On Seeking Alpha's StockTalks, it is not uncommon to read a comment that expresses the following general sentiment. "The VIX is down 20% today, but the VelocityShares Daily Inverse VIX Short-Term ETN (NASDAQ:XIV) is only up 10%. It has a lot of catching up to do!" Or, "it is a broken ETN!"
The most common response points out that these ETPs do not track the spot VIX, but are instead designed to track short-term VIX futures. A comment by one reader went on to assert that the ETPs follow their tracking index "perfectly."
Anyone who has done their due diligence knows that the five short-term volatility ETPs discussed in this article are not intended to track the spot VIX. But do these ETPs truly follow their tracking indexes perfectly?
To address the question, this article takes a close look at five popular ETPs that are linked to short-term VIX futures: the iPath S&P 500 VIX Short-Term Futures ETN (NYSEARCA:VXX), XIV, the VelocityShares Daily 2x VIX Short-Term ETN (NASDAQ:TVIX), the ProShares Ultra VIX Short-Term Futures ETF (NYSEARCA:UVXY), and the ProShares Short VIX Short-Term Futures ETF (NYSEARCA:SVXY). We evaluate both their long-term performance over more than four years, and their typical daily performance over the same timeframe.
Sources of Data and Information
This study analyzes historical data from October 4, 2011 (the inception date of both UVXY and SVXY) through January 15, 2016, inclusive. The results provide a head-to-head comparison over the entire timeframe when all five ETPs have been simultaneously available for trading. Historical settlement prices of VIX Futures (monthlies) were downloaded from the CBOE website. The historical closing prices of the five ETPs were downloaded from Yahoo! Finance.
The prospectus of each ETP can be accessed via one of the following links: UVXY and SVXY, XIV and TVIX, VXX. The methodology that underlies the calculation of S&P VIX Futures Indices is documented on the S&P Dow Jones Indices site. Four of the futures indices are related to this article: SPVXSPID, SPVXSP, SPVIXSTR, and SPVXSTR. SPVXSPID and SPVIXSTR are real-time indices, whereas SPVXSP and SPVXSTR are end-of-day indices. SPVXSPID and SPVXSP are "excess return," ER, indices, whereas SPVIXSTR and SPVXSTR are "total return," TR, indices. The daily change in a TR index is equal to the daily change in the corresponding ER index plus a daily three-month U.S. Treasury bill return.
Officially, the daily performances of VXX, XIV, and TVIX should track the appropriate relationship to SPVXSTR, whereas the daily performances of UVXY and SVXY should track the appropriate multiple of SPVXSP. Because two different indices are referenced, the gross annual fees, as published in the prospectuses, are not directly comparable.
On the other hand, when each of the ETP's published net fees are used, direct comparisons of the five ETPs are possible when the ER index (SPVXSP) is used to track the end-of-day performance.
Method of Evaluation
Because intraday ETP and Futures prices are not readily available, this analysis is based on closing prices. SPVXSP is used as the reference index, and each ETP's net fees, as published in its prospectus, are used without modification.
The following six paragraphs summarize the strategy used to analyze the performance of the ETPs. Readers who are not interested in the details should skip to the results in the next section.
- Calculate the Historical Daily Changes in the SPVXSP Index, based on historical settlement prices of the VIX Futures. In a recent Seeking Alpha article, I outlined two seemingly-different pathways for calculating the daily change in the index, and demonstrated that both pathways yield identical results. To ensure accuracy in this report, the daily changes were calculated separately by both methods, and the results were verified to be precisely equal.
- Calculate the Predicted Daily Changes Separately for each ETP. To do this, multiply the daily index change by the appropriate leverage factor, then subtract the ETP's daily net fee. The leverage factors are: +2 for UVXY and TVIX, -1 for XIV and SVXY, and +1 for VXX. The daily net fees were estimated as being the fee which, if compounded over 252 consecutive market days, would be equal to the ETP's net annual fee.
- Calculate the Actual Daily Changes of each ETP Separately. Note that this and the daily changes described in the preceding two paragraphs are fractional, and are relative to the previous day's closing price. The fractional change is found by subtracting yesterday's closing price from today's closing price, and then dividing the difference by yesterday's closing price.
- Calculate Running Prices for each ETP. Do this separately for each ETP, based first on predicted daily changes, and then based on actual daily changes. To enhance direct comparisons, all initial ETP closing prices for October 4, 2011 were set to a value of 100. The running price for a given day is found by multiplying the running price of the previous day times 1 plus the daily fractional change.
- Determine the Coefficient of Determination for Running Actual vs. Predicted Prices. Do this separately for each ETP. The coefficient of determination, also known as r-squared, is interpreted as representing the proportion of the variance in the dependent variable that is predictable from the independent variable. In the present context, an r-squared value of 0.98 indicates that 98% of the total variation in actual running prices can be explained in terms of the predicted prices. The remaining 2% of the total variation in running prices is unexplained. This result is used to assess how well the ETPs track expectations overall in the long run (Result 1).
- Determine the Coefficient of Determination for Predicted Daily Changes vs. Actual Daily Changes. Do this separately for each ETP. This result is used to assess how well the ETPs track expectations on a day-to-day basis (Result 2).
Result 1: Long-Term Tracking is Quite Good for All Five ETPs
Results for long-term tracking, over 4.28 years since October 4, 2011, are summarized in Table 1. The r-squared coefficients range from 0.9983 to 0.9999, indicating that in the long term, actual ETP prices can be reliably predicted from predicted prices that are based on the rules defined in the ETP's prospectus.
Four of the five ETPs had closing prices on January 15, 2016, that were better than the prediction. These are green-shaded in the table. Head-to-head, the actual compound annual growth rate, CAGR, of TVIX is the same that of UVXY, -87.81%. Based on the data, the actual CAGR of SVXY (33.29%) appears to be slightly better than that of XIV (32.87%), by a slim margin of 0.42%.
Overall, the historical vs. predicted performance has been strong for all five of the ETPs. The difference between XIV and SVXY may or may not be meaningful, given the high level of daily fluctuation, discussed in the next section.
Do the ETPs follow their tracking index perfectly? Based on an extended timeframe of more than four years, their overall long-term tracking has been quite good. Score a plus for the ETPs. Long-term holders of the ETPs can anticipate gains or losses fairly accurately. That said, no prudent investor would intentionally hold a long position in VXX, TVIX, or UVXY over any extended period of time, given that these ETPs have averaged annual losses of 55.4% to 87.8% (TVIX and UVXY) since October 4, 2011.
Result 2: Day-to-Day Tracking: Not So Good
Results for day-to-day tracking, assessed over 4.28 years since October 4, 2011, are summarized in Table 2. I offer comments on four of the entries (rows) in the table.
- SEM (Standard Error of Measurement) of the Daily Change Prediction. 70% of actual daily changes are expected to differ from the predicted change by this amount or less. For example, VXX has an SEM of 1.377%, indicating that the daily change in the price of VXX can be expected to differ from the predicted change by as much as 1.377% of the previous day's closing price. The smaller the SEM, the more reliable the prediction. While the SEMs of XIV and SVXY are comparable (1.45%), the SEM of TVIX (3.40%) is 23% larger than that of UVXY (2.76%), indicating lower confidence in the day-to-day prediction of TVIX price changes.
- Median Magnitude of Daily Change. Half of daily changes will be greater than this value (in either the positive or negative direction), and half will be less. The values for XIV and SVXY are comparable (2.14-2.16%). The median magnitude of UVXY (4.30%) is 13% larger than that of TVIX (3.81%), suggesting somewhat higher volatility in the daily price action of UVXY.
- SEM as a Percentage of Daily Change Prediction. This represents the SEM in (1), expressed as a percentage of the median magnitude of daily change in (2). To illustrate, for TVIX, the value is 89.10%. This implies that on a typical day, when the daily predicted change is a loss of 3.81%, the actual daily change can be expected to fall anywhere between a loss of 0.41% and a loss of 7.20%. The larger the value, the greater the daily tracking error, and the less reliably the ETP will act on any given day. On this basis, UVXY's daily tracking (64.2%) is more reliable than that of TVIX (89.1%), whereas XIV and SVXY are similar (67.2-67.7%). In all five cases, the typical error represents a large fraction of the predicted daily change percent!
- R-Squared for Predicted vs. Actual Daily Change. In this context, the r-squared value indicates the percentage of actual daily price changes that can be explained in terms of the predicted daily price changes. The r-squared value of UVXY (0.883) is 8.2% better than that of TVIX (0.801). The values of XIV and SVXY are comparable (0.872-0.873).
One issue potentially makes all of the tracking errors appear larger than they actually are. The equity markets normally close at 4 pm ET, whereas the futures markets close at 4:15 pm. This means that the futures settlement data actually predict hypothetical ETP results, as if the equity markets also closed at 4:15 pm. The 15-minute offset will degrade, at least to some extent, the correlation between actual and predicted ETP closing prices. We expect, however, that there should be no long-term directional bias in the price changes of VIX futures between 4:00-4:15 pm. As a result, long-term differences between the overall ETP averages are meaningful.
According to the prospectuses, fees accumulate every day, including weekends and market holidays. Our calculation assumed a 252 market-day year, and compounded the annual fee over 252 trading days. The long-term error introduced by our approximation is expected to be negligible.
Do the ETPs follow their tracking indexes perfectly? Based on day-to-day tracking errors evaluated over more than four years, the day-to-day tracking records are not very good. Score a minus for the five ETPs. This conclusion is consistent with comments to the effect that a given ETP is not living up to expectations. A high degree of uncertainty is associated with short-term trading - even when you can correctly guess the direction of volatility change.
Other Factors to Consider When Choosing a Short-Term Volatility ETF
To summarize the discussion so far, two key factors to consider when selecting a short-term volatility ETP are the accuracy of long-term tracking (Result 1), and the typically large day-to-day tracking errors (Result 2).
Additional characteristics that should be considered when choosing an ETP are summarized in Table 3.
ETF or ETN? An exchange traded fund, ETF, is a fund that holds the asset it tracks. An exchange traded note, ETN, is an unsecured debt note issued by an institution. To describe their legal structures, Morningstar uses the terms, "collateralized debt instrument" and "uncollateralized debt instrument" for the ETFs and ETNs respectively. The credit rating of the underwriter is an important consideration when trading ETNs, because in principle, if the underwriter were to go bankrupt, the investor could risk a total default.
A Schedule K-1 Tax Filing is required for both UVXY and SVXY. On the surface, this is a nuisance. In practice, it takes some amount of effort to figure out. However, if you were to hold SVXY for an extended period of time and you realized short-term capital gains, the K-1 filing could be worth your while. The reason is that a portion of the short-term gains reported by your broker are converted into long-term gains on the Schedule K-1, and are actually taxed by the IRS at the lower long-term capital gains rate. On the other hand, if the instrument were held more than one year, the tax advantage might be reversed. Be advised that the preceding comments do not constitute tax advice. Each reader should consult a certified tax advisor for information regarding their personal tax situation.
Average Volume, Dollars is the metric I like to use to evaluate liquidity. The greater the daily average, the easier it should be to buy or sell the instrument at, or very close to, the prevailing market price. On this basis, VXX is the most liquid ETP among the five.
Options are an important component of many trading strategies. Three of the ETPs currently offer options, whereas two do not.
The Ease of Shorting: Are shares available to short? If so, is there an additional fee? This is not included in Table 3, because the answer differs from broker to broker.
The head-to-head comparisons in this article provide a starting point for deciding which volatility ETP is best suited for a given investing/trading strategy. Happy hunting to all!
The information and analysis in this article are believed to be accurate. Any alleged error(s) will be examined, and will be corrected if confirmed to be valid.
ETP performance was analyzed on the basis of historical prices over a period of 4.28 years. Historical outcomes are dependent on the beginning and end dates chosen, and provide no guarantees of future performance.
All five ETPs involve a high level of investment risk! Know the product, and trade carefully.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.