$40k In UPRO on its Inception Date (‘09) Would be Worth $1.1 Million Today
Equal amplification of risk and return: enticing!
In general, pursuing larger raw returns requires sacrificing risk-adjusted performance. If you want more growth than a short-term bond fund offers, you can invest in an intermediate- or long-term bond fund, but you have to accept a worse risk-return profile. If you want stronger growth yet, you can invest in equities, but your Sharpe ratio will suffer.
It’s exactly what you’d expect: you sacrifice raw growth for a higher Sharpe ratio.
By the way, if equities had higher Sharpe ratios than bonds, it wouldn’t make sense to increase allocation to bonds as you near retirement. You’d be better off adding a cash allocation, to reduce risk as much as you like while retaining your same Sharpe ratio.
When people think about leveraged ETF’s, they usually only think about their ability to amplify returns. But these instruments in fact have a remarkable characteristic: they amplify returns without sacrificing risk-return.
Well, you have practical issues like expense ratio and tracking error, but for the most part, the Sharpe ratio of a well-running leveraged ETF is only slightly lower than the underlying.
UPRO performance since inception
ProShares UltraPro S&P 500 (UPRO) is a popular leveraged ETF that aims to multiply daily gains of the S&P 500 by a factor of 3. It was introduced in June 2009 and has a 0.92% annual expense ratio.
Let’s look at performance since inception for UPRO, alongside the S&P 500 ETF SPY and a theoretical “ideal” version of UPRO which multiplies SPY gains by exactly 3.
Roughly 27x growth in a shade over a decade, not bad! And that outrageous number is a considerable underperformance compared to an ideal 3x SPY, which would have had 36x growth. It’s a pretty big underperformance, corresponding to ~25% lost profits.
As you can see, the Sharpe ratio is worse than SPY’s, but not by too much.
if you had invested $40k in UPRO the day it was introduced, it’d be worth over $1.1 million today. (The magic number for exactly $1 million is $36,100.)
Good timing + bull market, but not unprecedented
Obviously, a big part of UPRO’s explosive growth is good timing. It was introduced right after the financial crisis of 2008 and rode a 10-year tailwind.
Yet historical S&P returns suggest it isn’t entirely unusual for a 3x daily ETF to turn $40k into a million in 10 years. To illustrate this, consider 10-year growth of $40k starting the beginning of each year from 1980 to 2009 (this time swapping in VFINX for SPY since its historical data goes back further):
To avoid overoptimistic estimates, I used UPRO’s alpha with respect to VFINX over UPRO’s lifetime (-0.0091% daily) for the synthetic 3x VFINX gains.
Of 30 full decades, 3x VFINX outperformed VFINX 20 times (66.7%), eclipsed the $1 million mark 4 times (13.3%), and even passed $3 million once.
Unsurprisingly, VFINX’s growth potential is much smaller. In the highest growth decade (1989–1998), the final balance for VFINX was just 1/14th that of 3x VFINX ($229k vs. $3.4M).
Max drawdowns for 3x VFINX were nasty: median 90.8%, range 49.3%-97.0%.
Leveraged ETF’s are unique in their ability to amplify returns without sacrificing Sharpe ratio (at least not much). They’re capable of generating life-changing returns.
I would never encourage somebody to invest in a fund that could easily wipe out 95%+ of their savings. But it’s an interesting sort of lottery ticket, one that carries a legitimate chance of turning modest savings (<$50k) into several million over a very short time frame. It’s also a good lottery ticket in that if you don’t hit the jackpot, you still might get something: over the past 30 years, a ~67% chance at outperforming the S&P.
Compared to trying to find the next AMZN, investing in a leveraged ETF seems less speculative and more likely to result in success. It’s a bet on high beta paying off, as opposed to a bet on one particular company beating ~99.99% of the others. Bull markets happen all the time, while picking a top-0.01% stock happens exactly 0.01% of the time.