Dorin Puha/iStock via Getty Images
With equity markets making daily new highs led by the ‘Magnificent 7’ stocks and bearish market pundits comparing NVIDIA Corp.’s (NVDA) meteoric rise with Cisco (CSCO) during the dot-com bubble era (Figure 1), should contrarian investors short the semiconductor sector with the Direxion Daily Semiconductor Bear 3X Shares ETF (NYSEARCA:SOXS)?
Figure 1 – Pundits comparing NVDA to CSCO (marketwatch)
Unfortunately, I believe the SOXS ETF is an investment vehicle even contrarians should avoid. Inverse levered ETFs like SOXS suffers from significant tracking error due to positive convexity and volatility decay. In fact, since volatility typically rises during crashes, even if investors are correct in their macro view and semiconductor stocks crash, the SOXS ETF will very likely underperform its theoretical returns over long holding periods.
Fund Overview
The Direxion Daily Semiconductor Bear 3X Shares ETF seeks daily returns that are 3x the inverse returns of the NYSE Semiconductor Index (“Index”). The SOXS ETF achieves its levered returns by holding total return swaps with large investment banks like Goldman Sachs that are reset nightly (Figure 2).
Figure 2 – SOXS portfolio holdings (direxion.com)
Inverse Levered ETFs Only Work As Intended In The Short-Run
Investors considering the SOXS ETF should be mindful that levered ETFs only work as intended on very short time frames, often measured in hours or days. When we consider holding periods beyond the 1-day horizon, volatility and convexity will start to introduce tracking error.
For example, if we started off with $100 invested in the SOXS ETF, if the index loses 5% on day 1, the position will grow to $115 (3 times 5% return). If the index declines by 5% again on day 2, the position will grow to $132.25, more than three times the theoretical 2-day compounded return of 10.25% or $130.75. This effect is commonly called ‘positive convexity’, i.e., the position grows as the bet is winning.
Conversely, if the semiconductor index returns were consecutive +5%, then the initial investment will end up at $72.25, versus three times the 2-day compounded loss of 9.75% or $70.75.
Finally, if the index returns a volatile -5% followed by +5%, investors end up with $97.75, significantly less than the three times the 2-day compounded loss of 0.25% or $99.25. This tracking error is commonly called ‘volatility decay’
While tracking error from ‘positive convexity’ and ‘volatility decay’ may seem small on a day-to-day basis, when compounded over many days, the effects can be very significant.
For example, the iShares Semiconductor ETF (SOXX) is a passive ETF that also tracks the NYSE Semiconductor Index. Over the past year, the SOXX has galloped to a 57.7% total return. So a theoretical 3x return would mean a loss of 173%. However, the SOXS has lost only 80.9% due to positive convexity we mentioned above (Figure 3).
Figure 3 – SOXS vs. SOXX returns, 1 year (Seeking Alpha)
Traders looking at the SOXS ETF should make sure they fully understand the tracking error risks involved with levered ETFs by consulting these FINRA and SEC warnings.
SOXS Will Likely Underperform During A Crash As Volatility Rises
Although the SOXS ETF suffers from significant tracking error as detailed above, swing traders may be attracted to the profit potential from its leveraged exposure on the NYSE Semiconductor Index.
For example, if traders have perfect foresight and are able to predict a -60% 1-year drawdown in the index (perhaps from a major recession) with low volatility (basically stocks falling in a straight line), then the SOXS ETF may return over 1,300%, far above the theoretical 180% returns (Figure 4).
Figure 4 – Theoretical returns of -3x leveraged ETFs (SOXS prospectus)
However, these figures are not realistic, since volatility typically skyrockets in a crash scenario. For example, in 2022, the SOXS ETF only returned 21.4% while the SOXX ETF declined by 36.2% (Figure 5).
Figure 5 – SOXS vs. SOXX, 2022 (Seeking Alpha)
Performance during the COVID crash was even worse. From January to April 2020, the SOXS ETF lost 52.6% while the SOXX ETF declined by -8.0% (Figure 6). So during major crashes, SOXS did the complete opposite of expectation.
Figure 6 – SOXS vs. SOXX, Covid Crash (Seeking Alpha)
The main reason for SOXS’s underperformance vs. expectations is due to volatility. Specifically, if we look at historical volatility of the SOXX ETF as a guide, we can see that 30-Day realized volatility for the SOXX ETF rose to over 50% during the 2022 bear market, and over 100% during the COVID crash (Figure 7).
Figure 7 – SOXX historical 30D volatility (alphaquery.com)
As volatility rises, the tracking error from volatility decay also increases dramatically, to the point where it can result in SOXS losing money even as the underlying index declines.
Are We Witnessing A Bubble?
While market naysayers include myself are drawing similarities between the current market environment and the dot-com bubble, it is also important to put the comparisons in perspective. For example, if we measure the current ‘Magnificent 7’ bubble, its rise has been rather tame compared to the gains in some other historical bubbles (Figure 8). So either this is a mini-bubble, or there are still more gains to come.
Figure 8 – Comparison of financial bubbles (Bank of America)
More importantly, there are key fundamental differences in the nature of the current rally compared to past bubbles. For example, while the dot-com bubble was about the meteoric rise of profit-less internet companies, currently the ‘Magnificent 7’ are collectively highly profitable and trade for ~30x earnings. While this is extended, it is not yet outrageously so.
if investors are concerned about valuations, I would recommend they look to reduce their holdings, switch to funds or companies with lower valuations, or play bearish bets using put options. At least with put options, the amount of capital at risk is well defined as you cannot lose more than the premiums upfront.
The main thing with bubbles is that timing the peak of a bubble is like buying the lottery. Sure, someone always wins, but it is highly unlikely to be you. We can try to profit from the bubble on the way up, try to reduce exposure to the bubble in our portfolios when valuations stop making sense, and finally, hedge with protective puts when it looks risky, but shorting a bubble with a decaying product like the SOXS is like financial suicide.
Conclusion
In conclusion, I believe investors should avoid inverse levered ETFs like the SOXS, as it only delivers on its stated levered returns over short time horizons. Holding periods longer than 1-day will see significant tracking error due to positive convexity and volatility decay.
In fact, since crashes are typically associated with rises in volatility, the SOXS ETF is basically designed to underperform its theoretical returns over long holding periods, even if the underlying index goes into a deep drawdown.
Even though I am personally cautious on semiconductor stocks due to their massive run-up and extended valuations, I would rather express my bearish views using put options and put spreads. At least with long put options, I have clearly defined risk, whereas inverse ETFs can decay perpetually.
I rate the SOXS ETF an avoid (or sell).
