The Case for Leveraged ETFs to Lower Risk
Leveraged ETFs often have a bad reputation. Take a 2x leveraged ETF, which is designed to deliver twice the daily return of a given index. Over the long term, however, its performance can differ considerably from simply doubling the return of the underlying index. In fact, there are periods where the index ends up with a positive return, while the leveraged ETF shows a loss. This divergence is primarily caused by what's known as decay - a result of daily rebalancing, compounding effects in volatile markets, and the inherent costs of using leverage. This issue is more pronounced during choppy sideways markets. To see this in action, you can compare leveraged ETFs like SSO (2x the S&P 500) or SPXL (3x the S&P 500) with SPY (which tracks the S&P 500) over random 3, 5 or or 10 year periods. You’ll often find that the leveraged ETF underperforms the expected multiple of the index’s return. The effect is even starker in more volatile indices or sectors. For example, compare TNA vs. IWM or NUGT vs. GDX.
The charts below show the ratio of SPXL returns to SPY returns over the past 10 years, calculated using 1-day, 20-day, 50-day, and 200-day rolling periods. In theory, this ratio should be close to 3 but as the timeframe increases it becomes noticeably more erratic. In particular, the last chart - based on 200-day rolling periods - reveals a large number of outliers, with the ratio staying within the 2.5 to 3.5 range only 40% of the time.
It just so happened that in 2013 I came across a Stanford paper titled The Floor-Leverage Rule for Retirement which describes an intriguing investment strategy: allocate 85% of assets to secure a sustainable lifetime income floor, and invest the remaining 15% in equities with 3x leverage. Surprisingly, the paper concludes that this approach is actually safer than many traditional strategies, such as William Bengen’s 4% safe withdrawal rule or split-account methods. While the strategy is designed for retirement spending, it sparked my curiosity about the broader role leveraged ETFs might play in long-term investing. At the time, I had been growing more interested in risk management and had already read Nassim Taleb’s The Black Swan, where he describes what he calls a barbel strategy: “If you know that you are vulnerable to prediction errors, and accept that most risk measures are flawed, then your strategy is to be as hyper-conservative and hyper-aggressive as you can be, instead of being mildly aggressive or conservative.”. I was also exploring risk-parity-based strategies like Harry Browne’s “Permanent Portfolio” and Ray Dalio’s “All Weather” portfolio.
And this led me to start experimenting with ideas like: why not invest half as much in a 2x leveraged ETF and hold the rest in safer assets? For example, instead of a traditional 60/40 portfolio (60% equities, 40% fixed income), what if I invested just 30% in a 2x ETF and increased my allocation to safe assets to 70%? Or even 20% in a 3x ETF, freeing up even more capital for conservative holdings?
The backtest below spans from the inception of SSO in 2006, allocating 50% to SSO and holding the remaining 50% in cash earning zero interest. SPY is used as the benchmark. The test uses weekly data and unadjusted closing prices. I chose these parameters (zero return on cash, no dividend reinvestment, and a weekly timeframe) to keep the analysis realistic and grounded. On average, the strategy required just 2.5 rebalancing trades per year and delivered some alpha without increasing overall risk. While the drawdown during the GFC was slightly deeper, the strategy’s overall risk profile remained stronger:
50% in SSO, rest in 0% cash, 06/23/06 - 05/30/24
I then tested a similar setup: 33% in SPXL and the remaining 67% in cash, going back to SPXL inception (Nov 2008). A bit more trades (6.64 / year) which is expected, but overall, the backtest produced even a better alpha than the 50% in SSO version:
33% in SPXL, rest in 0% cash, 11/07/08 - 05/30/25
The backtests above assumed all reserves capital was held in zero-interest cash, mainly to isolate the performance of the core strategy without any contribution from the reserve allocation. However, in practice, that freed-up cash can be put to better use. Rather than thinking of it as idle cash, one can treat it as a "safe basket" that might include high-interest savings accounts, gold, select currencies, managed futures, or even market-neutral funds. Here is an example of what an overall portfolio would look like:
A simple approach to allocating the reserves is to split them between short term treasury and gold. For example, in a modified 60/40 portfolio, one can allocate 30% in the 2x leveraged market ETF SSO (or 20% in the 3x leveraged one SPXL) and the remaining split equally between a 3-month treasury ETF (e.g. BIL) and gold (e.g., IAU). This results in an effective allocation of 30% SSO, 35.5% cash, and 35.5% gold. There is quite of bit of alpha in this, and it is partly due to gold’s stellar performance, but I'll leave it to the reader to draw their own conclusions about the merits of this setup.
30% in SSO, 35.5% in cash earning 2.5%, 35.5% in gold (IAU), 06/23/06 - 05/30/24
Similarly, this is what a 20% allocation to SPXL (with 40% in cash and 40% in gold) would’ve produced over the last 16.5 years:
20% in SPXL, 40% in cash earning 2.5%, 40% in gold (IAU), 06/23/06 - 05/30/24
I then began a series of backtests across different time periods, even going so far as to synthesize 2x and 3x leveraged data stretching back to the 1950s. The results were consistently compelling: over the long term, and more often than not, this strategy delivered superior risk-adjusted returns compared to fully investing in the underlying index alone.
The last backtest in this post is tailored for Canadian investors. Instead of having a 100% allocation to the TSX 60 index (e.g. TSE:XIU), consider this alternative portfolio: 50% in CNDU, with the remaining 50% divided equally among cash earning 2.5%, gold (CGL.C), and a market-neutral fund such as the Picton Mahoney Alternative Fund (TSE:PFMN). This results in an effective allocation of 50% CNDU / 16.67% cash / 16.67% CGL.C / 16.67% PFMN. This chart below shows the results of a ~5-yr backtest (07/24/20 - 05/30/25 which is all the data Google Finance has for for these funds). The high alpha, Sharpe and Calmar ratios justify the slightly deeper (yet shorter drawdown):
50% CNDU / 16.67% in cash earning 2.5% / 16.67% CGL.C / 16.66% PFMN, 07/24/20 - 05/30/25
I will conclude this post with a few more observations about using leveraged ETFs that are worth highlighting, beyond enhancing risk-adjusted returns:
Limited downside: Allocating, for example, 20% of a portfolio to a 3x leveraged ETF caps the maximum potential loss to that 20%, effectively placing a floor on overall portfolio losses.
Cash reserve independence: The remaining uninvested capital can - and ideally should - be allocated to assets independent of the stock market. This ensures that, even in extreme scenarios like a decade-long market closure, the portfolio remains viable.
Sequence risk mitigation for retirees: For those in retirement who are making withdrawals, this approach allows them to avoid liquidating equity holdings during downturns, helping to reduce sequence-of-returns risk.
