⚠️ Leveraged ETFs ⚠️
Every time we hear about leveraged ETFs, a compliance officer appears out of thin air to remind us that these products are for short-term trading only and not for buy-and-hold investors. The more thorough among them will spew off terms like “daily leverage resets” or even “volatility decay” and point to some 2x leveraged natural gas fund that lost over 75% on both the long and short side.
While I’m sure this speech keeps some people out of trouble (especially the sponsors of these products in the eyes of the regulators), wholesale writing off buy-and hold investors is taking it too far, as I will seek to demonstrate. On the way there, we’ll touch on the aforementioned terms and create some examples along the way.
For those new to leveraged ETFs, they are investment products that seek to mirror the returns of an underlying index at some multiple of the (usually) daily return. As of December 31, 2023, the largest product (by AUM) was the ProShares UltraPro QQQ ETF, which seeks to deliver 3 times the daily returns of the underlying index — the Nasdaq 100. Most sponsors of such funds also offer negative “gearing” (the leverage level), such as -1x, -2x, or -3x.
So let’s start off where most foreboding ends — discussing daily leverage resets. In order to deliver a multiple of the daily return of an underlying index, the fund must borrow money and purchase assets that track the value of the index. In practice, most of these products purchase futures or engage in swap agreements with embedded leverage. For example, a single contract of the E-mini Nasdaq 100 has a contract unit of $20 x the Nasdaq 100 Index while requiring relatively little cash as collateral. This implicit leverage allows the portfolio manager to increase or decrease the collateral backing these contracts in order to target the specific level of leverage desired.
If 3x leverage is targeted, a nominal $100 dollar investment would require $200 in borrowing, resulting in $300 notional exposure to the underlying index (3x the $100 investment). If the index returns 1% that day, the fund is up $3 ($300 x 0.01), closing the day with $103. However, if we started the next day with the same $200 in borrowed funds, we would only have $303 notional/$103 nominal, or 2.94x leverage, so we need to borrow more to get back to the target gearing. This daily reseting of leverage ensures we closely match the daily index return, but demonstrates that replication doesn’t hold constant for periods beyond 1 day.
As stated before, this characteristic leads to the wholesale warning against buy and hold, but what is the actual impact? To check, we’ll backtest the TQQQ ETF to see how closely it tracks the underlying index, which we’ll proxy as the QQQ ETF. Because we want to isolate the impact of the daily leverage reset, we’ll invest 1/3 in TQQQ and 2/3 in a cash equivalent, the SHY ETF. In order to make this realistic for a buy and hold investor, we’ll simulate quarterly rebalancing to get back to these target weights. Here’s the result from February 2010 through September 2023:
Despite all the warnings, the results are fairly similar. The excess return annualizes to 0.42% and rebalancing quarterly was responsible for 0.54% higher annualized returns. subtracting that out makes the remaining difference even more negligible. Said another way, had we invested $10,000 dollars at inception and compared our ending capital, the un-leveraged TQQQ investment would have earned $263 more than the index. That’s not a very big difference after more than a decade.
A fair retort to this backtest is that it only captures what happened historically, as the path of the returns could have looked way different. So let’s model the ending market value of a 17% return, 20% volatility product (per the actual backtest) with 3x gearing that has been “unleveraged” by investing 2/3 in cash (assumed to have zero return for simplicity). We’ll compare this return to a 100% investment in the underlying index — simulated over 252 days — and repeat the simulation 1,000 times.
Here’s a single run of the simulation, for reference:
Now here’s a histogram of the excess market value for the unleveraged geared product (referred to as “Product” from now on) versus the underlying index (“Index”) across 1,000 trials.
It turns out, the median ending market value of the Product is $164 lower than simply investing directly in the Index. This would represent a 1.64% drag on returns. However, a glance at the above histogram reveals a strong positive skew to this distribution. In other words, we occasionally observe outcomes where really large outperformance comes from the Product relative to the Index. These less common outcomes are so strong that the average performance of the product is a positive $428 in excess of the Index, per year.
The single-simulation line chart above is instructive, as it demonstrates where the Product tends to deviate from the Index. The long, upward-trend of the index from day 75 to 90 benefitted the Product, which was constantly resetting to higher levels of leverage through much of that run. This gave the Product’s return profile a positive convexity versus the index. Interestingly, this leverage reset also helps in a downward market, as the product reduces leverage daily to get back to targets. The worst outcome for these products, in contrast, is highly volatile periods that don’t trend, aka mean reverting periods. The below charts plot the excess market value in each of these three scenarios:
- Daily returns of 0.5% with no volatility:
2. Daily returns of -0.5% with no volatility:
3. Daily returns oscillating back and forth between 0.5% and -0.5%.
This is likely counterintuitive to most people, but we’ve just demonstrated that you’d literally be better off investing with a leveraged product in a consistent down market than the underlying index due to the convexity it demonstrated — holding notional exposure constant.
But wait, we ran this simulation using 17% returns. What about a more realistic scenario? Say, 8% returns? It’s actually quite similar.
In this case, the positive skew isn’t quite as high, but there’s still positive expectancy with an average excess market value of $72 for the Product over the Index, annually. After fees, this likely doesn’t stay positive; but it certainly isn’t the train-wreck everyone makes it sound like. Plus, uninvested cash would also be earning a positive yield in real life. So far, this analysis has been focused on the daily leverage reset of these products, but there was another cool sounding term referenced above, so let’s not overlook it. It’s time to talk about ⚠️ V o l a t i l i t y D e c a y ⚠️.
“Volatility decay” is a term I try to work into casual conversation daily due to how mysterious and genteel it makes me sound. Not to be anticlimactic, but volatility decay isn’t a property that emerges from the use of leveraged products — it’s just more noticeable with them. Even everyone’s favorite ETFs, like SPY and QQQ, are subject to volatility decay. The reality is that no risky asset can escape the erosive properties of this phenomenon. When an asset experiences ups and downs over time, the hurdle to a positive return becomes higher. As a result, any asset that doesn’t take a direct path to its final price has a realized holding period return that is lower than the sum of its periodic returns. This is a property of ALL assets. To pay any special attention to this attribute of geared products is simply to observe that they exhibit more volatility that their underlying index — which is literally their purpose in life.
Is investing 100% of your portfolio in TQQQ a good idea? I’m not here to opine on that. But if an asset with the same average return and volatility as TQQQ existed without any leverage, I’m simply arguing it would be a roughly equally good or bad thing to put 100% of your portfolio into it. Note, such a product would have resulted in an annualized standard deviation of 60% over the last decade. Yeehaw. The takeaway here is that sizing an investment in your portfolio is just as important with a geared ETF as it is with any other investment — leveraged or not.
We could stop here, but what about those 2x leveraged natural gas funds everyone talks about as the final nail in the coffin for these products? Let’s take a look.
Yikes, that really is bad. Let’s dig deeper into the worst of the two, which is the 2x leveraged long product. To start, let’s look at the index, which I’m proxying as an unleveraged Natural Gas ETF, UNG. It has produced an average return and volatility of about -16% and 47%, respectively. Plugging that into our simulator gives the following results:
As a quick aside, a 2x geared version of an index whose performance really sucked is, unsurprisingly, also going to suck. Above is one trial of the simulation that exemplifies this nicely. Yet, if we unleverage the geared product (“Product”) back to the same notional level as the index (“Index”), we get this result again:
The Product manages to outperform the index due to it’s constant deleveraging during the long down-trending environment. Here’s a look at all 1,000 trials.
The volatility is higher and so is the distribution of outcomes. But just like the other two examples, the massive positive skew results in positive expectancy of $129 per year. I find it hard to fault a geared ETF that actually outperformed its index on a notional basis when it seems like the real takeaway should be “Don’t invest 100% of your money into a natural gas ETF!”
We’ve demonstrated how these ETFs perform comparably on a notional basis, so we should talk about what it means to use them in such a way. Using TQQQ as an example, how would we size the product to obtain the same notional exposure as if we were using QQQ, the unleveraged version of the ETF? It’s fairly simple: If we wanted 30% of a portfolio invested in the Nasdaq 100, we could either put 30% of the portfolio in QQQ, or 10% in TQQQ — leaving 20% in cash. Both accomplish roughly the same thing (see above article for more details on “roughly”). What’s great about the second option is that you now have 20% of your portfolio in cash — increasing your liquidity during tight markets or allowing you to add more assets you previously didn’t have room for. Using the capital efficiency of leveraged products unlocks extra space to add diversifiers, a technique the folks at Resolve Asset Management refer to as “return stacking”.
For an example of how to utilize leveraged ETFs in a portfolio, you can check out a mock portfolio I built over at PortfolioTree.com I call the Accumulator’s Inflation-Hedged Portfolio, which uses leveraged ETFs to make room for commodities, bonds, and managed futures, while still maintaining an equity beta close to 1.
So, next time you see a compliance officer being summoned on behalf of a geared ETF, remember that it isn’t leverage that gets you in trouble — it’s concentration. Being mindful of the total notional exposure in your portfolio will go a long way towards finding appropriate ways to utilize more capital efficiency as a not-so-speculative buy-and-hold investor.
