RSI(14) Mean Reversion on TQQQ

ProShares UltraPro QQQ: 3x daily leveraged Nasdaq-100, with large trends and deep drawdowns. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.

RSI(14) mean reversion on TQQQ returned 12.08% a year from 2021-01-04 to 2026-10-02 and turned $10,000 into $19,242. Buy-and-hold of the same fund returned 25.35% and ended at $36,585, so the rule gave up a large part of the return. It also took a smaller worst fall: the maximum drawdown was 66.58% against 80.77% for holding. The Sharpe was 0.48 against 0.69. On TQQQ this rule ranks 8th of the 12 templates, and TQQQ is the 6th best of the 59 funds for this rule.

The rule is simple. It buys at the next open when the 14-day RSI falls below 30 and sells when RSI rises above 70. There is no stop and no profit target. It made 16 round trips, won 12 of them for a 75% win rate, and was invested 29.8% of sessions. The win rate looks strong. The average win was 16.71% and the average loss was 25.02%, and the profit factor was 1.83. Four losing trades, three of them in 2022, are the difference between the 75% win rate and a return far below holding.

The wins came in 2021, 2023, 2024 and 2026, when TQQQ dipped and bounced. The losses came when TQQQ was oversold in a falling market and the rule bought at a signal that did not mark the low. Each trade is listed below with its dates, because with 16 trades one entry changes the result.

These are backtest figures on one window with no fees or slippage in the headline run, $10,000 starting cash, no margin and daily decisions. Prices in trade lists are adjusted for splits and dividends.

Year by year

Holding TQQQ returned 88.9% in 2021, −78.1% in 2022, 187.6% in 2023, 57.1% in 2024, 33.9% in 2025 and 54.1% in 2026 so far. The rule returned 32.4%, −58.7%, 71.6%, 46.5%, −17.6% and 70%. It beat holding in two years, 2022 by 19.4 points and 2026 by 15.9, and trailed in the other four. The largest shortfall was 2023, when the fund returned 187.6% and the rule returned 71.6%, because the rule was flat from February to July and again from December.

In 2021 the rule made 3 trades and all won: 14.75% from 2021-02-26 to 2021-04-09, 9.17% from 2021-05-13 to 2021-06-03 and 6.28% from 2021-09-21 to 2021-10-25. The result was 32.4% in a year the fund returned 88.9%. The rule sat out the second half of 2021 after its last sale on 2021-10-25, which left it in cash for November and December and through the fund's peak on 2021-11-19. The rule's equity peak on 2021-10-21 came during that last trade, a month before the fund's own peak.

In 2022 the rule lost on every trade it closed that year. It bought on 2022-01-18 at an adjusted price of 32.48 and sold on 2022-03-28 at 27.17, a loss of 16.35%. It bought again on 2022-04-25 at 19.34 and sold on 2022-07-11 at 12.85, a loss of 33.56%. A third entry on 2022-08-31 at 13.76 ended on 2022-11-25 at 10.63, a loss of 22.75%. The year's return of −58.7% was better than the fund's −78.1% only because the rule was out of the market part of the time.

The first trade of 2023 was a recovery: bought on 2022-12-21 at 8.788 and sold on 2023-01-24 at 10.17, a gain of 15.73%. Then 2023 had two more wins, 16.46% from 2023-08-18 and 21.01% from 2023-09-22 to 2023-11-15. In 2024 there were three wins of 13.74%, 8.68% and 19.45%. In 2025 the rule lost 27.41% on a trade from 2025-02-28 to 2025-04-30, then gained 12.92% from 2025-11-19 to 2025-12-10. In 2026 the two trades gained 35.06% and 27.26%.

Month by month

Most months were flat. The rule was invested 29.8% of sessions, so in the monthly table most cells read 0%. The best month was November 2023 at 32.09% and the worst September 2022 at −29.88%. Holding had a best month of April 2026 at 51.68% and a worst of April 2022 at −36.68%. The rule's April 2026 return was 27.76%, and its August 2026 return was 19.78%.

The months that stand out for losses are all in 2022 and early 2025: June 2022 at −26.87%, February 2022 at −14.72%, April 2022 at −10.28%, and March 2025 at −22.85%. Each was a month when the rule held a position that kept falling. It does not exit on the way down. It waits for RSI to come back above 70, and RSI above 70 requires a rally.

The rule's good months, October 2021 at 14.73%, January 2023 at 22.55%, November 2023 at 32.09%, September 2024 at 19.06% and April 2026 at 27.76%, each fell inside one bounce. Several months in a row are at 0% between them, for instance from February to July 2023 and from October 2024 to January 2025.

Every trade

RSI mean reversion on TQQQ made 16 closed round trips, an average hold of 39 days, an average winner of 16.71%, an average loser of −25.02%, a profit factor of 1.83, a longest losing streak of 3. It held a position at the close on 29.8% of trading days.

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.

The 16 trades span 15 to 86 days, with a median hold of 34 days and an average of 39.2. The median return was 13.74%. That is a high median next to a CAGR of 12.08%, which tells us the winners were frequent and the losers were few and large. The four losing trades lost 16.35%, 33.56%, 22.75% and 27.41%. The best trades were 35.06% from 2026-03-30 to 2026-04-15, 27.26% from 2026-07-30 to 2026-08-17 and 21.01% from 2023-09-22 to 2023-11-15.

The best two trades are short, 16 and 18 days. TQQQ returned a lot in a short time after RSI fell below 30. The longest trades, 86, 77 and 69 days, were the 2022 losers. The pattern is that a trade that rebounds does so within roughly two to eight weeks and a trade that fails keeps falling for longer, because the sell signal needs RSI above 70.

The smallest win, 6.28% from 2021-09-21 to 2021-10-25, was entered at an adjusted 32.01 and sold at 34.02 after 34 days. It was the last trade of 2021.

The worst trades show the weakness of a rule with no stop. The 2022-04-25 entry at 19.34 was made with RSI below 30, and the position lost 33.56% over the next 77 days before RSI recovered enough to sell. The 2025-02-28 entry at 35.02 was 61 days long and lost 27.41%, which is the trade behind the 46.92% drawdown that ran from 2025-02-28 to 2025-04-08.

Largest drawdowns

Buy-and-hold's deepest drawdown ran from 2021-11-19 to 2022-12-28 and reached −80.8%.

The largest drawdown was 66.58%, from 2021-10-21 to 2022-11-03, with recovery on 2024-09-19. That is 378 days from peak to trough and 686 days from trough to recovery. Holding's largest was 80.77% from 2021-11-19 to 2022-12-28, recovered on 2024-12-04, so the rule's drawdown began earlier, ended earlier and was shallower by 14.19 points.

The second drawdown was 46.92% from 2025-02-28 to 2025-04-08, recovered on 2026-04-14 after 371 days. Holding fell 57.4% over a similar stretch, from 2024-12-16 to 2025-04-08, recovering on 2025-08-12. The rule recovered later than the fund even though its fall was smaller, because it bought on 2025-02-28 and held a position that lost 27.41% on exit, and it then stayed flat from 2025-04-30 until 2025-11-19 while the fund rose. The third drawdown was 20.65%, from 2021-03-01 to 2021-03-08, and recovered in 28 days.

Both big drawdowns come from a trade that was entered on an oversold signal and kept falling. The rule's design makes that likely on a 3x fund: the fund's average daily range was 4.74%, and RSI(14) can stay below 30 for many days. The caveat on the strategy's own page says the same thing: in a persistent downtrend RSI can stay oversold for weeks while the position keeps losing, and there is no stop-loss in this template.

With trading costs

The headline run fills at the bar price. These runs charge slippage on every fill.

The cost runs add 5 and 10 basis points per fill. At 5 basis points the CAGR was 11.8%, the drawdown 66.69% and the ending value $18,968. At 10 basis points the CAGR was 11.49%, the drawdown 66.81% and the ending value $18,673. The Sharpe went from 0.48 to 0.472 and 0.466.

The effect is small because the rule made only 32 fills over 5.7 years. The ending value dropped by a few hundred dollars across both cost runs. TQQQ trades heavily, with average daily dollar volume of $4,150,045,180 and a median minute volume of 285,786.5 shares, so the assumed costs are modest for a fund this liquid. The cost question does not change the ranking: the rule still trails holding at every cost level.

Changing the parameters

Four variants change one threshold. A tighter buy level of RSI below 25 gave 5.85% a year with a 63.86% drawdown over 12 trades, 8 of them wins. A looser buy level of RSI below 35 gave 8.79% with a 71.97% drawdown over 20 trades and 16 wins. Both returned less than the published rule at 12.08%.

Changing the sell level helped. Selling above 65 instead of 70 gave 14.43% with a 66.14% drawdown, 17 trades and 13 wins, the highest of the five versions. Selling above 75 gave 12.64% with a 63.05% drawdown, 14 trades and 10 wins. Selling earlier produced a higher return and selling later produced a slightly shallower drawdown.

The table does not show a stable optimum. Buying at 25 gave fewer trades and less return, buying at 35 gave more trades and a deeper drawdown, and the sell level changed return by about two points. With 12 to 20 trades per version, each result is driven by a handful of entries, such as the 2022 losers that every version shares in some form. The variants show sensitivity and do not prove that 65 is better than 70, because a threshold chosen after looking at this window is fitted to it. RSI(2) snapback on the same fund returned 39.31% with a 43.2% drawdown, which shows a shorter RSI and a quicker exit behaved differently.

How TQQQ behaved

TQQQ gives three times the daily move of QQQ, and the numbers show what that does. Annualized volatility was 67.11% and the maximum drawdown 81.68%. Calendar returns were 90.95% in 2021, −79.07% in 2022, 198.32% in 2023, 58.22% in 2024, 34.36% in 2025 and 54.6% in 2026 so far. The average daily range was 4.74% and the average up and down days were 3.01% and −3.22%.

RSI was a strong signal on this fund, stronger than the rule's result suggests. After 18 sessions with RSI(14) below 30, the median forward 5-day move was 10.3% and the 20-day move 6.9%, against baselines of 1.24% and 3.62%. After 152 sessions with RSI(2) below 10, the medians were 1.73% and 4.54%. The oversold bounce is large on average and shows up in the 12 winning trades. The problem is the exit, which waits for RSI above 70 and takes whatever comes between, and the entries in 2022, which landed in a fall that RSI(14) did not mark as the low.

The fund was above its 200-day average on 69.48% of sessions and crossed it 28 times. It had 56 falls of 10% or more from a 20-day high, covering 478 days. Compared with QQQ, TQQQ returned 291.92% in total over the window, against 151.07% for the underlying, while a simple 3x of the underlying is 453.22% and a daily rebalanced ideal is 568.66%. The gap to the simple figure was −161.29 points, with large positive and negative years: 33.9 points in 2023 and −27.96 in 2025.

Among leveraged funds, this rule did best on TECL at 23.89% and FAS at 22.26%, and lost money on UST at −5.04% and TMF at −25.04%. TQQQ at 12.08% came just behind SOXL at 14.43% and ROM at 12.76%, and ahead of QLD at 8.78%. The median across all 59 funds was 2.98% and across the leveraged category 12.08%, which is the figure for TQQQ itself. See the other strategies on TQQQ for comparison.

Overnight returns, seasonality and peers

TQQQ made most of its gain while the market was closed. The overnight log return was 97.97% and the intraday log return 32.89%, which puts 74.87% of the fund's return in the gap between close and open. With 29.8% exposure, the rule was invested for a minority of sessions and did not earn the gaps on the rest.

Month of the year gives a second view. May averaged 14.75% for TQQQ, November 12.06%, June 7.24%, October 6.79% and July 6.16%. September averaged -5.42%, February -3.82% and December -2.51%. The rule's worst month was September 2022 at -29.88%, which falls in the fund's weakest calendar month, and its best was November 2023 at 32.09%, in one of the strongest. Each calendar month has five or six observations, so this describes the window and carries no forecast. The day of the week mattered less: Monday averaged 0.53% and Wednesday 0.28%, while Tuesday, Thursday and Friday were between 0.03% and 0.06%. The fund rose on 54.47% of sessions.

The trade sequence also shows a streak pattern. The longest winning run was 6 trades and the longest losing run was 3, and the 3 losses were the three 2022 exits in a row. After those, the rule won 3 trades in 2023 and 3 in 2024. The rule's losing record is one cluster, so its overall result depends on that one year more than the 75% win rate suggests.

Next to the other templates on TQQQ, RSI mean reversion at 12.08% sat just ahead of the dip buyer at 11.91%, which had a 79.56% drawdown against this rule's 66.58%. The RSI(2) snapback led at 39.31% with a 43.2% drawdown, and momentum breakout was last at 0.57%. The 3-month momentum switch returned 7.66% and the golden cross 7.33%. A slow exit rule with no stop landed in the middle of the group, and the RSI(2) snapback, the weekly target, trend plus trailing stop and the 200-day filter all did better here.

Across the 10 leveraged funds in the category, the same rule made between 12 and 20 round trips, with TQQQ at 16. The median CAGR for the category was 12.08%, equal to TQQQ's, and the median across all 59 funds was 2.98%. The 2x peers QLD at 8.78%, SSO at 7.4% and SPUU at 5.38% had drawdowns of 48.1%, 34.85% and 34.59%, shallower than TQQQ's 66.58%.

The rules

Buy when the 14-day RSI drops below 30 (oversold), sell when it recovers above 70 (overbought).

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.

The rule has two lines and no stop, so its behaviour follows from how RSI(14) behaves on a 3x fund. RSI measures the ratio of recent gains to recent losses over 14 sessions. On TQQQ, with an average daily range of 4.74%, RSI swings quickly: it was above 70 on 120 sessions and below 30 on only 18. The entry is rare and the exit is common.

Entries are rare, but each position is held until RSI returns above 70, which takes weeks, and exposure was 29.8%. A fund that runs up for months, as TQQQ did in 2023 when it returned 187.6%, left the rule in cash from February to July. The rule is a dip buyer that sells the first strong rally. It captures dips, which pay on TQQQ, and misses trends, which pay more.

On a fund with deep trends and high volatility, the sell signal at 70 comes early in a rally and the entry at 30 comes early in a fall. That asymmetry shows in the trade list: winners of 6% to 35% in a few weeks and losers of 16 to 34% over two or three months. This template has no stop, and the variants in the parameter section change thresholds and not the exit structure. The comparison with a trend filter on the same fund shows the other approach: it returned 25.03%, and the 200-day regime filter returned 24.95%.

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.