Home › Explore Investing Strategies › Questions and answers › Leveraged ETF rules
HFEA (55% UPRO, 45% TMF, rebalanced quarterly): did it still beat SPY after 2022?
HFEA (55% UPRO, 45% TMF, reset quarterly) since 2009 vs. SPY, UPRO alone and 60/40, with 2022, when stocks and Treasuries fell together.
Summary
$10,000 split 55% UPRO and 45% TMF in July 2009 and reset every quarter was worth $367,740 at the as-of date, 23.1% a year. The same money in SPY reached $110,720 (14.9% a year), and in UPRO alone $1,235,026.
2022 was the turning point. SPY returned −18.2% and TMF −72.6% as stocks and long Treasuries fell together, and HFEA returned −63.8%. HFEA fell 70.6% from its December 2021 high to October 2023. At the as-of date it was still 34.7% below that high.
Over the last five years HFEA returned −6.7% a year and SPY 11.9%. HFEA beat SPY in 70% of the 149 five-year windows.
The rule behind this question: All Weather and the Permanent Portfolio: four-season allocations tested with ETFs
$10,000 once, full period
| Case | End balance | Total return | Annual return | Max drawdown | 2022 |
|---|---|---|---|---|---|
| HFEA: UPRO 55%, TMF 45%, quarterlyJul 2009 to Oct 2026 | $367,740 | +3,577.4% | 23.1% | −70.6% | −63.8% |
| Same mix, monthly | $248,110 | +2,381.1% | 20.4% | −70.2% | −62.9% |
| Same mix, yearly | $196,438 | +1,864.4% | 18.7% | −69.5% | −63.8% |
| UPRO 40%, TMF 60%, quarterly | $144,488 | +1,344.9% | 16.7% | −75.4% | −66.2% |
| Hold UPRO | $1,235,026 | +12,250.3% | 32.0% | −76.8% | −56.8% |
| Hold TMF | $4,313 | −56.9% | −4.7% | −94.3% | −72.6% |
| Hold SPY | $110,720 | +1,007.2% | 14.9% | −33.7% | −18.2% |
| SPY 60%, IEF 40%, quarterly | $54,242 | +442.4% | 10.2% | −21.3% | −16.7% |
Last 10 and 5 years
| Case | End balance | Total return | Annual return | Max drawdown |
|---|---|---|---|---|
| 10 years, HFEA: UPRO 55%, TMF 45%, quarterlyNov 2016 to Oct 2026 | $36,746 | +267.5% | 13.9% | −70.6% |
| 10 years, Hold UPRONov 2016 to Oct 2026 | $127,473 | +1,174.7% | 29.0% | −76.8% |
| 10 years, Hold SPYNov 2016 to Oct 2026 | $41,435 | +314.3% | 15.3% | −33.7% |
| 10 years, SPY 60%, IEF 40%, quarterlyNov 2016 to Oct 2026 | $25,247 | +152.5% | 9.7% | −21.3% |
| 5 years, HFEA: UPRO 55%, TMF 45%, quarterlyNov 2021 to Oct 2026 | $7,059 | −29.4% | −6.7% | −70.7% |
| 5 years, Hold UPRONov 2021 to Oct 2026 | $20,995 | +109.9% | 16.0% | −63.9% |
| 5 years, Hold SPYNov 2021 to Oct 2026 | $17,501 | +75.0% | 11.9% | −24.5% |
| 5 years, SPY 60%, IEF 40%, quarterlyNov 2021 to Oct 2026 | $13,784 | +37.8% | 6.6% | −21.3% |
Calendar-year returns, distributions reinvested
| Year | HFEA | UPRO | TMF | SPY | SPY 60%, IEF 40% |
|---|---|---|---|---|---|
| 2010 | +46.5% | +36.3% | +16.1% | +15.1% | +13.6% |
| 2011 | +63.2% | −11.9% | +109.2% | +1.9% | +8.3% |
| 2012 | +31.3% | +46.8% | +0.6% | +16.0% | +11.2% |
| 2013 | +28.8% | +118.5% | −39.0% | +32.3% | +15.7% |
| 2014 | +62.8% | +38.0% | +97.3% | +13.5% | +11.7% |
| 2015 | −3.9% | −5.2% | −13.7% | +1.2% | +1.6% |
| 2016 | +19.3% | +30.8% | −2.5% | +12.0% | +7.6% |
| 2017 | +48.8% | +71.4% | +22.7% | +21.7% | +13.8% |
| 2018 | −13.5% | −25.1% | −11.0% | −4.6% | −1.8% |
| 2019 | +75.7% | +102.3% | +34.8% | +31.2% | +21.8% |
| 2020 | +79.1% | +10.1% | +39.0% | +18.3% | +17.7% |
| 2021 | +38.9% | +98.6% | −19.8% | +28.7% | +15.2% |
| 2022 | −63.8% | −56.8% | −72.6% | −18.2% | −16.7% |
| 2023 | +28.3% | +68.5% | −13.0% | +26.2% | +16.8% |
| 2024 | +13.1% | +63.6% | −35.9% | +24.9% | +14.3% |
| 2025 | +18.4% | +31.9% | −2.9% | +17.7% | +14.0% |
Every five-year window, HFEA vs. SPY
| Five-year windows | HFEA ahead of SPY | HFEA ended with a loss | Median HFEA annual return | Median SPY annual return | Worst HFEA start |
|---|---|---|---|---|---|
| 149 | 70% | 11% | 23.4% | 13.9% | −6.7%Nov 2021 |
How it was calculated
- $10,000 went in once at the start, with nothing added later. Buy and hold bought with the same money on the same day and held to the end.
- UPRO 55% and TMF 45% were bought at the open on the first day and reset at the open on the first trading day of January, April, July and October. The monthly and yearly rows differ only in when the mix is reset, and SPY 60%, IEF 40% is also reset quarterly. Hold rows buy on the first day and hold to the end.
- Each buy or sell was charged 0.15% as a cost. Taxes, currency conversion costs and interest on cash are not included.
- Dividends were reinvested at the price on each ex-dividend date. Taxes on dividends are not deducted.
- Annual return is the compound annual growth rate (CAGR) of the end balance over the starting money. Average stock share is the daily share of the account held in stocks, averaged over the period.
- The as-of date is the last trading day of last month (October 30, 2026). The numbers are recomputed at the start of each month.
- The mix was reset at the open on the set days, selling first. Everything was sold at the last day's close. See the backtest spec
Try it yourself
Rules that keep a mix of two ETFs, like HFEA, are not in the backtest tool yet. Holding UPRO over the same period can be calculated there.
This page runs to its base date, and the button opens the app with the same dates. To see the latest trading day, change the period in the app.
Common questions
Where did HFEA come from?
It started with the 2019 Bogleheads forum thread "HEDGEFUNDIE's excellent adventure" and spread on Reddit's leveraged ETF boards. The idea is to mix 3× stocks with 3× long-term Treasuries so that bonds cushion falls in stocks. The original author's backtest figures are not quoted on this page.
How often did stocks and Treasuries fall in the same year?
From 2010 to 2025, UPRO and TMF both fell in 2015, 2018 and 2022. HFEA returned −3.9% in 2015, −13.5% in 2018 and −63.8% in 2022.
How much did the rebalancing schedule matter?
With the same 55/45 mix, quarterly rebalancing returned 23.1% a year, monthly 20.4% and yearly 18.7%. Max drawdowns were −70.6%, −70.2% and −69.5%, almost the same. No schedule avoided the 2022 fall.
Why did TMF fall so far?
TMF targets three times the daily return of US Treasuries over 20 years. When rates rise, long bond prices fall, and leverage, volatility and financing costs magnify the fall. TMF fell 94.3% from its March 2020 high to October 2026.
Was it better than holding UPRO alone?
Over the whole period, UPRO alone ended higher (32.0% a year vs. 23.1% for HFEA). Max drawdowns were −76.8% for UPRO and −70.6% for HFEA. The deepest fall bottomed in March 2020 for UPRO and in October 2023 for HFEA.
Are dividends included?
Yes. UPRO, TMF, SPY and IEF all reinvest each distribution at the price on its ex-dividend date. Taxes on distributions are not deducted.
Articles and questions on the same topic
Leveraged ETFs (TQQQ, QLD, SOXL)
- Strategy articles: Leveraged ETFs at a glance: monthly investing and five rules, Infinite buying: buy in 40 steps, sell at 10% above the average cost
- Backtest questions: Grid trading on QQQ and TQQQ: did it beat simply holding?, Investing in a Nasdaq 100 2x ETF through a Korean ISA: how much more was left after tax than QLD in an overseas account?, QQQ, QLD, TQQQ and SOXL at the 200-day, 50-week, 120-week and 200-week moving averages: what happened after each touch?
Asset allocation and rebalancing
- Strategy articles: All Weather and the Permanent Portfolio: four-season allocations tested with ETFs, Dual momentum (GEM): switching between stocks and bonds on 12-month returns
- Backtest questions: TQQQ value rebalancing (VR): did it beat buy and hold or regular investing in a backtest?, Does selling below the 200-day moving average reduce drawdowns in monthly investing?
Related questions
- TQQQ, QLD and QQQ cross-rebalancing: how does it compare with simply holding?
- How much less does a 70% stock, 30% cash mix fall than 100% stocks?
- Holding TQQQ only while QQQ is above its 200-day line: did it beat buy and hold?
These results are calculated from past prices. They do not guarantee future returns and do not recommend any product.
About & methodology · Explore Investing Strategies · Glossary · Questions and answers · Rule status · Sample analysis · Pricing · Terms · Investment disclosure · Privacy policy · Open the analyzer · 한국어 · 日本語 · 繁體中文 · 简体中文