Over the same ten years, KRW 100 million became about KRW 519.96 million in SPY, KRW 1,034.58 million in SSO and KRW 1,583.86 million in UPRO. The effect of greater leverage is clear in those ending balances.
But the UPRO account had grown to KRW 352.70 million in February 2020, then fell to KRW 87.08 million on March 23. Years of gains disappeared and the balance dropped below the original investment. Could you have continued holding through that decline?
SPY vs SSO vs UPRO: fund structure
| Item | SPY | SSO | UPRO |
|---|---|---|---|
| Benchmark | S&P 500 | S&P 500 | S&P 500 |
| Daily objective | Index price and yield performance | 2x before expenses | 3x before expenses |
| Annual gross / net expenses | 0.0945% / 0.0945% | 0.88% / 0.84% | 0.88% / 0.88% |
| Structure | UIT · equity holdings | Stocks, swaps and futures; daily exposure reset | Stocks, swaps and futures; daily exposure reset |
| Distributions | Quarterly | Quarterly | Quarterly |
All three funds use the S&P 500 as their reference. SPY holds US large-cap stocks; SSO and UPRO combine stocks and derivatives to target twice and three times the daily move before costs. Those multipliers apply to one day, not to a promised long-term return.
What did KRW 100 million become over ten years?
- Investment period
- 2016-09-08 ~ 2026-09-08
- Investment
- KRW 100 million each · no further contributions
- Basis
- Distributions reinvested · valued in KRW · pretax
| Metric | SPY | SSO | UPRO |
|---|---|---|---|
| Ending value | ₩519,960,778 | ₩1,034,577,863 | ₩1,583,864,433 |
| Cumulative return | +419.96% | +934.58% | +1,483.86% |
| CAGR | +17.92% | +26.32% | +31.82% |
| MDD | −29.65% | −56.70% | −75.31% |
| Annualized volatility | 19.08% | 36.09% | 53.66% |
| Lowest account value | 98,124,657 KRW | 95,641,211 KRW | 87,075,120 KRW |
UPRO’s maximum drawdown was −75.31%. That does not mean the original 2016 investment lost that percentage. It measures the decline from the account’s higher, later peak. The balances and recovery dates below show what that meant.
How deep were the worst declines?
During the COVID selloff, the SSO account also fell from KRW 255.43 million to KRW 110.60 million. Even the 2x fund lost more than half its peak balance. It still held more than the original 2016 capital, while UPRO fell below it.
Recovery differed too. These KRW accounts regained their previous peaks on 2020-08-12 for SPY, 2020-09-02 for SSO and 2021-02-08 for UPRO. SSO had built a smaller balance before the crash, but its percentage loss and wait for recovery were also smaller.
| ETF | MDD · peak → trough | Recovery · days from peak | KRW 100m at peak: trough balance |
|---|---|---|---|
| SPY | −29.65%
2020-02-20 → 2020-03-23 |
2020-08-12
174 days |
70,350,318 KRW |
| SSO | −56.70%
2020-02-19 → 2020-03-23 |
2020-09-02
196 days |
43,298,202 KRW |
| UPRO | −75.31%
2020-02-19 → 2020-03-23 |
2021-02-08
355 days |
24,688,045 KRW |
The last column asks what would remain from KRW 100 million at each peak. It is different from the actual 2016 account balances described above. Recovery time counts calendar days from the peak.
The wait after the 2022 decline was longer. In USD, UPRO regained its 2022 peak on 2024-06-17, 896 days after that peak. Being able to tolerate a loss and being able to leave the money invested for that long are separate constraints.
View USD drawdowns and recovery dates for COVID-19 and 2022
| Window / ETF | MDD | Peak → trough | Recovery / days from peak |
|---|---|---|---|
| 2020 / SPY | −33.69% | 2020-02-19 → 2020-03-23 | 2020-08-10 / 173 |
| 2022 / SPY | −24.51% | 2022-01-03 → 2022-10-12 | 2023-12-13 / 709 |
| 2020 / SSO | −59.34% | 2020-02-19 → 2020-03-23 | 2020-09-02 / 196 |
| 2022 / SSO | −46.73% | 2022-01-03 → 2022-10-12 | 2024-03-01 / 788 |
| 2020 / UPRO | −76.82% | 2020-02-19 → 2020-03-23 | 2021-01-08 / 324 |
| 2022 / UPRO | −63.94% | 2022-01-03 → 2022-10-12 | 2024-06-17 / 896 |
COVID peaks and troughs are selected within January 1–June 30, 2020; the 2022 window covers that calendar year. Recovery is tracked through September 8, 2026. Unlike the KRW table above, these results exclude exchange-rate changes.
Did 3x leverage deliver three times the gain?
UPRO’s cumulative gain over this period was about 3.70 times SPY’s gain. The realized long-term ratio differed from the daily 3x target. The table uses USD to exclude exchange-rate effects and divides cumulative gains, not ending balances.
| ETF | Cumulative return | Return divided by SPY return |
|---|---|---|
| SPY | 311.82% | 1.00x |
| SSO | 719.41% | 2.31x |
| UPRO | 1,154.46% | 3.70x |
Leveraged funds reset exposure to their target each day. Sustained advances compound on a growing balance, so long-term gains can exceed an arithmetic multiple. Repeated reversals can leave a loss even after the index recovers.
Swap and futures trading and financing costs also affect returns. The ProShares prospectus explains that those costs sit outside the operating-expense table. These results use actual ETF prices, so costs reflected in those prices were not deducted again.
How much did the entry date change the result?
Five years of holding produced different outcomes. From 2015-03-23 to 2020-03-23, USD total returns were 17.41% for SPY, 3.30% for SSO and -19.48% for UPRO. If the money was needed during the COVID crash, five years had still left UPRO with a loss.
The table shows the lowest return for each holding period when the entry date is moved through every observed trading day since 2009. A long holding period does not remove the crashes experienced along the way.
| Holding period | SPY | SSO | UPRO |
|---|---|---|---|
| 1 year | -20.06% | -43.46% | -63.00% |
| 3 years | 0.94% | -18.36% | -40.18% |
| 5 years | 17.41% | 3.30% | -19.48% |
| 10 years | 132.49% | 237.28% | 290.82% |
UPRO’s observed history starts in 2009 and excludes the 2008 financial crisis. The loss frequencies in the full table below describe that history, not the probability of future losses.
View full rolling results, worst entry dates and calculation details
| ETF / years / windows | Worst / median / best | Loss frequency | Worst start → end |
|---|---|---|---|
| SPY / 1 / 4076 | -20.06% / 15.79% / 77.52% | 10.08% | 2019-03-21 → 2020-03-23 |
| SPY / 3 / 3576 | 0.94% / 45.83% / 114.35% | 0.00% | 2017-03-22 → 2020-03-23 |
| SPY / 5 / 3073 | 17.41% / 92.29% / 176.99% | 0.00% | 2015-03-23 → 2020-03-23 |
| SPY / 10 / 1815 | 132.49% / 251.66% / 388.35% | 0.00% | 2010-03-23 → 2020-03-23 |
| SSO / 1 / 4076 | -43.46% / 26.42% / 194.59% | 15.43% | 2019-03-21 → 2020-03-23 |
| SSO / 3 / 3576 | -18.36% / 82.17% / 266.51% | 0.25% | 2017-03-22 → 2020-03-23 |
| SSO / 5 / 3073 | 3.30% / 176.33% / 430.75% | 0.00% | 2015-03-23 → 2020-03-23 |
| SSO / 10 / 1815 | 237.28% / 645.26% / 1,400.51% | 0.00% | 2010-03-23 → 2020-03-23 |
| UPRO / 1 / 4076 | -63.00% / 35.57% / 362.30% | 21.22% | 2019-03-21 → 2020-03-23 |
| UPRO / 3 / 3576 | -40.18% / 115.50% / 467.92% | 1.57% | 2017-03-22 → 2020-03-23 |
| UPRO / 5 / 3073 | -19.48% / 241.88% / 843.01% | 0.07% | 2015-03-23 → 2020-03-23 |
| UPRO / 10 / 1815 | 290.82% / 1,069.85% / 3,638.65% | 0.00% | 2010-03-23 → 2020-03-23 |
Each observed entry is held for 1, 3, 5 or 10 years. An ending anniversary on a closed market day moves to the next observed trading day; incomplete windows are excluded. Windows overlap. The worst entry minimizes ending return and need not produce the largest interim drawdown.
What if you bought before a crash? Try your entry date
Try February 19, 2020 with a one-year holding period. Follow how far each balance fell and how it recovered. Enter your own investment amount to see what the same decline means in money.
This explorer rebases historical reinvested returns to your entry date and amount. It can differ slightly from a backtest that recalculates purchases and distribution settlement.
Loading explorer. The article tables show the fixed ten-year results.
| ETF | Value | Return | MDD |
|---|---|---|---|
| SPY | — | — | — |
| SSO | — | — | — |
| UPRO | — | — | — |
SPY is teal, SSO amber and UPRO purple. Focus a chart and use arrow keys or Home/End to read daily values. Drawdown starts from the high within the selected period.
TradingView Lightweight Charts™ · TradingView
Why can sideways markets hurt leveraged funds?
The 2022 decline followed by the 2023 rebound shows the difference. From late 2021 to late 2023, USD total returns were 3.24% for SPY, -10.51% for SSO and -27.24% for UPRO. SPY ended near its starting level, while SSO and UPRO still had losses.
Recovering a loss requires a larger percentage gain than the percentage lost. Daily leverage resets compound that effect. The S&P 500’s sector breadth does not remove the cost of large market-wide reversals.
View observed returns through advances, declines and recovery
| Window / ETF | Observed start → end | Cumulative return |
|---|---|---|
| 2017 / SPY | 2017-01-03 → 2017-12-29 | 20.77% |
| 2022 / SPY | 2022-01-03 → 2022-12-30 | -18.65% |
| 2022~2023 / SPY | 2021-12-31 → 2023-12-29 | 3.24% |
| 2017 / SSO | 2017-01-03 → 2017-12-29 | 42.26% |
| 2022 / SSO | 2022-01-03 → 2022-12-30 | -39.69% |
| 2022~2023 / SSO | 2021-12-31 → 2023-12-29 | -10.51% |
| 2017 / UPRO | 2017-01-03 → 2017-12-29 | 67.74% |
| 2022 / UPRO | 2022-01-03 → 2022-12-30 | -57.58% |
| 2022~2023 / UPRO | 2021-12-31 → 2023-12-29 | -27.24% |
Each window compares its first and last observed trading days. The 2022–2023 example was selected after the fact and is not a perfectly flat market. Actual results include expenses, financing costs and price differences as well as volatility.
View a hypothetical two-day example starting with KRW 100 million
| Index path | 1x balance | 2x balance | 3x balance |
|---|---|---|---|
| +10% → +10% advance | 121m | 144m | 169m |
| −10% → −10% decline | 81m | 64m | 49m |
| +10% → −9.0909…% round trip | 100m | 98.18m | 94.55m |
An index that rises 10% and then falls 9.0909…% returns to its starting point. An idealized, cost-free 3x fund loses about 5.45%. Two consecutive 10% gains instead give the index 21% and the 3x fund 69%, above the arithmetic 63%. The sequence and size of daily moves change the compounded result.
Did lump-sum and monthly investing differ?
What if you invested KRW 1 million a month instead of KRW 100 million at once? The table covers the same ten years, with KRW 1 million initially plus monthly contributions of KRW 1 million, totaling KRW 122 million. Distributions are reinvested.
| ETF | Principal / ending value | Cumulative return | Annualized XIRR |
|---|---|---|---|
| SPY | ₩122,000,000 / ₩325,373,420 | +166.70% | +18.52% |
| SSO | ₩122,000,000 / ₩509,337,871 | +317.49% | +26.73% |
| UPRO | ₩122,000,000 / ₩676,691,385 | +454.67% | +31.94% |
Monthly contributions spread purchase dates, but accumulated wealth still faces the next crash. A decline early in the plan is different from one after a large balance has built up. Monthly investing cannot by itself avoid UPRO’s deep drawdowns.
Principal and contribution timing differ from the lump sum. Ending balances alone cannot rank the methods. Unlike lump-sum CAGR, monthly XIRR accounts for the dates money was invested.
View the monthly contribution schedule
There are 121 monthly contributions from September 2016 through September 2026, scheduled on the first. The first contribution moves to the September 8, 2016 start, combining with the initial investment to put KRW 2 million to work that day. Closed-market purchases move to the next trading day.
What matters more than the leverage number?
For money intended to own the growth of S&P 500 businesses over many years, I would first consider a conventional index ETF such as SPY. Before choosing SSO or UPRO, I would set a balance I could tolerate seeing during a decline. UPRO’s 896-day USD recovery from its 2022 peak gives that planning question a concrete scale.
View calculation assumptions, formulas and source dates
The calculation uses split-adjusted prices and cash distributions. Without payment-date information, distributions enter cash on the ex-date and are reinvested on the next trading day. Taxes, brokerage charges and FX spreads are excluded. Fund costs reflected in actual prices are not deducted again.
Account balances are valued in KRW. Return multiples, market windows and rolling returns use USD to exclude exchange-rate effects. Currency movements are not leveraged two or three times.
For daily index returns r, an idealized cost-free 3x path compounds as ∏(1+3r)−1, rather than 3×[∏(1+r)−1]. The reported gain ratio is not a ratio of ending balances and is omitted when SPY’s cumulative return is below 0.01%.
Volatility is the sample standard deviation of daily simple returns times √252. Drawdowns use closing valuations and exclude potentially larger intraday losses. Rolling returns and the explorer rebase reinvested wealth paths, which can differ slightly from a fresh purchase-and-settlement backtest.
Expenses were checked on October 22, 2026. The September 28, 2026 SSO and UPRO prospectuses specify expense agreements through September 30, 2027. Current net expenses were not applied retrospectively to all years. Funding costs also interact with collateral income, so they were not reduced to one assumed rate.
Frequently asked questions
Does the UPRO history include the 2008 crisis?
No. UPRO launched in 2009, and the common observed history here starts June 25, 2009. No synthetic pre-inception 3x prices were added.
Are SSO and UPRO distributions multiples of SPY’s?
No. The multiplier applies to daily index performance. Distributions depend on fund income and distribution policy. This article compares total returns with distributions reinvested.
Do strong ten-year rolling returns establish future safety?
No. Entry windows overlap and reflect a particular market history since 2009. Interim drawdowns, recovery periods and future interest rates and volatility still matter.
Does monthly investing avoid large UPRO losses?
It spreads purchase dates, but invested wealth remains exposed. Spreading contributions is different from reducing leverage on the accumulated portfolio.
Compare your own contribution schedule
Compare UPRO and SSO separately with SPY using the contributions you plan to make. Follow declines and recovery on your own payment schedule.
Open investment backtestQQQ vs QLD vs TQQQ · SOXX vs SOXL · Leveraged ETF effect calculator · MDD calculator
These are pretax historical simulations, not investment advice or a guarantee. You can lose all or a substantial part of your investment. Methodology