Leveraged ETFs in Low-Volatility Environments
Leveraged ETFs (such as SPXL – (Direxion Daily S&P 500 Bull 3X Shares) offer amplified exposure to the S&P 500, promising high returns but exposing investors to volatility drag caused by daily rebalancing. This effect can significantly erode performance over longer horizons, particularly during periods of elevated market volatility. Inspired by recent research, The Volatility Edge, A Dual Approach For VIX ETNs Trading, focused on volatility-linked ETNs, we propose a volatility filter that adjusts ETF exposure based on the relationship between short-term realized volatility and implied volatility. By reducing exposure in high-volatility periods and maintaining it in calmer markets, this approach aims to harness leverage effectively while mitigating the most damaging drawdowns.
Introduction
In recent years, exchange-traded products have become increasingly sophisticated, offering investors exposure not only to broad equity indices but also to more complex strategies that amplify or even invert daily returns. Among these instruments, leveraged ETFs stand out for their promise of magnified gains but also for the hidden risks embedded in their design.
A prominent example can be found in SPXL (Direxion Daily S&P 500 Bull 3X Shares) and SPXU (Direxion Daily S&P 500 Bear 3X Shares), which seek to deliver three times the daily return of the S&P 500, in the long and inverse directions, respectively. To achieve this leverage, the funds rebalance their derivative exposure on a daily basis. While this mechanism ensures that the leverage target is met each day, it also introduces a well-documented phenomenon known as volatility drag: over extended horizons, compounding of daily returns causes realized performance to diverge substantially from the “theoretical” triple return of the index. Periods of high volatility tend to accelerate this erosion, making such products especially fragile during turbulent markets.
Recent research The Volatility Edge, A Dual Approach For VIX ETNs Trading by Zarattini, Mele, and Aziz (2025) suggests a potential path to mitigating this problem. Their study on volatility-linked ETNs introduced the idea of a volatility filter, which compares short-term realized volatility, estimated using a rolling window of recent returns, with the implied volatility reflected in the VIX index. The VIX, often called the “fear index”, is a financial indicator that measures the expected volatility of the U.S. stock market over the next 30 days based on S&P 500 option prices. It reflects short-term market uncertainty or risk, with higher values indicating greater expected volatility and increased investor fear, while lower values suggest a calmer market with smaller price fluctuations. Importantly, the VIX does not predict the market’s direction, only how much prices are expected to move. When implied volatility exceeds the short-term realized measure, the strategy maintains exposure, when the filter signals elevated realized risk, exposure is reduced or cut entirely. This simple mechanism proved effective at avoiding some of the most damaging drawdowns in their backtests.
This approach gave us an idea that leveraged equity ETFs like SPXL and SPXU could benefit from a similar filter. Rather than maintaining exposure unconditionally, one could use the relationship between short-term realized volatility and implied volatility as a guide for investing. In this way, the strategy aims to harness leverage during favorable trends while avoiding some of the most damaging periods of volatility drag.
Methodology
In this study, the backtest covers the period from January 17, 2013, to July 31, 2025. We used daily data for SPXU and SPXL obtained from EODHD.com – the sponsor of our blog. EODHD offers seamless access to +30 years of historical prices and fundamental data for stocks, ETFs, forex, and cryptocurrencies across 60+ exchanges, available via API or no-code add-ons for Excel and Google Sheets. As a special offer, our blog readers can enjoy an exclusive 30% discount on premium EODHD plans..
As a benchmark to assess the risk-adjusted returns for our trading strategies, we employed SPY’s total daily data returns from the same source. We consider SPY the most appropriate benchmark, given that the analyzed ETFs provide three times the daily return of the S&P 500.
VIX data were obtained from the FRED, covering the same period.
As mentioned in the Introduction, the research The Volatility Edge, A Dual Approach For VIX ETNs Trading inspired us to design the following strategy. Since our goal is to compare short-term realized volatility with implied volatility, it is necessary to define both terms. Short-term realized volatility was calculated as the annualized standard deviation of SPY returns over recent days, which is straightforward given the availability of the data. By contrast, implied volatility is represented by the VIX, which is available with a one-day lag. We then use an average of recent VIX values to smooth SPY’s implied volatility.
Once both variables are obtained, their comparison determines whether or not to invest in leveraged ETFs such as SPXL (or SPXU for the purpose of hedging). This procedure is repeated on a daily basis. The final step is to optimize the length of the time window for both metrics in order to achieve the best performance relative to the benchmark. We therefore begin by examining the ETFs separately.
Basic performance characteristics in tables are presented as follows: the notation perf represents the annual return of the strategy, st dev stands for the annual standard deviation, max dd is the maximum drawdown, adjusted Sharpe r is calculated as the ratio of perf to st dev and adjusted Calmar r as the ratio of perf to max dd.
SPXL in Low-Volatility Enviroment
Let’s begin with the ETF SPXL (Direxion Daily S&P 500 Bull 3X Shares). SPXL, as a triple-leveraged ETF on the S&P 500 index, performs best during periods of steady growth when daily market moves are relatively small and the trend is clear. In such conditions, leverage works in favor of the investor without being offset by volatility drag, which otherwise erodes the fund’s value in choppier markets.
It is therefore crucial to identify a simple indicator that helps distinguish favorable environments from turbulent ones. One approach is to compare short-term realized volatility with implied volatility as measured by the VIX index. When VIX is higher than realized volatility, it indicates that the market expects larger moves than are actually occurring—in other words, investors are “overpaying for insurance.” For SPXL, this creates a supportive setting: realized swings remain contained, volatility drag is limited, and if the market is trending upward, the triple exposure amplifies those gains.
Therefore, we designed a strategy for investing in SPXL based on the relationship between realized and implied volatility. Specifically, when the average VIX value (implied volatility) exceeds SPY’s annualized X-day standard deviation (realized volatility), we take this as a signal to invest in SPXL. Otherwise (realized volatility is higher then implied), we remain out of the market. Each position is held for one day, after which the procedure is repeated.
Building on the paper that guided our thinking, the initial strategy’s parameters are set up as SPY’s 10-day annualized standard deviation, while the moving average of the VIX is calculated over the most recent 60 days.
Both the graph in Figure 3 and the results in Table 3 indicate that the SPXL strategy delivered significantly higher performance, with returns double those of the benchmark. However, these gains were accompanied by increased standard deviation and larger drawdowns, which are clearly visible in the graph. While these negative aspects slightly reduce the Sharpe ratio, the Calmar ratio still reflects the efficiency of the approach. Nevertheless, this outcome did not fully align with our objectives, and we therefore decided to test the robustness and modify the strategy.
The strategy was adjusted through two modifications. First, we fixed the 60-day window for the moving average of VIX values and varied only the length of SPY’s annualized standard deviation, specifically using the 5-, 10-, and 20-day measures.
The results show that extending the number of days used to calculate the standard deviation leads to higher annualized returns, while drawdowns and volatility decline. This effect can be explained by the fact that shorter windows (5 or 10 days) are highly sensitive to one-off swings in SPY, which results in more frequent exits from the market, whereas a longer window provides a smoother and more stable estimate of realized volatility. As a result, it is more common for the average VIX to exceed realized volatility, leading to more frequent entries into SPXL during upward-trending periods and thus higher returns, while the more stable signal simultaneously reduces volatility and drawdowns.
This leads to higher Sharpe and Calmar ratios, reaching values exceeding 1 and 0.80, respectively, in the strategy based on the 20-day standard deviation. These results outperform the benchmark and meet our expectations. While the modification of realized volatility yielded favorable outcomes, the question remains whether adjusting the implied volatility will prove equally successful.
The second modification of the SPXL strategy focuses on implied volatility, represented by the average VIX value over the most recent 60 days. In this case, the realized volatility is fixed at a 10-day window, while the averaging period for VIX values varies across 10, 20, 30, 40, 60, 80, and 120 days.
In the case of implied volatility represented by the VIX index, shorter averages produced better results, which can be explained by the different nature of this measure. While realized volatility is backward-looking and requires longer periods to smooth out random fluctuations, the VIX responds immediately to current events and investor sentiment. When a long averaging window is applied, the signal becomes overly smoothed and loses timeliness. By contrast, shorter windows, such as 10 or 20 days, capture prevailing market expectations more accurately and allow quicker identification of situations where implied volatility exceeds realized volatility. This increases the likelihood of timely entries into SPXL during favorable periods, thereby enhancing the strategy’s performance, while the additional noise introduced by shorter windows is less detrimental in this context. The strongest results were achieved with the 10-day moving average of the VIX, yielding a Sharpe ratio close to 1 and a Calmar ratio of almost 0.9. These outcomes are comparable to the best results obtained in the previous modification, although the Calmar ratio is noticeably higher.
The results indicate that the effectiveness of the SPXL strategy depends on a careful balance between the horizons used for realized and implied volatility. Longer windows for realized volatility reduce noise and stabilize the signal, while shorter windows for implied volatility preserve its responsiveness. However, combining the longest realized-volatility window (20 days) with the shortest VIX window (10 days) does not yield the best performance, suggesting that the two parameters interact and must be jointly optimized. Overall, the strategy performs best when realized volatility is measured over a moderate horizon (10 days) while implied volatility is averaged over a short window (10–20 days), balancing stability and adaptability.
Our approach for SPXL proved effective, successfully identifying periods when it is favorable to invest based on the relationship between realized and implied volatility. But what if the market moves in the opposite direction? In that case, SPXU, with its inverse and leveraged exposure, could potentially benefit, raising the question of whether a similar volatility-based indicator can be applied to guide investment decisions for SPXU.
SPXU in Low-Volatility Enviroment
In this part, we focused on the SPXU. SPXU (ProShares UltraPro Short S&P 500) is a triple-leveraged inverse ETF designed to deliver three times the inverse daily return of the S&P 500 index. This means that SPXU gains when the S&P 500 declines and loses when the index rises. Due to its leverage, SPXU is highly sensitive to short-term market fluctuations and daily volatility. It performs best during sustained downward trends in the market, while in choppy or bullish periods, volatility drag and compounding effects can significantly erode returns.
SPXU, as a triple-leveraged inverse ETF, gains when the S&P 500 declines and loses when the index rises. When short-term realized volatility of SPY exceeds the average VIX, it indicates that the market is moving more than investors had anticipated, reflecting heightened turbulence and unexpected swings. Such conditions often coincide with downward movements in the market, creating a favorable environment for SPXU. In other words, high realized volatility (SPY’s standard deviation) relative to implied volatility (VIX moving average) could signal short-term uncertainty and potential declines in the S&P 500, which an inverse leveraged ETF like SPXU can exploit.
Therefore, the strategy for investing in SPXU is designed as the opposite of the SPXL approach. Specifically, when the average VIX value (implied volatility) is lower than SPY’s standard deviation (realized volatility), this serves as a signal to invest in SPXU, otherwise, the position is avoided. Each position is held for one day, after which the procedure is repeated. Again, for this analysis, SPY’s 10-day annualized standard deviation is used as the basis, while the moving average of the VIX is calculated over the most recent 60 days. Benchmark also remains unchanged.
As we can see from the results, this strategy did not deliver highly profitable outcomes, but compared to the performance of the SPXU ETF alone, it achieved improvements across all measures. Therefore, in this case as well, we attempt to modify the strategy with respect to both implied and realized volatility.
We begin again by first fixing the window for implied volatility, calculating the VIX moving average over 60 days, while varying realized volatility, measured as SPY’s standard deviation, across 5, 10, and 20 days.
We can see that the modification led to an improvement when the number of days used to calculate the standard deviation was reduced, specifically to 5 days. Since this strategy operates on the opposite principle to SPXL, where longer windows produced better results, it is logical that here shorter windows work more effectively. In this case, the modification improves investment in SPXU, however, it is still not sufficient to outperform the benchmark.
Now let us test the case where realized volatility, measured as SPY’s standard deviation, is fixed at 10 days, while the implied volatility is varied by changing the window for the VIX moving average, specifically across 10, 20, 30, 40, 60, 80, and 120 days.
We can see that none of the modifications prove to be effective, as the ratios remain negative in all cases, clearly indicating inefficiency. While the performance of SPXU improves somewhat through these adjustments, it still does not come close to surpassing the benchmark.
It’s hard to overcome the performance of SPY by just using the triple-leverage short ETF in the strong bull market (as was the case for the period of 2013 to 2025). All of the variants had results from -8% to +1%. However, what we can expect is that these variants will have a negative correlation with the SPY ETF itself, and signals can therefore be used as entries to employ a selective hedge. The price of the hedge (-8% to +1%) is not excessively high. However, we do not plan to pursue these selective hedging strategies further, as we will dedicate an independent article on the topic of hedging in the future.
Conclusion
Our analysis demonstrates that volatility-based filters can improve the performance of leveraged ETFs by identifying favorable conditions for exposure. For SPXL, comparing realized and implied volatility proved to be an effective approach. The strategy delivered returns significantly above the benchmark, particularly when realized volatility was measured over longer horizons and implied volatility over shorter ones. This combination allowed for stable yet responsive signals, reducing noise while capturing timely opportunities.
In contrast, applying the same framework to SPXU produced weaker results, suggesting that bearish leveraged exposure is more difficult to exploit systematically. While the strategy improved performance relative to holding SPXU outright, it didn’t produce consistently positive results. Still, the performance of the SPXU systematic strategies suggests that we may be able to use this approach as a selective hedge in the future.
Author: Sona Beluska, Junior Quant Analyst
Are you looking for more strategies to read about? Sign up for our newsletter or visit our Blog or Screener.
Do you want to learn more about Quantpedia Premium service? Check how Quantpedia works, our mission and Premium pricing offer.
Do you want to learn more about Quantpedia Pro service? Check its description, watch videos, review reporting capabilities and visit our pricing offer.
Are you looking for historical data or backtesting platforms? Check our list of Algo Trading Discounts.
Would you like free access to our services? Then, open an account with Lightspeed and enjoy one year of Quantpedia Premium at no cost.
Or follow us on:
Facebook Group, Facebook Page, Twitter, Linkedin, Medium or Youtube
Share onLinkedInTwitterFacebookRefer to a friend
