DeployQuantQLD trading strategies, backtested
ProShares Ultra QQQ: 2x daily leveraged Nasdaq-100, between QQQ and TQQQ in leverage. Every DeployQuant template run on QLD over 5.7 years of minute data, same engine, same window, sorted by return.
QLD is ProShares Ultra QQQ, a fund that targets twice the daily move of the Nasdaq-100. From 2021-01-04 to 2026-10-02 buy-and-hold on QLD made 23.6% a year, a total of 236.8%, and turned $10,000 into $33,676. The cost was a maximum drawdown of 62.8%, from the peak on 2021-11-19 to the trough on 2022-12-28. The Sharpe ratio was 0.72.
All 12 DeployQuant templates ran on QLD over the same window. Two of them beat holding on return: the RSI(2) snapback at 26.2% a year and the weekly 7% target at 25.6%. Eleven of the 12 had a shallower drawdown than holding, and all 12 were positive. The other ten finished behind the fund on return, and their gap was widest for the slower rules, which sat in cash through the 2023 rebound.
A leveraged fund is a different test from an index fund. QLD's volatility was 44.8% a year against 14.0% for a broad value fund such as VOOV. The average daily range was 3.14%. Rules that decide once a day see large moves between decisions, and the drawdowns they report are measured in tens of percent rather than single digits. The same rules on the unlevered fund, QQQ, can be compared on that page.
The headline runs carry no fees or slippage.
| Strategy | CAGR | max DD | Sharpe | trades | win rate | final value |
|---|---|---|---|---|---|---|
| RSI(2) Dip Snapback | 26.2% | −29.5% | 0.93 | 164 | 68% | $37,984 |
| Weekly Entry + 7% Target | 25.6% | −44.4% | 0.83 | 190 | 43% (+1 open) | $37,040 |
| 20-Day Momentum + Trailing Stop | 22.5% | −20.1% | 0.97 | 20 | 50% (+1 open) | $32,046 |
| 200-Day SMA Regime Filter | 21.6% | −28.8% | 0.84 | 12 | 42% (+1 open) | $30,694 |
| First-to-Last Day of Month | 20.6% | −65.0% | 0.67 | 69 | 61% (+1 open) | $29,277 |
| SMA 10/50 Trend | 18.8% | −32.3% | 0.76 | 19 | 37% (+1 open) | $26,838 |
| Drawdown Dip Buyer + 8% Target | 18.6% | −62.0% | 0.63 | 12 | 100% (+1 open) | $26,598 |
| SMA-200 Trend + 15% Trailing Stop | 17.2% | −40.1% | 0.70 | 17 | 47% (+1 open) | $24,829 |
| Golden Cross (SMA 50/200) | 12.0% | −39.9% | 0.53 | 3 | 67% (+1 open) | $19,160 |
| EMA 12/26 Trend | 10.8% | −38.1% | 0.51 | 24 | 38% (+1 open) | $18,066 |
| 3-Month Momentum Switch | 10.8% | −36.1% | 0.52 | 18 | 44% (+1 open) | $18,039 |
| RSI(14) Mean Reversion | 8.8% | −48.1% | 0.42 | 15 | 73% | $16,210 |
| Buy & hold QLD | 23.6% | −62.8% | 0.72 | – | – | $33,676 |
Which templates beat holding QLD and which did not
The RSI(2) snapback made 26.2% a year with a 29.5% drawdown, 164 round trips and a 68% win rate. It ended at $37,984, which is $4,308 above holding. The drawdown was less than half of holding's 62.8%. By year it made 58.6% in 2021, lost 15.3% in 2022 against a loss of 59.7% for the fund, and made 23.6%, 39.0%, 38.2% and 19.1% in the years after. In 2023 holding made 113.2%, so the rule gave up most of that rebound. It beat holding in 2021, 2022 and 2025 and trailed in the other three years.
The weekly 7% target made 25.6% with a 44.4% drawdown and 190 round trips. Only 43% of its trades won, but the average win of 7.24% was larger than the average loss of 3.87%. It lost 31.7% in 2022 and made 90.0% in 2023. At the end of the test it held an open position entered on 2026-09-28.
The momentum breakout made 22.5% a year with a 20.1% drawdown and a Sharpe ratio of 0.97, the highest of the 12 and above holding's 0.72. It made only 20 round trips, with a 50% win rate and a profit factor of 3.78. Its best trade ran from 2025-05-05 to 2025-11-07 for a gain of 50.1% in 186 days. It lost 6.1% in 2022.
The 200-day regime filter made 21.6% with a 28.8% drawdown and 12 trades. It won 42% of them, with an average win of 24.98% against an average loss of 4.36%. The monthly cycle made 20.6% but kept a 65.1% drawdown, since it is invested 95.2% of the time. The SMA 10/50 trend made 18.8% with a 32.3% drawdown. The dip buyer made 18.6% with a 62.0% drawdown, no different from holding on risk, because it buys the dips and then sits through the long ones.
The weakest templates were the trend with a trailing stop at 17.2%, the golden cross at 12.0% with 3 round trips, the EMA 12/26 trend at 10.9%, the 3-month momentum switch at 10.8%, and the RSI(14) mean reversion at 8.8% with a 48.1% drawdown. The RSI(14) rule won 73% of its 15 trades but its average loss of 15.06% outweighed an average win of 11.13%. It lost 39.7% in 2022 because it bought declines that kept going.
The pattern across the table is that slow trend rules protected against 2022 and then came back into the fund late. The golden cross made 63.6% in 2023 against 113.2% for holding, and lost 15.5% in 2025 while the fund gained 29.9%. The rules that stayed close to the fund's return were the ones that traded often and stayed in the market.
Drawdown and recovery by template
Holding QLD produced three drawdowns of note: 62.8% into 2022-12-28, 41.9% from 2024-12-16 to 2025-04-08, and 25.9% from 2024-07-10 to 2024-08-07. The first took 404 days to reach the trough and 517 more to recover. The second took 113 days down and 100 to recover.
The momentum breakout kept its worst drawdown to 20.1%, from 2022-08-15 to 2023-03-22, and recovered by 2023-05-26. The second was 19.5% in 2024. The 200-day regime filter had its worst drawdown, 28.8%, in the 2024-07-10 to 2024-08-07 slide, which the rule did not avoid because the fund was above its 200-day average at the time. That is a reminder that a trend filter reacts to a slow decline and cannot dodge a sudden one.
The weekly 7% target had the longest recovery of the group of fast rules. Its 44.4% drawdown began on 2022-02-09, bottomed on 2022-11-09 and recovered on 2023-07-12. It also finished the window inside a 28.2% drawdown that started on 2026-06-02 and had not recovered by 2026-10-02. The trend with a trailing stop and the 200-day regime filter were in a similar position, with drawdowns of 27.5% and 22.5% from the same peak.
The golden cross shows the cost of slow rules most clearly. Its deepest drawdown, 39.9%, came in 2025 after the rule had held through the first drop in the fund, because the 50-day average stayed above the 200-day average until the decline was well advanced. Holding made the same period's 41.9% drawdown, so the filter protected almost nothing in that episode.
Months, streaks and yearly gaps for each rule
April set both extremes for holding QLD. April 2026 was the best month at 32.53% and April 2022 the worst at negative 25.58%. The templates mostly followed. The monthly cycle made 28.9% in April 2026, the dip buyer 28.66% and the weekly target 24.83%. In April 2022 the dip buyer lost 25.3% and the momentum breakout 12.54%. The trend rules had November 2023 as their best month: the golden cross made 21.45%, the 200-day filter 21.37%, the trailing stop strategy 21.35% and the RSI(14) rule 20.68%. January 2022 was the worst month for the monthly cycle at negative 21.9%, the RSI(2) snapback at negative 13.38%, the 200-day filter at negative 13.35%, the golden cross at negative 16.86% and the trailing stop strategy at negative 20.54%.
The streak figures show what each rule's edge looks like in sequence. The dip buyer won 12 round trips in a row and never lost one, each near its 8% target. The RSI(2) snapback won 8 in a row and lost 3 in a row at most. The weekly target won 6 in a row and lost 8 in a row, a more balanced split than the weekly target shows on the quieter funds. The RSI(14) rule won 5 in a row and lost 3. The 200-day filter and the trailing stop strategy lost 3 in a row at most, and the SMA 10/50 trend and the EMA 12/26 trend lost 4.
The yearly gaps to holding are very wide for a leveraged fund. The RSI(2) snapback beat holding by 44.4 points in 2022 and trailed it by 89.6 points in 2023, when QLD gained 117.25%. The 200-day filter beat holding by 46.4 points in 2022 and trailed by 59.2 in 2023. The EMA 12/26 trend gap was positive 32.1 in 2022 and negative 62.4 in 2023, and the trailing stop strategy's was positive 39.2 then negative 63.8. Every trend rule swung between the two years in the same way, which is the cost of sitting out the decline and then rejoining the rebound late. The momentum breakout trailed by 54.5 points in 2021 and 57.1 in 2023 and led by 53.6 in 2022 and 15.7 in 2025.
Open trades carry part of the reported returns. The 200-day filter has been in since 2026-04-09 at an adjusted 67 and is up 46.46%. The golden cross has been in since 2026-04-30 at 80.12 and is up 22.48%. The trailing stop strategy has been in since 2026-07-20 at 87.66 and is up 11.94%. The SMA 10/50 trend, the momentum breakout and the dip buyer entered between June and September 2026 and are up between 5.71% and 8.61%. These are marks to the last close.
How each strategy traded QLD
| Strategy | Time in market | Avg hold (days) | Best trade | Worst trade | Profit factor | With 10 bps slippage |
|---|---|---|---|---|---|---|
| RSI(2) snapback | 36.6% | 5 | 22.7% | −16.4% | 1.79 | 19.3% |
| weekly 7% target | 70.0% | 7 | 10.0% | −14.8% | 1.37 | 19.2% |
| momentum breakout | 50.3% | 51 | 50.1% | −9.9% | 3.78 | 21.7% |
| 200-day regime filter | 62.3% | 94 | 48.2% | −9.3% | 3.59 | 21.1% |
| monthly cycle | 95.2% | 28 | 29.5% | −22.4% | 1.61 | 17.8% |
| SMA 10/50 trend | 61.6% | 67 | 59.0% | −12.0% | 2.72 | 18.0% |
| dip buyer | 74.0% | 120 | 8.9% | 8.0% | – | 18.5% |
| trend + trailing stop | 62.6% | 73 | 37.1% | −16.6% | 2.26 | 16.5% |
| golden cross | 62.3% | 384 | 67.6% | −17.4% | 4.35 | 11.8% |
| EMA 12/26 trend | 63.4% | 55 | 45.6% | −11.2% | 1.77 | 9.9% |
| 3-month momentum | 60.0% | 70 | 51.0% | −12.5% | 2.36 | 10.1% |
| RSI mean reversion | 29.1% | 41 | 22.6% | −21.1% | 1.95 | 8.2% |
How the trades looked and what costs did
The RSI(2) snapback held a position for 4.7 days on average and was invested 36.6% of the time. Its best trade ran from 2025-04-07 to 2025-04-10 for a gain of 22.7%, and the worst ran from 2025-02-21 to 2025-03-06 for a loss of 16.4%. The longest losing streak was 3 trades. Its profit factor was 1.79. The rule works by buying a sharp decline and selling on the recovery, and the April 2025 trade is the clearest example, since the fund's best single day, 2025-04-09 at 23.5%, fell inside it.
The long-hold rules show a different table. The golden cross held an average of 384 days, its best trade lasted 746 days for a gain of 67.6%, and it had one loser of 17.4%. The dip buyer's best trade lasted 916 days from 2021-12-09 to 2024-06-12 and returned only 8.9%, because the rule exits at its 8% target and the position sat through the whole 2022 decline before reaching it. A 100% win rate and an 8% best trade describe a rule that holds losing positions rather than selling them.
Costs changed the ranking. At 10 basis points of slippage the RSI(2) snapback fell from 26.2% to 19.3%, and at 5 basis points it made 22.7%. Its drawdown rose to 31.6%. The weekly 7% target went from 25.6% to 19.2% at 10 basis points. The monthly cycle went from 20.6% to 17.8%. After costs of 10 basis points the momentum breakout at 21.7% and the 200-day regime filter at 21.1% were ahead of both fast rules. The slow rules barely moved because they trade rarely: the golden cross went from 12.0% to 11.9%.
The cost runs apply a flat figure to every trade. QLD trades about $285,882,203 a day on average and the median minute bar has 8,796 shares, so a small test account meets a liquid book. A much larger order would not.
Most templates ended the window holding a position. The weekly 7% target entered on 2026-09-28, the monthly cycle on 2026-10-01, and the 3-month momentum switch on 2026-10-02. Their returns include that open position at the last price. Read the trend with trailing stop page for the trade list of a rule that lost 8.26% on its average loser.
How QLD behaved
| Measure | QLD |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | 254.3% |
| Annualized volatility | 44.8% |
| Deepest drawdown | −63.7% (2021-11-19 to 2022-12-28) |
| Up days | 54.3% |
| Average daily range | 3.14% |
| Average overnight gap | 1.19% |
| Correlation to SPY | 0.94 |
| Correlation to QQQ | 1.00 |
| Correlation to TLT | 0.09 |
| Sessions above the 200-day average | 72.3% |
| Crossings of the 200-day average | 24 |
| Falls of 10% or more from a 20-day high | 50 |
Calendar years
| Year | Return |
|---|---|
| 2021 | 59.1% |
| 2022 | −60.5% |
| 2023 | 117.3% |
| 2024 | 42.8% |
| 2025 | 30.3% |
| 2026 (part) | 39.6% |
Biggest single days
| Best day | Move |
|---|---|
| 2025-04-09 | 23.5% |
| 2022-11-10 | 14.7% |
| 2022-11-30 | 9.2% |
| 2022-07-27 | 8.5% |
| 2025-05-12 | 8.1% |
| Worst day | Move |
|---|---|
| 2025-04-04 | −12.2% |
| 2022-09-13 | −11.1% |
| 2025-04-03 | −10.7% |
| 2022-05-05 | −9.9% |
| 2022-05-18 | −9.9% |
Average return by calendar month
| Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2.5% | −2.2% | 0.8% | 1.6% | 10.0% | 5.0% | 4.1% | 0.6% | −3.6% | 4.9% | 8.3% | −1.5% |
Most and least correlated funds
| Most correlated | Least correlated | ||
|---|---|---|---|
| TQQQ | 1.00 | SQQQ | -1.00 |
| QQQ | 1.00 | QID | -1.00 |
| QQQM | 1.00 | PSQ | -1.00 |
| VOOG | 0.98 | TECS | -0.97 |
| ROM | 0.97 | REW | -0.96 |
Leverage and decay against QQQ
| Year | QLD | QQQ | 2× QQQ, reset daily |
|---|---|---|---|
| 2021 | 59.1% | 29.2% | 61.4% |
| 2022 | −60.5% | −32.4% | −58.8% |
| 2023 | 117.3% | 54.8% | 132.0% |
| 2024 | 42.8% | 25.6% | 52.6% |
| 2025 | 30.3% | 20.8% | 38.4% |
| 2026 | 39.6% | 22.4% | 45.3% |
The last column compounds 2× QQQ's daily return with no fees or financing: the return a perfect daily-reset fund would have had. It leaves out the fund's fees, its borrowing costs and the interest it earns on cash, which put a real fund below or above this line. A daily-reset fund does not aim for 2× the underlying's return over a year.
Over 2021-01-04 to 2026-10-02, QLD returned 254.3% while QQQ returned 151.1% and a perfect daily-reset 2× version would have returned 373.2%. Its measured daily beta to QQQ was 2.00.
What QLD's own behaviour explains
QLD's daily moves are large and close to even in direction. Up days were 54.3% of sessions with an average gain of 2.02%, and the average down day was negative 2.15%. The lag-1 autocorrelation was negative 0.04, so there is almost no day-to-day persistence. The average overnight gap was 1.19% and overnight moves carried 64.1% of the log return, with 35.9% coming during the session.
That split matters for rules that enter at the open. A rule that buys the open and sells the open collects the overnight gap, which was the larger share.
The mean reversion signals had a measurable edge on QLD. After RSI(2) closed below 10, which happened on 151 sessions, the median 5-day return was 1.18% against a baseline of 0.94%, and the median 20-day return was 3.48% against 2.81%. After the 14-day RSI closed below 30, on 18 sessions, the median 5-day return was 6.11% and the 20-day return was 3.83%. The sample of 18 is small, but the RSI(14) mean reversion rule bought into 2022 declines and lost, which shows the signal does not hold in a prolonged slide.
The fund fell 10% or more from a 20-day high on 50 occasions, covering 309 days, and it spent 72.3% of sessions above the 200-day average, crossing it 24 times. The longest drawdown lasted 630 sessions, from 2021-11-19 to a recovery on 2024-05-28. A buy-and-hold investor waited 517 days from the trough to recover. Every rule that avoided part of 2022 had an advantage in that window, and every rule that missed 2023 gave it back.
Calendar years were 59.1% in 2021, negative 60.5% in 2022, 117.3% in 2023, 42.8% in 2024, 30.3% in 2025 and 39.6% for the partial year 2026. The best day was 2025-04-09 at 23.52%, then 2022-11-10 at 14.69%. The worst days were 2025-04-04 at negative 12.19% and 2022-09-13 at negative 11.06%. By calendar month, May averaged 9.97% and November 8.28%, while September averaged negative 3.65% and February negative 2.20%. With 5 or 6 observations per month these are descriptions of this window.
Leverage and decay against QQQ
QLD tracks two times the daily return of QQQ, and the fund's correlation to QQQ is 1.00 with a realized beta of 2.0 and an R-squared of 1.00. Its beta to SPY was 2.57. Over the full window QQQ returned 151.1% and QLD returned 254.3%. Twice the QQQ total would be 302.1%, so the fund finished 47.9 percentage points below that figure. A daily-rebalanced model of 2x gives 373.2%, which is above both. The difference between the fund and the model comes from fees, financing and the path of the underlying.
By year, the gap between the fund and twice the underlying was small in 2021, at 0.74 points, and positive in 2022 and 2023 at 4.25 and 7.64 points. It turned negative in 2024, 2025 and 2026, at negative 8.36, negative 11.26 and negative 5.29 points. In the years when QQQ trended with few reversals, the fund returned less than twice the underlying. The compounding effect works against the holder in choppy markets and for the holder in steady trends, and the data shows both in this window.
The fund is most correlated with TQQQ, QQQ and QQQM, and inversely with SQQQ, QID and PSQ. Its sibling SSO tracks the S&P 500 at 2x, and the best template there was also the RSI(2) snapback, at 24.6% against 21.9% for holding. On TQQQ the same rule made 39.3% against 25.4%, so the RSI(2) edge is larger on the higher-leverage fund.
The test stops at one window of 5.74 years that contains one bear market. A rule that looks good against the 2022 decline of 60.5% might not look as good in a window without one. None of these figures is a forecast.
Calendar months, weekdays and the leveraged peers
May averaged 9.97% on QLD and November 8.28%, with June at 4.96%, October at 4.87% and July at 4.12%. September averaged negative 3.65%, February negative 2.2% and December negative 1.53%. Each month has 5 or 6 observations, and a single year such as 2023 moves a calendar average by several points, so these describe the window and carry no seasonal claim. Monday averaged 0.35% and Wednesday 0.19%, against 0.04% on Tuesday, 0.03% on Thursday and 0.05% on Friday.
The leveraged peers show how much the best template depends on the underlying. TQQQ, at three times the Nasdaq-100, held for 25.35% and its best template, the RSI(2) snapback, made 39.31%. The same rule was best on SSO, where holding made 21.92% and the rule 24.56%, and on SPUU at 22.48% held and 24.12% for the rule. On SOXL the golden cross was best at 45.93% against 33.32% for holding, and on TECL the monthly cycle was best at 35.97%, below its 38.16% hold. The RSI(2) snapback was the best template on TQQQ, SSO, SPUU and TMF, and on each of those it beat holding. On QLD its 2-day signal fired on 151 sessions and the rule was invested 36.6% of the time.
Build it from blocks (or type it in English), backtest it on 5.7 years of minute data in seconds, tweak any parameter, then paper trade it on live data. No card, no broker needed to start.
Frequently asked questions
What was the best strategy for QLD?
Of the 12 templates tested on QLD over 2021-01-04 to 2026-10-02, the strongest by CAGR was RSI(2) snapback at 26.2% (max drawdown 29.5%), versus 23.6% for buy-and-hold. The best result in hindsight is not a forecast. Check drawdowns and trade counts before drawing conclusions.
Did any strategy beat buying and holding QLD?
2 of 12 templates beat QLD buy-and-hold (23.6% CAGR) on this window; 11 of 12 had a shallower maximum drawdown than holding (62.8%).
Did any strategy beat buy-and-hold on QLD?
Two did. The RSI(2) snapback made 26.2% a year and the weekly 7% target made 25.6%, against 23.6% for holding QLD from 2021-01-04 to 2026-10-02. The other 10 templates finished below holding.
What was QLD's maximum drawdown?
Holding QLD fell 62.8% from 2021-11-19 to 2022-12-28 and did not recover until 2024-05-28. Eleven of the 12 templates had a shallower drawdown. The momentum breakout had 20.1%.
How does RSI(2) mean reversion do on QLD?
It made 26.2% a year with a 29.5% drawdown and 164 trades, and won 68% of them. At 10 basis points of slippage the return fell to 19.3%. The rule made 58.6% in 2021 and lost 15.3% in 2022.
Does QLD return exactly twice QQQ?
No. QQQ returned 151.1% over the window and QLD returned 254.3%. Twice the QQQ figure is 302.1%, so QLD finished below it. The gap varied by year, from 7.64 points above in 2023 to 11.26 points below in 2025.
How volatile is QLD?
Annualized volatility was 44.8% and the average daily range was 3.14%. The fund fell 10% or more from a 20-day high on 50 occasions. Its calendar years ranged from negative 60.5% in 2022 to 117.3% in 2023.
Is a trend strategy better than buy-and-hold on QLD?
Not on return. The 200-day regime filter made 21.6% and the SMA 10/50 trend made 18.8% against 23.6% for holding. They cut the drawdown to 28.8% and 32.3% and gave up most of the 2023 rebound.
Other leveraged etfs
Backtests are hypothetical, computed by DeployQuant's engine on minute-resolution consolidated US market data (2021-01-04 to 2026-10-02, $10,000 starting capital, no margin, no fees or slippage in the headline run; buy-and-hold puts 98% of the account in at the first open, as the templates do) and do not guarantee future results. Nothing on this page is investment advice. Live trading involves risk of loss.