Several related recent historical backtests follow here (actual UPRO and TMF results, not the sim data) to address a couple a points at once. One is the oft-made claim that TMF and the excellent adventure will fail in an era of rising rates, and the other point is just to numerically demonstrate what mrspock and others have been pointing out about quarterly rebalancing providing a lot of the return.
is a brief 37-month period from December, 2015 through December, 2018. Why that seemingly cherry-picked and short range? Because the Fed raised rates nine times in those 3+ years, 25 bp at a time, from near zero. Below is how 100% TMF, 100% UPRO, and 55/45 UPRO/TMF with quarterly rebalancing (calendar-aligned) fared:
Even in a time of rising rates TMF in isolation eked out a small positive return. The overall 55/45 returned (2.9%) less annually than the pure UPRO did, but 11.75% more than pure TMF. The rebalancing made the 55/45 adventure a lot more successful than an unbalanced weighted average would have been.
Extending the hypothetical adventurer's portfolio that had begun in December, 2015 through July, 2020 (by March, 2020, interest rates had returned to the near zero they had been at prior to December, 2015) produces an excellent example
of the return boost from the quarterly rebalancing:
TMF and UPRO had similar good results over this 4⅔ period, but the 55/45 blend rebalanced quarterly beat each individual component by over 10% CAGR. With a good deal less volatility, a lower max drawdown, and much higher Sharpe and Sortino Ratios.
The last example demonstrates what the same strategy as above but without any rebalancing
would have looked like, again from 201512 through 202007:
The non-rebalanced 55/45 would still be preferable to either in isolation as far as having noticeably less volatility and a better max drawdown than UPRO or TMF alone, but the return is just a 55/45 weighted return of UPRO's and TMF's returns. The rebalancing juices it up from a $14,350 profit in less than five years to $26,805.