Summary Over the last 10 years, the number of oil ETFs has exploded with an increasing number of complex instruments available to investors to gain exposure to crude oil. Many such ETFs appear attractive to the profit-minded trader, but it is up to educated investors to determine which product is most appropriate given his/her objective, risk appetite, and timeframe. This article analyzes the most popular commodity and oil ETFs to determine which most effectively tracks the price of oil over a series of different timeframes. Commodities has arguably been the most challenging sector in which to turn a predictable profit over the past 10 years. Crude oil, the most popular commodity in the sector, has seen its price double, lose 75% of its value, double again and, most recently, drop by 50%. However, with great volatility comes great opportunity, and it is no surprise that oil prices earn front-page headlines on all major financial websites on a daily basis. For years, most small, individual traders were unable to trade crude oil. Direct trading of oil requires buying and selling of futures contracts, with one futures contract usually representing 1,000 barrels. With oil trading at an average price of $80/barrel over the past decade, a single contract would cost $80,000 — too risky for most recreational traders. Even with the necessary pocketbook, trading futures contracts is particularly dangerous in that they expire every 30 days, requiring a trader to cash out at undesirable prices or be forced to take physical delivery of the oil. That all changed in 2006 with the arrival of the United States Oil Fund (NYSEARCA: USO ), an ETF that bought and held oil futures contracts itself, and allowed traders to buy shares for under $100. Over the next 5 years, an explosion of new commodity ETF products hit the market that allowed investors myriad increasingly complex opportunities to gain direct exposure to oil. With so many products available, many investors do not understand exactly what sort of exposure they are purchasing and how closely it will actually track oil. This article does not attempt to convince you, the reader, to buy oil. Rather, it assumes that you have already made the decision to do so, and instead will discuss the most effective way to go long oil without buying futures contracts. With a market capitalization of $3.2 billion and average daily volume of 28 million shares, the United States Oil Fund is among the most the most popular commodity ETFs, and by far the most popular pure oil ETF. The ETF was launched in April of 2006 and was the first of its kind. It allocates about 75% of its holdings to oil futures contracts. Each month, it buys near-term futures contracts–which best approximate the spot price of oil–and then a week or two prior to expiration, sells them and simultaneously uses these funds to buy the next month’s contracts, thereby avoiding taking physical deliver of more than $2 billion worth of oil (or 40 million barrels) and maintaining constant exposure to the commodity. For this service, the fund charges an annual fee of around 0.7%. However, this process of buying and selling contracts is not without it complications. More on this in a moment. After witnessing the popularity of USO and its cousin the US Natural Gas Fund (NYSEARCA: UNG ), other ETF companies were quick to jump on the bandwagon with increasingly innovative and volatile products. In late 2008, ProShares upped the ante and introduced the Ultra Bloomberg Crude Oil ETF (NYSEARCA: UCO ), which utilized leverage to deliver 2x the daily movement of oil. That is, if oil (and USO) gained 2% in a day, UCO would gain 4%, and if oil lost 2%, UCO would lose 4%. Unsurprisingly, this product was embraced by daytraders due to the enhanced volatility that is their lifeblood. However, it was also traded by longer-term traders looking to capitalize on a prolonged rally in crude oil. Like clockwork, 4 years later in late 2012, the company VelocityShares decided that 2x volatility just wasn’t cutting it and released the exceptionally volatile VelocityShares 3x Long Oil ETF (NYSEARCA: UWTI ). As its name suggests, this ETF was designed to move 3x the daily price of oil. Despite their differences in leverage, all three products work similarly in that they buy futures contracts and roll them over each month, aiming to track the price of oil on a daily basis. That being said, the devil is in the details–and the interworkings of these ETFs have a lot of details that dictate whether these ETFs are effective in accurately tracking the price of oil. Let’s start simple. Figure 1 plots the price of oil versus the price of USO since its inception in April 2006. (click to enlarge) Figure 1: Crude Oil versus USO since inception in 2006, showing underperformance of ETF versus its underlying commodity Data source: Yahoo Finance; c hart created by author. Conveniently, both began the period at nearly identical prices of $68 per share or per barrel. Since then, oil has slid to $46/barrel as of September 22, 2015 while USO has slid much steeper to just $15/share. What explains this underperformance? While the previously discussed process by which USO rolls over its futures contracts each month guarantees continuous exposure to oil, it is not without its drawbacks. Were subsequent futures contracts equally priced, it would not be an issue. The fund would sell X number of soon-to-expire contracts and use these funds to buy X number of next-month contracts. However, futures contracts of commodities such as oil frequently trade in a structure known as contango where later contracts are more expensive than near contracts. This is understandable, particularly after oil has taken a large fall, that investors expect prices to rebound in the long term as uncertainty increases. Unfortunately for funds such as USO, this means that each month the fund is selling X number of contracts and buying X-Y number of contracts. Effectively, the fund is selling low and buying high. And as contango can routinely reach 1-2% per month during periods of wide contango, the fund sees a price-independent degradation of roughly this percentage. While this is relatively minor in the short term, it adds up and can be relatively devastating for long term holders, as seen in Figure 1. What about UCO and UWTI? Figure 2 below plots the performance of oil versus USO versus UCO versus UWTI since December 10, 2008. 2008 was used as it encompasses the full history of both USO and UCO. The price history of UWTI from 2008 to 2012–when it debuted–was reconstructed based on price history of USO and UCO. (click to enlarge) Figure 2: Crude Oil versus USO, UCO, and UWTI since 2008, showing massive underperformance of leveraged ETFs versus USO and crude oil Data source: Yahoo Finance; c hart created by author. If USO “underperformed,” then UCO and UWTI were decimated. UWTI dropped 99.3% from an estimated $1841 per share to just $11 per share while UCO dropped 92% despite oil squeaking out a 5% gain. This dramatic underperformance versus both oil and USO occurred for two reasons. First, the impacts of rollover discussed above are compounded due to leverage. If the monthly contango in the futures market is 2%, the attributable loss increases to 4% for UCO and 6% for UWTI, which adds up very quickly. Second, due to the leveraging process a phenomenon known as “leverage-induced decay” also weighs on performance. I will spare you all the math, but suffice to say, large moves in one direction followed by sharp reversals leads to under-performance of leveraged ETFs independent of the effects of contango. What does this mean for oil traders? Figure 3 below uses the data in Figure 1 and 2 above to calculate average, expected underperformance versus the price of oil sustained from holding USO, UCO, and UWTI over a yearlong period. Overall, 2000 different 1-year periods are used to generate this data (click to enlarge) Figure 3: Expected underperformance of USO, UCO, and UWTI based on the number of days the ETF is held, from 2008-2015 data Data source: Yahoo Finance; c hart created by author. A 22-day hold in USO is predicted to result in a 1% underperformance versus oil. That is, if oil gains 5% during this period, the ETF would be predicted to yield around 4%. On the other hand, it would take just 9 days to reach a 1% underperformance holding UCO and a mere 6 days to see a 1% underperformance holding UWTI. Over a typical year-long period, USO is expected to underperform by 10.9% compared to 22.2% for UCO and 37.4% for UWTI. It should be noted that the underperformances for UCO and especially UWTI are somewhat deceptive and in many cases may actually be much lower. For UWTI, when oil falls greater than 33.3% in a year, UWTI will inevitably “outperform” oil given that it cannot fall more than its predicted 100%, which skews the mean underperformances shown in Figure 3 to the upside. However, when sitting on an 80-90% loss, I expect any such “outpeformance” feels rather pyrrhic. Based on this analysis, it is clear that USO outperforms UCO and UWTI and comes the closest to accurately tracking the price of oil. UCO and UWTI have their uses among the day-traders and swingtraders, but should not be used as investment tools as the long-term drawdown is simply too great to justify its use. Sure, should oil double in a year, the 37% underperformance is acceptable given the predicted 300% gain, but if oil is flat on the year–which occurs much more frequently than that edge case–you are sitting on an inexcusable loss. Of the 3 ETFs, USO offers the best risk/reward profile and, in my opinion, is the superior product and the only one that should be considered for long-term investors. So far, I’ve limited this discussion to popular commodity ETFs that are designed to mimic the spot price of oil–so-called “pure oil” ETFs. As discussed, the big drawback of these products is that you CAN’T mimic the spot price of oil, not over the long term. Let’s now consider oil companies themselves. Major producing companies derive a substantial portion–if not all–of their income from oil sales. Therefore their share prices should be closely tied to the price of oil. The advantage of oil stocks over pure oil ETFs, of course, is that they are not subject to the same rollover losses as USO. If it can be determined that oil companies effectively track the price of oil on a day-to-day basis, it can be expected that they would do so over the long-term and not be subject to decay. Rather than analyze individual companies whose stocks are intermittently subject to forces not directly related to the price of oil such as earnings reports, lawsuits, and legislation, let’s instead consider a basket of oil companies to smooth out these events i.e. the oil sector ETFs. The 3 most popular oil sector ETFs are the Energy Select Sector SPDR (NYSEARCA: XLE ), the MarketVectors Oil Services ETF (NYSEARCA: OIH ), and the SPDR S&P Oil & Gas Exploration ETF (NYSEARCA: XOP ). XLE’s diverse holdings include large cap oil companies involved in all aspects of the petroleum industry such as Exxon Mobil (NYSE: XOM ), Chevron (NYSE: CVX ), and Schlumberger (NYSE: SLB ). OIH’s largest holdings, on the other hand, are more focused on oil service companies alone and include SLB, Halluburton (NYSE: HAL ), and Baker Hughes (NYSE: BHI ). Finally, XOP’s largest holdings include major exploration companies such as HollyFrontier (NYSE: HFC ), PBF Energy (NYSE: PBF ), and CVR Energy (NYSEMKT: CVR ). Figure 4 below plots the performance of each versus Oil and USO since 2009. (click to enlarge) Figure 4: Crude oil versus select oil sector ETFs Data source: Yahoo Finance; c hart created by author. Notice that the price of oil tends to form the upper bounds of this chart while USO forms the lower bounds with the 3 oil sector ETFs somewhere in between. Of the 3, XLE seems to be the best, handily outperforming both oil and USO over the 10 year period. This suggests that the oil sector ETFs are superior to USO in their ability to track oil without price-independent losses, as predicted. However, the key concept is correlation. Apple Computer (NASDAQ: AAPL ) has certainly outperformed oil and USO over the past decade as well, but given none of its businesses are related to oil, it has no correlation to the petroleum industry and is not a useful analogue. Correlation can be determined by looking at beta and the R-squared value. Figure 5 below shows a scatterplot between the daily percent change of the price of oil versus USO and XLE. (click to enlarge) Figure 5: Scatterplot comparing the daily percent performance of crude oil versus XLE and crude oil versus USO, showing a tighter correlation between oil and USO Data source: Yahoo Finance; c hart created by author. It can be easily appreciated that oil vs USO (the red dots) forms a tighter linear relationship than oil vs XLE (the blue dots), which is much more diffuse. Further, notice that the slope of the oil vs USO relationship is closer to 1:1 on the x- and y- axes while the oil vs XLE relationship is flatter. This illustrates the twin concepts of correlation and beta, respectively. Correlation is the idea that two entities are related. If entity A moves a certain magnitude, entity B moves a predictable magnitude in response. However, it does not have to be 1:1. For example, for every 10% that A moves, entity B might move 25%. Predictable, but not equal. Correlation is measured by the R-Sq value. In finance, beta is traditionally thought of as a measure of the volatility of a security or portfolio in comparison to the market as a whole. A stock with a beta of 1.0 indicates that a stock’s price movement will mimic that of the market – if the S&P 500 gains 5%, the stock will gain 5%; if the market is flat, the stock will be flat; and if the market falls 5%, the stock will fall 5%. A stock with a beta of 2 is more volatile than the market – a tech stock, for example – and will gain or lose twice that of the S&P 500 or whatever index is used as the benchmark. A beta of 0.5 is comparatively less volatile – a utilities stock, for example – and will gain or lose half of the market’s performance. While the beta is typically applied to compare a stock to a market or index it is a relatively simple calculation and can be used to compare any two equities or funds against each other. Equation 1 below shows the equation used to calculate beta: Equation 1: Beta = Covariance (Daily % Chg stock for which beta is being calculated, Daily % Chg underlying index)/Variance (Daily $ Chg Underlying index) In this case, we will be comparing the price of oil versus each of our ETFs. An ETF with a beta of 1.0 means that the ETF tracks oil on a 1:1 basis on a daily basis. Figure 6 below shows the R-Sq and beta values for USO, XLE, OIH, and XOP compared to oil. (click to enlarge) Figure 6: Betas and R-Sq values for USO, XLE, OIH, and XOP showing USO trumps the 3 oil sector ETFs by a large margin Data source: Yahoo Finance; chart created by author. Again, USO comes out on top in both categories. USO’s R-Squared with oil is 0.81, handily beating XOP which comes in second with an RSQ of 0.57 while its beta is 0.80, crushing XOP’s 0.43. Thus, while all three oil sector ETFs may outperform USO, they do so due to factors not directly related to the price of oil. This article is not about picking good investments. It is about selecting the ETF that best accomplishes a certain objective: to track the price of oil accurately over the short and long term. In conclusion, this analysis of several popular oil ETFs has determined that the United States Oil Fund is the best long-term investment in terms of accurately tracking the price of oil as well as minimizing losses due to futures contract rollover. That is not to say that the other ETFs might not have niche uses. UWTI and UCO are certainly effective trading vehicles for those trying to capitalize on an oversold bounce or socioeconomic-driven event over 3-5 days. Likewise, XOP, XLE, and OIH may be superior to USO for super-long terms investors with a Warren Buffet-like mindset who plan to hold for well-over 2 years and care more about historical performance than accuracy in tracking an underlying commodity. However, for the typical investor who is looking to capitalize on a steady rise in oil prices from a week to 2 years or so, I firmly believe the USO is the most effective trading vehicle to do so.