in these markets, a question of tremendous relevance for academic research and practitioners related to the price building processes in energy futures markets is whether historical price patterns can be exploited by savvy investors and whether there could be profitable trading opportunities in four very popular energy futures markets by using simple trading rules.

Mixed messaged for an efficient market?

The discussion among academics and practitioners about the merits of technical analysis has been ongoing for decades. The comprehensive survey by Park and Irwin (2007) shows that the main focus of attention of academic research is on equity and foreign exchange markets, not on commodities. The profitability of technical trading rules for commodity markets has been addressed by a few studies, mainly related to momentum and simple moving average rules.

Shambora and Rossiter (2007) use an artificial neural network model with moving average crossover inputs to predict crude oil futures prices and document significant profitability which is at odds with the expectation of an efficient oil futures market. In the same year, Tabak and Cajueiro published a paper that analysed the efficiency of Brent and West Texas Intermediate (WTI) Crude Oil using Hurst analysis and found evidence that the oil market has become more efficient over time.

A year later, Alvarez-Ramirez considered auto correlations of the inter-national crude oil price process documented that over long horizons the crude oil market is consistent with the efficient market hypothesis by Fama (1970) but cannot exclude the possibility of market inefficiencies at short time horizons.

“..showing that a particular strategy based on exploiting stale information on average earns a positive cash flow over some period of time is not, by itself evidence of market inefficiency. To earn this profit, an investor may have to bear risk and his profit may just be a fair market compensation for risk bearing”

Two papers in 2010 (Wan and Liu; Lean) test for the efficiency of the WTI Crude Oil market by showed that crude oil prices are becoming more efficient over time for all considered time horizons. The second study considered WTI Crude Oil spot and futures prices using mean-variance and stochastic dominance approaches and found that there are no arbitrage opportunities between spot and futures prices. However, this was not the case if we considered the high-frequency dynamics of futures price data of Crude Oil, Heating Oil, Gasoline, and Natural Gas, where Wang and Yang (2010) identified market inefficiencies, especially for Heating Oil and Natural Gas mostly during periods of bullish markets.

Going with the flow: momentum strategies

Millre and Rallis (2007) apply contrarian and momentum strategies to US commodity futures contracts. While contrarian strategies are found to perform poorly, momentum strategies appear to be profitable over time periods of up to 12 months. Shen (2007) compare the performance of momentum strategies for commodity and equity markets and document highly significant momentum profits for holding periods up to 9 months with magnitudes comparable to those realised with equity trading.

Contrary to the studies claiming that technical strategies in commodity futures earn profits that cannot be considerably weakened by the relatively low transactions costs prevailing in these markets, Marshall et al. (2008) apply a large set of technical trading rules to 15 major commodity futures series, and find that the resulting profits cannot consistently exceed those expected to emerge due to random variation. More recently, Fuertes (2010) examined combinations of momentum and term structure trading signals in commodity futures markets and find momentum and term structure strategies to be profitable when implemented individually while Szakmary (2010) document pure trend-following strategies generally to outperform momentum strategies.

Trading on mean reversion

Apart from the controversy about the profitability of technical trading rules in commodity markets, none of these studies apply Bollinger Bands which are an easy to implement tool for technical analysis. Despite the fact that these trading rules enjoy extensive popularity among practitioners, the academic literature investigating their performance is rather limited. Moreover, these studies cast doubts on the profitability of the Bollinger Bands. Using data of equity indices and the forex market, Lento et al. (2007) establish that the Bollinger Bands are consistently unable to earn profits in excess of the buy-and-hold trading strategy.

Leung and Chong (2003) compare the profitability of Moving Average Envelopes and Bollinger Bands for a broad sample of equity market indices and find that Bollinger Bands under-perform the Moving Average Envelopes

All is not lost for Bollinger Bands?

Given the recent empirical evidence that pure Bollinger Bands has not been as effective as other technical indicators, worked turned to consider more layered mean-reversion strategies with Lubnau & Todorova looking at trading rules for hedge portfolios consisting of futures on the same underlying asset but for different maturities for four NYMEX contracts; WTI Crude Oil, Heating Oil, Gasoline and Natural Gas.

They tested mean-reversion strategies involving calendar spreads constructed of those four futures. Spreads are particularly interesting because they are less likely to suffer from information shocks, as the movements of the two legs often offset each other.

Using daily data from 1992 to 2013, they found most combinations involving the front-month and second- month futures to be significantly profitable for all futures under consideration. The best risk-adjusted results were documented for WTI Crude Oil and Natural Gas, with Sharpe Ratios in excess of 2 for most combinations and a rather even performance for all tested combinations. This re-affirmed previous work which demonstrated that it was not possible to rule out the possibility of short-term market inefficiencies in the Oil Markets.

Returns of WTI mean-reversion strategies
Trading system example for front-month and second-month WTI futures. Note: This figure shows the z-score calculated for WTI front-month and second-month futures with a moving average and rolling standard deviation of 200 days for the five year period 2009–2013. The first z-score is calculated with the first 200 days of data available late in 2009. The solid red lines indicate the entry score of 2 and−2, the dashed red lines are the exit scores of 0.2 and−0.2. The gray shaded area indicates periods where the futures were in backwardation, where as the white area indicates contango

The Bollinger Band and mean-reversion strategies employed in their study assumed that shocks changing the spreads are short-lived and there is mean reversion in the hedge portfolios. Following the definition of Fama (1970), a market is called efficient if prices reflect all available information. In a more detailed view on the information sets used, weak-form efficiency is defined to comprise only historical data. The results gave conclusive evidence that it is possible to use technical trading systems profitably in the energy futures market

Over the study period, Sharpe Ratios in excess of 2 are obtained with almost all WTI Crude Oil and Natural Gas combinations, while significant results for Gasoline and Heating Oil futures are achieved mainly by using shorter Moving Average periods and by trading front-month and second- or third-month futures. A detailed discussion of the results show that events triggering an entry in the market vary randomly so that the over-all profitability does not arise from a strategy always buying front contracts and selling back contracts or the other way around. Overall, recent work has cast serious doubts that energy futures markets are efficient in the short-term.


Overall, a number of papers have shown that over long horizons the crude oil market is consistent with the efficient market hypothesis by Fama (1970) but cannot exclude the possibility of market inefficiencies at short time horizons. A number of different studies appear to confirm the existence of economically exploitable short-term inefficiencies. Whilst most do not consider transaction costs, their main result – there are inefficiencies in the crude oil market in the short-term – can be confirmed in a real-world setting using a conservative transaction cost rate. Moreover, this conclusion is shown to hold not only for the crude oil market, but can be extended to the case of other popular energy futures markets.

  • Alvarez-Ramirez, J., Alvarez, J., Rodriguez, E., 2008. Short-term predictability of crude oil markets: a detrended fluctuation analysis approach. Energy Econ. 30, 2645–2656.
  • Alvarez-Ramirez, J., Alvarez, J., Solis, R., 2010. Crude oil market efficiency and modeling: insights from themultiscaling autocorrelation pattern. Energy Econ. 32, 993–1000
  • Chan, E., 2013. Algorithmic Trading:Winning Strategies and Their Rationale.Wiley, New Jersey
  • Fama, E., 1970. Efficient capitalmarkets: a reviewof theory and empiricalwork. J. Financ. 25, 383–417.
  • Lean, H.H., McAleer, M., Wong, W.-K., 2010. Market efficiency of oil spot and futures: a mean-variance and stochastic dominance approach. Energy Econ. 32, 979–986
  • Lento, C., Gradojevic, N.,Wright, C.S., 2007. Investment information content in Bollinger Bands? Appl. Financ. Econ. Lett. 3, 263–267.
  • Lubnau T, Neda T. Trading on mean-reversion in energy futures markets Energy Economics 51 (2015) 312–319
  • Leung, J., Chong, T., 2003. An empirical comparison of moving average envelopes and Bollinger Bands. Appl. Econ. Lett. 10, 339–341.
  • Wang, Y., Liu, L., 2010. IsWTI crude oilmarket becomingweakly efficient over time?New evidence from multiscale analysis based on detrended fluctuation analysis. Energy Econ. 32, 987–992.
  • Wang, T., Yang, J., 2010. Nonlinearity and intraday efficiency tests on energy futures markets. Energy Econ. 32, 496–503.

I'd love to hear what your thoughts are...please feel free to leave a reply