DeployQuantRSI(14) Mean Reversion on QQQ
Invesco QQQ Trust: tracks the Nasdaq-100, a tech-heavy growth index. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.
The RSI(14) mean reversion rule buys QQQ when the 14-day RSI falls below 30 and sells the whole position when it rises above 70. It has no stop and no profit target. Over 2021-01-04 to 2026-10-02 it turned $10,000 into $14,276, a CAGR of 6.4% with a max drawdown of 24.35% and a Sharpe ratio of 0.48. Buy-and-hold on QQQ ended at $24,265, a CAGR of 16.7% with a max drawdown of 34.23%.
The rule made 15 round trips, won 11 of them, and was invested 29% of the time. A 73% win rate is high, and it still finished far behind holding. The reason sits in the trade averages: the average win was 5.89% and the average loss was 6.14%, and the profit factor was 2.63. Each trade was held for 40.7 days on average. The rule collected a few percent per trade and spent most of the window in cash while QQQ climbed.
Among the 12 templates tested on QQQ, this one ranks 11th by CAGR. Across the 59 funds the rule ranks QQQ 19th, and its median CAGR over all 59 is 2.98%, so QQQ is a better-than-typical fund for it. The only template below it on QQQ is the dip buyer at 5.37%. The best on this fund is the weekly 7% target at 16.52%, which finished just below holding. The headline run has no fees or slippage and the results are hypothetical.
Year by year
| Year | RSI mean reversion | buy & hold |
|---|---|---|
| 2021 | 11.2% | 28.4% |
| 2022 | −17.9% | −31.7% |
| 2023 | 21.9% | 53.0% |
| 2024 | 9.6% | 25.1% |
| 2025 | −2.5% | 20.4% |
| 2026 | 19.9% | 22.1% |
Year by year against buy-and-hold
The rule returned 11.2% in 2021 against 28.4% for holding, and 17.9% negative in 2022 against 31.7% negative. In 2023 it returned 21.9% against 53%, in 2024 9.6% against 25.1%, in 2025 it lost 2.5% against a gain of 20.4%, and in 2026, covering the part of the year to 2026-10-02, it returned 19.9% against 22.1%.
It beat holding in one year, 2022, by 13.8 points. In 2021 it trailed by 17.2 points, in 2023 by 31.1, in 2024 by 15.5, in 2025 by 22.9 and in 2026 by 2.2. The 2023 gap is the largest. QQQ gained 54.81% that year on its price series, and the rule held a position only in January and from August to November, taking 7.52% in January and 6.02% in August and 9.82% in November. The rest of the year it was in cash.
The 2022 result is the one that needs a closer look. A loss of 17.9% is better than holding's 31.7% loss, but it is still a loss, and it came from three trades that all lost: 3.62% from 2022-01-18 to 2022-03-28, 9.44% from 2022-04-25 to 2022-07-11 and 4.91% from 2022-08-31 to 2022-11-25. The rule had no stop and no filter, so it bought each washout in a falling market and held through it. All three trades ended with the fund lower than it began.
The 2025 result is the second weak spot. The rule's only large trade of the year was the entry at 494.95 on 2025-02-28, which lost 6.58% by 2025-04-30. A later trade from 2025-11-19 to 2025-12-10 gained 4.51%. That is two trades in a year when QQQ gained 20.4% by holding. Prices in these trade lists are adjusted for splits and dividends, so they are lower than the quotes printed at the time.
Month by month
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | 0.0% | −0.6% | 1.6% | 4.5% | 4.0% | −0.9% | 0.0% | 0.0% | −2.4% | 4.7% | 0.0% | 0.0% |
| 2022 | −2.7% | −4.3% | 3.7% | −3.0% | −1.5% | −8.6% | 4.1% | −1.3% | −10.1% | 3.9% | 3.3% | −1.6% |
| 2023 | 7.5% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 6.0% | −0.7% | −2.0% | 9.8% | 0.0% |
| 2024 | 0.0% | 0.0% | 0.0% | 1.7% | 4.0% | 0.0% | 1.9% | 1.8% | 0.0% | 0.0% | 0.0% | 0.0% |
| 2025 | 0.0% | 1.8% | −7.4% | −0.7% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 3.5% | 0.7% |
| 2026 | 0.0% | 0.0% | 1.7% | 8.5% | 0.0% | 0.0% | 1.9% | 6.5% | 0.0% | 0.0% | – | – |
The months that made and lost the money
The monthly table is mostly zeros. In 2021 the rule was flat in January, July, August, November and December. In 2023 it was flat for six months in a row and then again in December, and in 2024 it was flat in 8 of 12 months. In 2025 it traded in February, March, April, November and December. The long runs of zero explain the exposure of 29%.
The worst month was 2022-09 at 10.07% negative, the month the 2022-08-31 trade was open, which ended 4.91% negative, and it was followed by 2022-06 at 8.61% negative. In 2025, 2025-03 lost 7.39%. Holding's worst month was 2022-04 at 13.14% negative, and the rule's April 2022 figure was 3.02% negative. The rule's best months were 2023-11 at 9.82%, 2026-04 at 8.54%, 2023-01 at 7.52% and 2026-08 at 6.49%. Holding's best month was 2026-04 at 15.44%, and the rule captured 8.54% of that month by being long from 2026-03-30 to 2026-04-15.
The pattern is that good months come after a washout, while the bad ones come from a position opened too early in a falling market. Of the largest losing months, 2022-06, 2022-09 and 2025-03 each fell inside one of the three losing trades above or the 2025 loser. A rule that buys weakness takes its worst months when weakness continues.
Every trade
RSI mean reversion on QQQ made 15 closed round trips, an average hold of 41 days, an average winner of 5.89%, an average loser of −6.14%, a profit factor of 2.63, a longest losing streak of 3. It held a position at the close on 29.0% of trading days.
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2021-02-26 | $305.19 | 2021-04-09 | $322.53 | 5.7% | 42 |
| 2021-05-13 | $309.99 | 2021-06-03 | $319.68 | 3.1% | 21 |
| 2021-09-21 | $355.42 | 2021-10-25 | $363.91 | 2.4% | 34 |
| 2022-01-18 | $362.56 | 2022-03-28 | $349.45 | −3.6% | 69 |
| 2022-04-25 | $314.05 | 2022-07-11 | $284.40 | −9.4% | 77 |
| 2022-08-31 | $295.06 | 2022-11-25 | $280.57 | −4.9% | 86 |
| 2022-12-21 | $265.20 | 2023-01-24 | $280.91 | 5.9% | 34 |
| 2023-08-18 | $349.13 | 2023-09-15 | $369.27 | 5.8% | 28 |
| 2023-09-22 | $353.09 | 2023-11-15 | $380.61 | 7.8% | 54 |
| 2024-04-22 | $411.89 | 2024-05-10 | $436.40 | 6.0% | 18 |
| 2024-07-26 | $456.68 | 2024-08-26 | $474.15 | 3.8% | 31 |
| 2025-02-28 | $494.95 | 2025-04-30 | $462.39 | −6.6% | 61 |
| 2025-11-19 | $594.07 | 2025-12-10 | $620.86 | 4.5% | 21 |
| 2026-03-30 | $565.59 | 2026-04-15 | $627.65 | 11.0% | 16 |
| 2026-07-30 | $673.73 | 2026-08-17 | $733.02 | 8.8% | 18 |
Prices are adjusted for splits and dividends, so they sit below the quotes printed at the time. An open position is marked at the last close.
All 15 trades
The best trade gained 10.97% from 2026-03-30 to 2026-04-15, a hold of 16 days, entering at an adjusted 565.59 and exiting at 627.65. The second best gained 8.8% from 2026-07-30 to 2026-08-17. The third gained 7.79% from 2023-09-22 to 2023-11-15, a hold of 54 days. These three are all within 2023 and 2026, years when the washouts were followed by quick recoveries.
The worst trade lost 9.44% from 2022-04-25 to 2022-07-11, a hold of 77 days. The next worst lost 6.58% from 2025-02-28 to 2025-04-30 and 4.91% from 2022-08-31 to 2022-11-25 over 86 days, the longest hold in the list. The shortest hold was 16 days. That tells the story of the rule. Winners take two to eight weeks, and losers take two to three months, because the exit waits for RSI to climb past 70 and a falling market delays it.
The median trade gained 4.51% over 34 days. Eleven of 15 trades were winners, and the 4 losers were all in 2022 or 2025. The yearly counts show the pattern. In 2021 all 3 trades won, in 2022 none of 3 won, in 2023 all 3 won, in 2024 both won, in 2025 one of 2 won and in 2026 both won. A rule with this record has a win rate that depends almost entirely on whether the year held a prolonged decline.
The winning streak ran 5 trades and the longest losing streak ran 3, which is the 2022 sequence. The first trade was 2021-02-26 to 2021-04-09 for 5.68%, entering at 305.19. The last trade closed on 2026-08-17 and no position was open at the end of the window. The trades are few enough that a handful decide the CAGR: removing the two best would change the picture, and the page does not test that.
Largest drawdowns
| Peak | Low point | Depth | Days to low | Recovered | Days to recover |
|---|---|---|---|---|---|
| 2021-10-21 | 2022-11-03 | −24.4% | 378 | 2024-04-22 | 536 |
| 2025-02-28 | 2025-04-08 | −17.5% | 39 | 2026-04-08 | 365 |
| 2024-07-31 | 2024-08-07 | −7.5% | 7 | 2024-08-15 | 8 |
Buy-and-hold's deepest drawdown ran from 2021-11-19 to 2022-11-03 and reached −34.2%.
Drawdowns of the rule and of the fund
The rule's largest drawdown was 24.35%, from a peak on 2021-10-21 to a trough on 2022-11-03, with recovery on 2024-04-22. That is 378 days to the trough and 536 days to recover. Holding had a deeper drawdown of 34.23% over almost the same dates, from 2021-11-19 to 2022-11-03, which recovered on 2023-12-12. The rule's drawdown is shallower and lasted longer, since it recovered later than the fund did.
The second drawdown was 17.5%, from 2025-02-28 to 2025-04-08, and it took 365 days to recover, ending on 2026-04-08. Holding lost 22.38% in the same episode and was back by 2025-06-24. This is the sharper contrast. The rule entered on 2025-02-28, shortly after the fund's 2025-02-19 high, and sold on 2025-04-30. The equity curve did not return to its prior peak until a year after the trough. The third drawdown was 7.46% from 2024-07-31 to 2024-08-07, recovered in 8 days.
The pattern across the three is that the rule bought into declines that kept going. Its drawdowns were shallower than holding's, and it paid for that with a lower return in every rising year. The Sharpe ratio of 0.48 compares with 0.83 for holding, so the shallower drawdown did not offset the lower return on a risk-adjusted basis in this window.
With trading costs
The headline run fills at the bar price. These runs charge slippage on every fill.
| Slippage per fill | CAGR | Max drawdown | Final value | Sharpe |
|---|---|---|---|---|
| None (headline) | 6.4% | −24.4% | $14,276 | 0.47 |
| 5 basis points | 6.2% | −24.6% | $14,104 | 0.47 |
| 10 basis points | 5.9% | −24.9% | $13,914 | 0.45 |
What 5 and 10 basis points changed
With 15 round trips and 30 fills, the rule pays little for trading. At 5 basis points the CAGR was 6.17%, with a max drawdown of 24.56% and a Sharpe ratio of 0.465. At 10 basis points the CAGR was 5.92%, the drawdown was 24.88% and the Sharpe ratio was 0.448. The end value fell from $14,276 in the headline run to $14,104 and then $13,914.
The loss in CAGR is small at both levels. QQQ is a very liquid fund, with an average daily dollar volume of about $18,772,024,579 and a median minute volume of 77,976 shares, so the cost assumption is reasonable for it. Costs are not the reason the rule trails holding. The gap in CAGR against buy-and-hold is 10.3 points, and it was already there in the headline run with no costs.
Changing the parameters
| Version | CAGR | Max drawdown | Round trips | Win rate | Final value |
|---|---|---|---|---|---|
| Published rules | 6.4% | −24.4% | 15 | 73% | $14,276 |
| RSI < 25 / > 70 | 2.5% | −22.5% | 10 | 60% | $11,557 |
| RSI < 35 / > 70 | 6.5% | −28.7% | 20 | 80% | $14,382 |
| RSI < 30 / > 65 | 6.3% | −24.0% | 16 | 75% | $14,198 |
| RSI < 30 / > 75 | 6.7% | −20.2% | 13 | 69% | $14,517 |
Moving the RSI thresholds
Four variants changed one threshold at a time.
Buying below 25 instead of 30 gave a CAGR of 2.55% with a drawdown of 22.49% on 10 trades, 6 of them winners. A stricter entry means fewer trades and a lower return. Buying below 35 gave 6.53% with a drawdown of 28.66% on 20 trades, 16 of them winners. A looser entry bought earlier, made more trades and won more of them, but it carried a deeper drawdown. Its CAGR was nearly the same as the base 6.4%.
Selling above 65 instead of 70 gave 6.3% and a 24.02% drawdown on 16 trades with 12 winners. Selling above 75 gave 6.71% and a drawdown of 20.17% on 13 trades with 9 winners. The later exit was the best of the four variants on both return and drawdown, with a Sharpe ratio of 0.483 against 0.48 for the base. Holding for a larger rebound cost nothing here.
The four variants span CAGRs from 2.55% to 6.71%, and three of them sit within half a point of the base. The entry threshold at 25 is the one setting that clearly hurts. None of the variants approach the 16.7% for buy-and-hold. The variants were run one at a time, so combinations such as buying below 35 and selling above 75 were not tested, and the differences of a few tenths of a point rest on 13 to 20 trades.
How QQQ behaved
| Measure | QQQ |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | 151.1% |
| Annualized volatility | 22.4% |
| Deepest drawdown | −35.0% (2021-11-19 to 2022-11-03) |
| Up days | 54.8% |
| Average daily range | 1.58% |
| Average overnight gap | 0.59% |
| Correlation to SPY | 0.94 |
| Correlation to TLT | 0.09 |
| Sessions above the 200-day average | 75.2% |
| Crossings of the 200-day average | 20 |
| Falls of 10% or more from a 20-day high | 21 |
QQQ against the other broad index funds and the other rules
QQQ tracks the Nasdaq-100. In this window its CAGR was 17.39% on its own price series, with annualized volatility of 22.39% and a max drawdown of 35% from 2021-11-19 to 2022-11-03. It recovered on 2023-12-12, after 516 sessions. Calendar-year returns were 29.18% in 2021, 32.39% negative in 2022, 54.81% in 2023, 25.59% in 2024, 20.77% in 2025 and 22.44% in 2026 to date. QQQ's beta to SPY is 1.29.
QQQ rose on 54.82% of sessions. The average up day was 1.01% and the average down day was 1.06% negative. It had 21 separate cases of a 10% fall within 20 days, covering 82 trading days. It was above its 200-day average on 75.18% of sessions and crossed it 20 times. A fund that spent three-quarters of the time above its long average gives a rule that buys weakness a steady supply of dips that recovered and few that did not.
Among the 12 broad index funds, this rule's CAGR was highest on EEM at 9.51%, then VOOV at 8.01% and QQQE at 7.22%. On QQQM, which holds the same index, it returned 6.46% on 15 round trips, close to QQQ's 6.4%. On VV it returned 5.02%, on SPY 4.37% and on IWM 0.73%. The category median is 6.46%, which QQQ almost matches.
Against the other rules on QQQ, the monthly cycle returned 14.73% with a 36.07% drawdown, the RSI(2) snapback returned 14.17% with a drawdown of 14.27%, the golden cross 13.6%, and the 200-day regime filter 12.9% with a 19.93% drawdown. The rules that held QQQ more of the time did better. The RSI(2) snapback is also a mean-reversion rule, and it did better here, which shows that the choice of RSI length and thresholds matters more than the idea of buying weakness. The RSI(2) snapback, the golden cross and the 200-day regime filter pages show the same window. The strategy overview and the QQQ page have the full rankings.
The rules
Buy when the 14-day RSI drops below 30 (oversold), sell when it recovers above 70 (overbought).
- WHEN the market opens · IF not invested AND RSI(14) < 30 · THEN buy with 98% of the sleeve
- WHEN the market opens · IF invested AND RSI(14) > 70 · THEN sell the whole position
A standard mean-reversion setup. The Relative Strength Index measures how stretched recent price action is. Readings under 30 have historically marked short-term washouts in uptrending assets. This template buys at the next session open and holds until RSI crosses back above 70. It has no profit target and no stop.
Good for: assets that trend up over time but overshoot on the way, such as broad index ETFs.
Watch out: in a persistent downtrend, RSI can stay oversold for weeks while the position keeps losing; there is no stop-loss in this template.
How an RSI(14) threshold behaves on a growth index fund
RSI(14) measures how stretched the last 14 sessions are. A reading under 30 means the fund has fallen on most of those sessions or fallen hard on a few. The rule buys at the next open and holds until the reading passes 70. Because the exit needs a strong rebound, the hold can run for weeks, and the template has no stop to cut a position that keeps falling.
QQQ's own data shows how rarely the entry fires. RSI(14) fell below 30 on only 15 sessions in the whole window, and the rule made 15 round trips, so each oversold episode produced about one trade. The median forward 5-day return after those 15 readings was 3.69% against a baseline of 0.56%, and the 20-day median was 1.42% against 1.74%. The signal worked at five days and gave nothing extra at twenty. This rule holds for roughly a month, which puts most trades between those two horizons.
The other side of the rule is the exit. RSI(14) stood above 70 on 145 sessions. The sell test is easy to meet in a rising fund, but the rule can only sell after it has bought, and it only buys after a washout. In a stretch like 2023 or 2025 where the fund rose without a deep dip, the rule had no position at all. In the monthly table it shows 0 for every month from 2023-02 to 2023-07 and again for 2025-05 to 2025-10.
The template's description fits assets that trend up but overshoot on the way, such as broad index funds. QQQ did trend up, with a total return of 151.07% on its price series, and the overshoots did occur. The cost is the time out of the market. The rule's exposure of 29% means that most sessions were spent in cash while the fund earned its 17.39% CAGR.
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
Did RSI mean reversion beat buy-and-hold on QQQ?
Over 2021-01-04 to 2026-10-02, RSI mean reversion on QQQ returned 6.4% annualized vs 16.7% for buy-and-hold: it trailed buy-and-hold by 10.3% per year, with a maximum drawdown 9.9 points shallower than holding (24.4% vs 34.2%).
How many trades did it make?
15 completed round trips over 5.7 years (30 fills), with 73% of round trips closing profitably.
Why RSI 30/70?
They are the conventional oversold and overbought bands from Welles Wilder's original formulation. Treat them as a starting point. In DeployQuant you can change them to 25/65 or anything else and re-backtest in seconds.
Does this strategy use a stop-loss?
No. The only exit is RSI recovering above 70. Adding a stop or a take-profit block is a one-block edit in the Lab.
How did the RSI mean reversion strategy do on QQQ?
It returned a CAGR of 6.4% over 2021-01-04 to 2026-10-02, ending at $14,276 from $10,000, with a max drawdown of 24.35% and a Sharpe ratio of 0.48. Buy-and-hold on QQQ returned 16.7% with a 34.23% drawdown. The rule ranks 11th of 12 templates on QQQ.
Why does a 73% win rate still lag buy-and-hold?
The rule won 11 of 15 trades, but the average win of 5.89% was no larger than the average loss of 6.14%, and it was invested only 29% of the time. QQQ gained 142.65% in total over the window, and the rule sat in cash while much of that happened.
Which trades lost money?
Four lost. They were 3.62% from 2022-01-18 to 2022-03-28, 9.44% from 2022-04-25 to 2022-07-11, 4.91% from 2022-08-31 to 2022-11-25 and 6.58% from 2025-02-28 to 2025-04-30. The rule has no stop and holds until RSI passes 70.
What happens with different RSI thresholds?
Buying below 25 gave 2.55%, buying below 35 gave 6.53%, selling above 65 gave 6.3% and selling above 75 gave 6.71%. The 75 exit had the shallowest drawdown at 20.17%. None came near buy-and-hold's 16.7%.
Do trading costs matter for this rule?
Very little. At 5 basis points the CAGR was 6.17% and at 10 basis points it was 5.92%, from 6.4% in the headline run. The rule makes only 15 round trips.
Did the rule protect against the 2022 decline?
Partly. It lost 17.9% in 2022 against 31.7% for holding, and its drawdown was 24.35% against 34.23%. All three of its 2022 trades lost money, so the protection came from being in cash part of the year.
How reliable is a test with 15 trades?
It is a small sample. A few trades decide the result, and the test covers one window with one bear market. The results are hypothetical and are not a forecast.
Related
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.