Trading Strategies

The Definitive Guide to Shorting Leveraged ETFs

This post documents some of my research in creating a trading strategy centered around shorting leveraged exchange-traded funds (ETFs). I present the following thought experiment to motivate readers:

  1. Suppose an underlying instrument increases by 25% on day 1 and decreases by 20% on day 2. The return of the underlying instrument is (1 + 0.25) * (1 – 0.20) – 1 = 0%.
  2. Now suppose I construct a leveraged ETF designed to track 3x the daily return of the underlying instrument. The return of the leveraged ETF is (1 + 0.75) * (1 – 0.60) – 1 = -30%.

The intuition here is that in flat but volatile markets, leveraged ETFs exhibit price decay due to the effect that volatility has on cumulative returns.

Below I present a real example of this price decay by plotting SPY and SPXL. SPY is an ETF that tracks the S&P 500, and SPXL is a leveraged ETF that tracks 3x the daily return of the S&P 500. These two securities are analogous to the two securities in the thought experiment.

SPY returned +3% over this time period, but SPXL returned -4% because of price decay inherent in all leveraged ETFs.

SPY, the underlying security, has remained fairly unchanged but with moderate volatility over this time period. Overall return is approximately +3%. SPXL, the 3x leveraged ETF, has returned -4% over the same time period (instead of +9% which some would expect).

A market-neutral portfolio of going long 3x of SPY and going short 1x of SPXL would result in a small profit over this time period.

The price decay observed in leveraged ETFs also exists in inverse leveraged ETFs. For every long 3x ETF out there, there often exists a short 3x ETF. What happens when you construct a portfolio that is comprised of going short equal amounts of both the long 3x ETF and the short 3x ETF? Intuitively, one would expect that the two securities would be mirror images of each other and that the sum of the cumulative returns for both securities would be 0%.

Both the long 3x leveraged ETF and the short 3x leveraged ETF had a negative return over this time period.

But both the long 3x ETF and the short 3x ETF have decreased in value over this time period and going short equal amounts of both would have resulted in an annualized return of around 12%. In addition, this trading strategy is exposed to low risk because shorting this pair of securities is market neutral. The chart above presents the motivation behind this trading strategy, and when I first saw it, it seemed like a very profitable arbitrage opportunity.

The remainder of this post is organized as follows: Section 1 presents background on ETFs and leveraged ETFs for uninitiated readers. Section 2 examines the source of alpha in shorting leveraged ETFs. Section 3 describes exactly how to execute this trading strategy. Section 4 recommends specific leveraged ETF pairs. Section 5 describes the risks to this strategy. Section 6 contains links to further resources.

1. Introduction to ETFs and Leveraged ETFs

An exchange-traded fund an investment fund traded on a stock exchange that owns assets such as stocks, commodities, and bonds and divides ownership of these assets into shares. ETFs are attractive investments because they provide easy diversification, low costs, tax efficiency, while retaining all the features of ordinary stock, including short selling and options.

ETFs track the price of its holdings by trading extremely close to the net asset value of its holdings throughout the trading day. The price of a share of the ETF and the value of its holding are kept close because authorized participants (usually large financial institutions) are allowed to buy or sell shares from the ETF fund manager by giving or receiving the underlying basket of securities.

When the price difference between the net asset value of an ETF share and the underlying basket of securities gets too large, authorized participants can either purchase or redeem shares.  In addition, an ETFs holdings are publicized and investors on the secondary market can also engage in arbitrage by buying or selling the ETF and the underlying basket of securities.

Leveraged exchange-traded funds are a special type of ETFs that are designed to be more sensitive to daily market movements than non-leveraged ETFs. Leveraged ETFs can either be long or short the underlying basket of securities and can be designed to return 2x or 3x the daily return  for long ETFs or 2x or 3x the inverse of the daily return for short ETFs.

2. Source of Alpha in Shorting Leveraged ETFs

The hedge fund I previously worked at could be described as a fundamental global macro fund. It was fundamental in the sense that we believed that there were timeless and universal cause-and-effect linkages that drive financial markets. In other words, we used a first principles approach by identifying  certain unconditional propositions or assumptions that couldn’t be deduced from any other propositions and then reasoned upward from there. When designing any trading strategy, it’s important that any sources of alpha have strong support from first principles.

Let’s return to a hypothetical example to fully understand how volatility affects cumulative returns. Suppose there is a leveraged ETF with initial price $100, and it will increase or decrease by 10% a day with equal probability. Also suppose that this leveraged ETF will trade for two days. What are the four potential outcomes for this leveraged ETF?

  1. $100 * 1.10 * 1.10 = $121 with 25% probability
  2. $100 * 1.10 * 0.90 = $99  with 25% probability
  3. $100 * 0.90 * 1.10 = $99 with 25% probability
  4. $100 * 0.90 * 0.90 = $81 with 25% probability

It is important to note that the expected value for this leveraged ETF is still $100 since the daily expected return is 0%. But even with an expected value that is unchanged, the price has declined below the initial price of $100 in 75% of the outcomes.

The problem with volatility is that you lose more when you’re up and you win less when you’re down — both are bad. That’s why the most common outcome (with 50% probability) is that the ending price is $99. When you move beyond a two-period world, more and more of the outcomes will be drawn towards $0 with a few lucky outcomes with very large prices.

Below I plot a histogram that displays the ending price of 10,000 simulated leveraged ETFs identical to the hypothetical example except that they trade for a period of 1,000 days instead of 2.

Over 95% of simulated leveraged ETFs have an ending price below their initial price.

Of the 10,000 simulated leveraged ETFs, 95% of them ended up with an ending price of less than $100 (which was the initial price). Of these 95%, the majority of them ended up being very close to $0.

But there is an important caveat — 5% of the simulated leveraged ETFs exceeded their initial price, and some exceeded it by a very large amount, and therefore these simulations were not plotted in the histogram. The simulation that got the luckiest ended up with an ending price of $91,087. These extremely lucky simulation outcomes are necessary because the expected mean ending price of all 10,000 simulations is still $100.

This histogram shows exactly why leveraged ETFs are not to be held for long periods of time — the price decay has a significant effect that pulls the most likely outcome towards zero. It shows why financial regulators are considering eliminating leveraged ETFs. It shows why things like the Kelly criterion exist. And it shows why shorting leveraged ETFs can be a profitable trading opportunity. The alpha in this trading strategy stems from the impact that volatility has on cumulative returns.

3. How to Short Leveraged ETF Pairs

Executing the strategy is straightforward.

  1. Identify a pair of leveraged ETFs, ideally from the same ETF fund manager. One of the leveraged ETFs should be the long ETF and the other should be the short ETF.
  2. Short the same notional amount of both ETFs. This means that if the long and short ETFs have different prices, calculate the number of shares to short of each such that you are short the same dollar amount amount of each.
  3. As prices change, the notional amount that you are short each ETF will begin to be unbalanced. For example, if the underlying basket of securities increases in price by a significant amount, the notional amount that you are short the long leveraged ETF will be greater than the notional amount that you are short the short leveraged ETF. For example, suppose you short $10,000 each of the long and short ETF. After some price movement, you could end up being short $13,000 of one ETF and $7,000 of the other ETF.  Re-balance periodically by going short or covering the proper ETF to bring the notional amount into balance.
  4. Cover both shorts to exit the trade after sufficient time has passed.

Regarding re-balancing, there really aren’t any specific rules I can offer. My personal feeling is that once the notional amount begins to be unbalanced by around 20%, you should re-balance. This provides a reasonable cushion while preventing excessive transaction costs.

It is important to re-balance because having an unbalanced notional amount exposes you to either a long or short bias on the underlying. This trading strategy does not have a view on the future direction of the underlying (in fact we are hoping that the underlying remains at a similar price but with high volatility around this price), so you should not be exposed to this bias.

4. Recommended Leveraged ETF Pairs

Selecting leveraged ETF pairs to short should be done based on three criteria: the interest rate charged on borrowed shares (the fee rate), the number of shortable shares available through your broker (usually inversely correlated with the fee rate), and the future volatility of the underlying basket of securities. The ideal leveraged ETF pair would be a pair of ETFs with a low fee rate, high number of shortable shares, and high future volatility.

I obtained lists of leveraged ETF from here and here and looked up the fee rate, shortable shares, and historical volatility in Interactive Brokers. Below is a screenshot containing some of the leveraged ETFs that I considered.

List of shortable leveraged ETFs.

Based on the three criteria established above, I recommend the following leveraged ETF pairs:

  1. NUGT and DUST (Direxion Daily Gold Miners Bull and Bear 3x ETF)
  2. ERX and ERY (Direxion Daily Energy Bull and Bear 3x ETF)
  3. FAS and FAZ (Direxion Daily Financial Bull and Bear 3x ETF)
  4. TNA and TZA (Direxion Daily Small Cap Bull and Bear 3x ETF)
  5. TQQQ and SQQQ (Proshares UltraPro Long and Short QQQ ETF)

All these pairs have relatively low borrowing costs (5% or less per year), are easily shortable, and have moderate to high historical volatility.

5. Risks to Shorting Leveraged ETFs

  1. Shorting pairs of leveraged ETFs is not an arbitrage opportunity. Despite what the initial charts suggest, there is no arbitrage opportunity. Returning to the hypothetical example in Section 2, while the most likely outcome is for leveraged ETFs to be drawn towards zero, there are outcomes in which the leveraged ETF increases significantly in price. This means that there is the risk for extreme loss. Extending the hypothetical example to being short a pair of leveraged ETFs does not eliminate this risk — sustained moves in either direction without much volatility lead to extreme loss. For example, the same SPXL and SPXS ETF pair that was plotted earlier returned about -90% during this 1.5 year time period in which the market trended in one direction:Shorting equal amounts of these two leveraged ETFs would have returned -90% over this time period.
  2. Returns to this strategy are asymmetrical. The returns to this strategy are similar to several options-related strategies that are designed to generate a small positive return with high probability but a large negative return with low probability. In addition, as with any short-selling strategy, maximum losses are unbounded.
  3. The position is not market neutral. Although this strategy involves shorting equal notional amounts of leveraged ETF pairs with periodic re-balancing, this strategy is not truly market neutral. In between re-balancings, the notional amount of each leg will become unequal and necessarily expose you to either a long bias or short bias on the underlying.
  4. Drawdowns can be significant. Be prepared for the potential of large drawdowns. Since this strategy relies on going short on margin, your account must contain sufficient funds to maintain your position.
  5. Transaction costs and costs to borrow can be significant. This strategy relies on re-balancing periodically which incurs additional transaction fees. Costs to borrow (the fee rate) are a function of how hard it is for your broker to locate shares to short and can fluctuate over time. The fee rate may exceed 10% per year for certain leveraged ETFs.

6. Further Resources

This section contains recommended links to additional resources that contain discussion about shorting leveraged ETFs:

  1. and 3x ETFs provide lists of leveraged ETFs.
  2. Shorting Leveraged ETF Pairs: Easier Said Than Done examines the impact of different re-balancing thresholds have on returns.
  3. Another Look At The NUGT-DUST Double Short examines the distribution of returns based on different start dates and with a holding period of one year.
  4. Darwin’s Inverse Leveraged Short ETF Strategy describes some of the risks involved with this strategy.
  5. Shorting Leverage plots some equity curves for common leveraged ETF pairs.

The code underlying this post can be viewed at my Github repository.

I have an email list where I occasionally send updates to readers on the trading systems that I’m developing. If you are interested, please enter your email below.

Get updates from Signal Plot in your inbox.


I am interested in building investing systems, and this blog contains my research and analysis on this topic. I previously worked as an analyst at Bridgewater Associates, a hedge fund that utilizes a systematic, global macro investing style.

14 comments on “The Definitive Guide to Shorting Leveraged ETFs

  1. Have you tried this out? Would be curious how it has gone. NUGT and DUST would be my favorite pair for this since they are so volatile and relatively mean-reverting. Over the past year, DUST is down 27% and NUGT down 55%–but if you shorted them through the last year you’d also find yourself sweating through a 90% rise in DUST with only a corresponding 67% drop in NUGT in mid-December.

    Also just making sure I’m reading this correctly–your leveraged ETF simulations were a stochastic using a 50-50 chance of going up or down 10% each day played out over 1,000 days, right? At first I thought you might have actually used market returns of indices and leveraged them.

    It’s a nice thought exercise that does demonstrate the decay in these leveraged ETFs pretty well (although I don’t know any leveraged ETFs that generally move as much as 10% a day), but I would caution against drawing the conclusion that leveraged ETFs will “generally go down” over long periods of time. I already shared this link in an off-topic comment on your latest bitcoin post, but other folks have done actual market-based simulations to show that leveraged ETFs generally perform well over long periods of time. Not 2x or 3x as well as their indices, but generally better than their indices (and not dwindling inevitably to zero!). Yay for the market’s positive skew!

    Leveraged inverse ETFs, however, may go towards zero over time–I will investigate that one. Should be a pretty easy tweak to my model to find out, actually….

    • OK, that did not take long. In my 30 year simulation of inverse leveraged ETFs using real data (from Yahoo! Finance, so it contains some errors), it’s not looking so hot for those inverse leveraged ETFs.

      Inverse leveraged ETF CAGRs from 30 year Historical Simulation:
      TMV (-3x long term treasuries): -25%
      SQQQ (-3x NASDAQ 100): -43%
      SPXU (-3x S&P): -38%
      SICK (-3x healthcare): -30%
      FAZ (-3x financials): -44%
      DGLD (-3x gold): -23%

      Ouch! Now to compare with their bullish brothers over the same time-frame:

      TMF: 18%
      TQQQ: 10%
      UPRO: 18%
      CURE: 28%
      FAS: 8%
      UGLD: 1%

      Clearly wouldn’t be market neutral, but just shorting inverse leveraged ETFs doesn’t seem like a bad long-term play. Even after shorting fees you’d have generally done better than buying their bullish brothers (with the exception of SICK/CURE).

      • Thanks for your comment. I did try the shorting leveraged ETF pairs for a little bit, but I found that even through Interactive Brokers, it’s hard to maintain short positions for long periods in these ETFs. Interactive Brokers automatically bought in some of my positions, even though I was trading in small size. So I decided to stop using this strategy.

        Thank you for sharing those links and your research about leveraged ETFs. You’ve given me a lot to think about.

        • My pleasure–looking forward to your future posts!

          I’ve been working on a balanced portfolio including some leveraged ETFs.


          That’s a particularly aggressive example (my actual portfolio is more conservatively tilted towards bonds, utilizes less overall leverage). A sharpe optimized portfolio over the last 30 years would be quite heavy in long term treasuries (but of course that data set starts around the beginning of the great bond bull market).

          Rebalancing is critical to bank the volatility gains–these portfolios require more frequent rebalancing than most to get optimal results (about 8 rebalances a year works well).


  2. Great article.

    Can we optimize based on historical data what percentage would be the perfect percentage to rebalance? Say with your 20% rule, if there is a 20% difference between the notional balances of each ETF (i.e. -$12k and -$10k), that would trigger a balance. Or also maybe put in some kind of rule that would automatically rebalance after say 1 month, regardless of the percentage difference.

    Instead of eyeballing these rules, I’m sure we can create a model that will optimize for this, as well as figure out the best strategy when rebalacing does take place. Would we increase the short balance of the smaller investment (i.e. increase the -$10k balance to match it’s corresponding $12k peer?). This would effectively have us keep increasing our exposure on the table, even though we are technically market neutral. Or would we want to decrease the cover and increase the $-12k to $-10k.

    I really want to implement this to see how it goes. I’d like to backtest it as well. Any suggestions on how to do this?


    • Try Googling “quantopian shorting leveraged etfs”. There’s some people on Quantopian that have implemeneted backtests with a certain criteria. You may be able to modify their code and implement a grid search where you iterate over a reasonable set of criteria and find out what works best. Ideally, you would want to repeat this process over multiple leveraged ETF pairs.

      Backtesting this strategy is kind of hard though because your results vary wildly depending on exactly what day you start the strategy. Try going on Google Finance and compare two leveraged ETF pairs and you can see what I mean. So your backtest would ideally randomize the start date instead of starting on some arbitrary day.

      If you want to actually use this strategy, I actually recommend using options now instead of using the underlying. I tried trading this strategy on my personal account, but my broker bought in one leg of my short position without warning before I could profit. Had that not happened I would be up by quite a bit though.

      It’s also a bit of pain to deal with short sale restrictions, varying short fee rates, and wondering if there are shares available to short when you need to rebalance.

      This strategy sounds really awesome in theory, but in practice it’s less awesome. Then again, the best performing mutual fund in 2015 used this strategy (see link below), so who knows.

  3. Pingback: “These Aren’t The Gains You’re Looking For” (Leveraged ETFs) – Throwing Good Money After Bad

  4. One could conceivably write long-dated calls or buy in-the-money puts (also long-dated) and take advantage of this. I wonder if that’s more efficient? i don’t have the means to test options for this concept unfortunately. Your post is very timely though! I was just about to write something related to leveraged ETF decay. I’ll try to remember to refer back to this post!

  5. It would be interesting to explore how this strategy would fit into a portfolio. For example, does the return pattern correlated with certain systematic factors? Is it possible to characterize the environment in which the drawdowns occur and relate that to drawdowns in other strategies?

    Nick de Peyster

    • It’s definitely a mean reversion play, so it’s anti-momentum. Also it benefits from high volatility, so it’s anti-low volatility (some research suggests low volatility is a factor that can be exposed). Drawdowns will occur when the market trends in one direction, either positive or negative. The strategy performs best in highly volatile but trendless markets. Thank you for continuing to read and comment on my posts, I really appreciate it.

  6. Pingback: Quantocracy's Daily Wrap for 01/30/2017 | Quantocracy

  7. In a world with lots of superstars, you write one of my favorite financial blogs. Great to have you posting again.

    I’m not very smart, and leveraged ETFs boggle my mind. In your histogram, the returns are negatively skewed with 95% of instances losing money. But if the 5% of winners win enough to compensate for the losses with an overall expected value of 0, how can this be exploited through shorting? Won’t you break even in the long run? What am I missing?

    • Thank you for the kind words. You’re right in that in that simulation the overall expected value is zero and you would break even in the long run.

      The idea behind shorting both the long 3x and the short 3x leveraged ETFs is that you are market neutral (until the two sides become unbalanced), and if you rebalance at the correct times, you can remain market neutral.

      This means that in the 5% of the time that the underlying trends in one direction, you’ll be largely immune from that movement. You’ll be close to break even in 5% of the time and make money the other 95% of the time.

      In practice, this strategy doesn’t work as elegantly due to shorting fees, the threat of your broker closing out your position, knowing when to rebalance, and so on, so the threat of loss is still real. But I have heard of professional investors making money using this strategy.

Leave a Reply