DeployQuant3-Month Momentum Switch
Hold while the trailing 3-month return is positive (above +5% to enter, below 0% to exit).
The 3-month momentum switch holds a fund while its trailing 63-day return is positive. It buys at the open after the 63-day return closes above 5%, with 98% of the sleeve, and sells everything at the open after that return closes below 0%. The test runs the same two rules on 59 ETFs from 2021-01-04 to 2026-10-02, 5.74 years, starting each run with $10,000. The headline run has no fees or slippage and no margin. Cost runs add 5 and 10 basis points.
The median CAGR across the 59 funds is 0%. That figure is pulled to zero by a cluster of funds where the rule never traded or barely traded, and the spread around it is wide. Positive CAGR appears in 29 funds. The median max drawdown is 26.23%, the median Sharpe is 0.09, the median fund made 12 round trips, and the median exposure was 44.7%. The rule beat buy-and-hold of the same fund in 20 of 59 cases and drew down less in 55.
Two features of the result recur through the page. The rule is invested more of the time than a dip-buying rule, so its drawdowns are closer to those of the funds it trades. It also reacts slowly, so it tends to buy after a rise has started and sell after a fall has begun. The RSI mean reversion page tests the opposite style on the same 59 funds and has a higher median CAGR of 2.98%, and the compare page for the pair sets the two next to each other.
This is one window with one deep bear market. The numbers describe what the rule did here and do not describe what a momentum rule does in general.
The rules
- WHEN the market opens · IF not invested AND the 63-day return > +5% · THEN buy with 98% of the sleeve
- WHEN the market opens · IF invested AND the 63-day return < 0% · THEN sell the whole position
Time-series momentum on a quarterly lookback, the horizon much of the academic momentum literature uses. The template enters after a +5% three-month run and exits when the same measure turns negative. The gap between entry (+5%) and exit (0%) is a buffer against flip-flopping around a single threshold.
Good for: assets with long, persistent cycles, such as index, sector and managed-futures ETFs.
Watch out: a three-month lookback is slow; V-shaped crashes and recoveries can see it exit near the bottom and re-enter well off the low.
How the entry and exit buffer works
The 63-day return is the percentage change in price over the last 63 trading days, about one quarter. The entry threshold of 5% and the exit threshold of 0% leave a gap. A fund that has risen 3% over the quarter is not bought, and a fund already held at that reading is not sold. The gap is there to stop the rule from entering and exiting on small wiggles around a single line.
The buffer has a cost that shows up in the data. Because the rule waits for a 5% run, it misses the first part of any recovery. Because it waits for the full quarter to turn negative, it holds through the first part of any decline. On a fund that rises in a straight line this does little harm. On a fund that falls fast and rebounds fast, the rule exits late and re-enters late. The strategy page's own caveat describes this: V-shaped moves can leave it selling near the bottom and buying back well off the low.
The three parameter variants in the facts all move the entry threshold. Setting entry at 0% removes the buffer and produces the most trades. Setting entry at 10% or 15% widens it and produces far fewer. The results of those variants appear further down.
There is no stop-loss between the 0% exit and the current price. If a fund falls 15% in two weeks after a strong quarter, the 63-day return can remain positive and the position stays open. That explains why the median drawdown of 26.23% is not much better than what many of these funds produced when held. The 200-day regime filter and the trend plus trailing stop rule are the closest siblings in the facts, with median CAGRs of 1.84% and 2.61%.
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.
Results on every ETF
| ETF | CAGR | buy & hold | max DD | Sharpe | trades | win rate |
|---|---|---|---|---|---|---|
| SOXL | 16.4% | 33.3% | −74.2% | 0.58 | 26 | 42% |
| SOXX | 14.3% | 31.1% | −35.4% | 0.64 | 16 | 44% |
| CLSE | 11.5% | 19.5% | −8.0% | 1.18 | 7 | 71% |
| QLD | 10.8% | 23.6% | −36.1% | 0.52 | 18 | 44% (+1 open) |
| SPUU | 9.8% | 22.5% | −38.6% | 0.58 | 18 | 56% (+1 open) |
| VOX | 8.9% | 9.1% | −13.0% | 0.77 | 11 | 64% (+1 open) |
| VOOG | 8.8% | 15.7% | −20.8% | 0.74 | 13 | 62% (+1 open) |
| SSO | 8.6% | 21.9% | −39.5% | 0.53 | 18 | 50% (+1 open) |
| IAU | 8.5% | 13.7% | −21.7% | 0.63 | 9 | 56% (+1 open) |
| ROM | 8.4% | 30.2% | −50.3% | 0.41 | 21 | 38% (+1 open) |
| IOO | 8.4% | 16.7% | −15.5% | 0.84 | 11 | 55% (+1 open) |
| VOO | 8.3% | 14.4% | −18.4% | 0.86 | 9 | 78% (+1 open) |
| SPY | 8.2% | 14.6% | −18.6% | 0.86 | 9 | 78% (+1 open) |
| QQQ | 8.2% | 16.7% | −21.5% | 0.65 | 12 | 50% (+1 open) |
| QQQM | 8.0% | 16.7% | −21.8% | 0.63 | 12 | 50% (+1 open) |
| VV | 8.0% | 14.3% | −17.8% | 0.83 | 11 | 64% (+1 open) |
| TQQQ | 7.7% | 25.4% | −55.4% | 0.39 | 23 | 48% (+1 open) |
| TECL | 7.6% | 38.2% | −69.9% | 0.40 | 26 | 38% (+1 open) |
| XLK | 6.1% | 22.1% | −30.9% | 0.45 | 18 | 33% (+1 open) |
| VTV | 5.7% | 13.4% | −11.8% | 0.72 | 10 | 60% |
| ALTY | 4.5% | 7.6% | −8.3% | 0.83 | 6 | 67% |
| XLF | 2.7% | 12.6% | −23.0% | 0.30 | 15 | 60% |
| VOOV | 2.4% | 12.6% | −16.5% | 0.33 | 13 | 69% |
| QAI | 2.1% | 3.9% | −5.0% | 0.61 | 4 | 50% |
| KMLM | 1.3% | 7.1% | −28.7% | 0.19 | 9 | 33% (+1 open) |
| TBF | 1.2% | 12.2% | −23.1% | 0.16 | 12 | 33% (+1 open) |
| FXE | 1.0% | −0.9% | −8.1% | 0.28 | 4 | 50% |
| USDU | 0.5% | 5.2% | −6.5% | 0.16 | 3 | 33% |
| IGIB | 0.3% | 0.2% | −5.0% | 0.13 | 4 | 50% |
| SGOV | 0.0% | 3.2% | −0.0% | 0.00 | 0 | – |
| RINF | −0.4% | 6.6% | −12.8% | -0.00 | 10 | 50% |
| UDN | −0.5% | −1.1% | −9.7% | -0.15 | 4 | 50% |
| IEI | −0.6% | −0.4% | −3.8% | -0.51 | 2 | 0% |
| AGG | −0.7% | −0.8% | −5.7% | -0.40 | 3 | 0% |
| BND | −1.0% | −0.8% | −6.5% | -0.55 | 3 | 0% |
| IEF | −1.4% | −2.3% | −9.1% | -0.52 | 4 | 25% |
| CTA | −1.6% | 8.9% | −23.3% | -0.06 | 9 | 44% (+1 open) |
| IWM | −1.6% | 7.5% | −26.8% | -0.06 | 20 | 40% |
| EEM | −1.6% | 6.6% | −32.7% | -0.06 | 13 | 23% |
| XLP | −1.7% | 5.8% | −23.6% | -0.16 | 13 | 23% |
| TLT | −2.3% | −8.2% | −18.6% | -0.33 | 7 | 29% |
| FAS | −2.5% | 18.3% | −58.1% | 0.09 | 25 | 40% |
| QQQE | −2.9% | 9.8% | −27.6% | -0.20 | 19 | 37% |
| XLY | −3.9% | 6.3% | −34.6% | -0.24 | 17 | 35% |
| UST | −4.0% | −8.5% | −26.2% | -0.57 | 9 | 11% |
| PSQ | −4.3% | −13.8% | −35.3% | -0.22 | 10 | 20% |
| SPDN | −4.8% | −9.7% | −29.5% | -0.41 | 8 | 13% |
| SH | −4.8% | −10.0% | −29.5% | -0.42 | 8 | 25% |
| EEV | −9.5% | −16.1% | −55.3% | -0.32 | 17 | 18% |
| SDS | −11.6% | −21.6% | −55.4% | -0.46 | 13 | 8% |
| TMF | −13.2% | −31.2% | −63.6% | -0.59 | 17 | 18% |
| VIXM | −13.8% | −16.1% | −58.2% | -0.71 | 20 | 15% |
| VXZ | −14.2% | −15.1% | −58.9% | -0.78 | 22 | 14% |
| REW | −15.7% | −36.1% | −70.8% | -0.34 | 18 | 11% |
| QID | −17.2% | −29.5% | −71.8% | -0.49 | 22 | 14% |
| TECS | −20.7% | −46.7% | −77.9% | -0.28 | 18 | 17% |
| SQQQ | −25.8% | −42.3% | −86.7% | -0.45 | 25 | 16% |
| SOXS | −41.1% | −48.3% | −96.0% | -0.61 | 27 | 22% |
| UVXY | −49.8% | −48.7% | −98.1% | -1.05 | 27 | 11% |
Which funds the rule worked on
SOXL is the best result at 16.43% a year, with a 74.16% max drawdown and 26 round trips. Buy-and-hold on SOXL made 33.32% in the same window, so the rule kept part of the return on the leveraged semiconductor fund at a drawdown that is still deep. SOXX, the unleveraged semiconductor fund, returned 14.3% with a 35.43% drawdown against 31.11% for holding.
CLSE is the cleanest entry in the table. It returned 11.46% with a 7.95% drawdown over 7 round trips and a 71% win rate, and it was invested 61% of the time. Buy-and-hold on CLSE made 19.49%. The shallow drawdown reflects the fund's own design as a long and short fund, and the rule added little risk to it.
The broad index funds sit in a tight band. SPY returned 8.21% with an 18.58% drawdown, QQQ returned 8.21% with a 21.53% drawdown, and VOO returned 8.25% with an 18.39% drawdown. Holding made 14.56% on SPY, 16.7% on QQQ and 14.42% on VOO. SPY and VOO each made 9 round trips with a 78% win rate and were invested 65.7% of the time. The rule kept part of the buy-and-hold return on these funds while sitting out a share of the days.
The weak results cluster in three places. The first is funds that went sideways or had short cycles: IWM lost 1.59% a year across 20 round trips with a 40% win rate, while holding made 7.5%. EEM lost 1.62% with a 23% win rate. The second is the inverse and volatility funds, where UVXY lost 49.75% a year with a 98.08% drawdown, SOXS lost 41.12% with a 96.01% drawdown and SQQQ lost 25.78% with an 86.72% drawdown. UVXY finished slightly behind holding, which lost 48.74%. The third is the leveraged funds that whipsawed: FAS lost 2.46% a year over 25 round trips with a 40% win rate, while holding FAS made 18.34%, and TECL made 7.63% against 38.16% for holding.
Win rates are low across the board for a momentum rule. SOXL won 42% of its 26 trades, QQQ won 50% of 12, and XLK won 33% of 18. This is the usual shape of trend rules: many small losses on false starts and a few large gains. The payoff of the winning trades has to cover the losers, and on the leveraged funds it did in some cases and not in others.
One fund has no trades. SGOV is a Treasury bill fund and never produced a 5% quarterly return, so the rule stayed in cash, with 0% CAGR and 0% drawdown against 3.22% for holding.
Results by fund type
| Fund type | ETFs | Median CAGR | Median buy & hold | Median max DD | Beat holding |
|---|---|---|---|---|---|
| Broad index ETFs | 12 | 8.0% | 14.4% | −20.8% | 0 of 12 |
| Sector ETFs | 6 | 6.1% | 12.6% | −30.9% | 0 of 6 |
| Leveraged ETFs | 10 | 8.4% | 23.6% | −55.4% | 2 of 10 |
| Inverse ETFs | 11 | −11.6% | −21.6% | −55.4% | 10 of 11 |
| Bond ETFs | 7 | −0.7% | −0.8% | −5.7% | 4 of 7 |
| Commodity ETFs | 1 | 8.5% | 13.7% | −21.7% | 0 of 1 |
| Currency ETFs | 3 | 0.5% | −0.9% | −8.1% | 2 of 3 |
| Volatility products | 3 | −14.2% | −16.1% | −58.9% | 2 of 3 |
| Alternative-strategy ETFs | 6 | 2.1% | 7.6% | −12.8% | 0 of 6 |
Results by fund category
The leveraged group has a median CAGR of 8.45% against a median buy-and-hold of 23.55%. Its median drawdown is 55.44%, and the rule beat holding on 2 of 10. Sector funds have a median of 6.06% against 12.58% with a 30.88% median drawdown, and none of the 6 beat holding.
Broad index funds have the most useful median for a reader choosing a core holding. It is 8% against 14.42% for holding, with a 20.82% median drawdown and none of the 12 ahead of holding. Alternative-strategy funds have a median of 2.11% against 7.61% and a 12.8% median drawdown, again with no wins. The commodity group is IAU alone, at 8.51% against 13.66% with a 21.73% drawdown.
Currency funds have a median of 0.49% against negative 0.9% for holding, and 2 of 3 beat holding. Bond funds have a median of negative 0.74% against negative 0.78% for holding, with 4 of 7 ahead and a 5.72% median drawdown, the shallowest of the equity-like and bond groups. The bond funds spent little time invested, with exposure as low as 8% for IEI and 10.6% for AGG, so the rule behaved close to cash on them.
Inverse funds have a median CAGR of negative 11.61% against negative 21.57% for holding, with a median drawdown of 55.35%. The rule beat holding on 10 of 11. Volatility products have a median of negative 14.19% against negative 16.06%, a 58.88% median drawdown, and 2 of 3 ahead. In both groups the rule lost money. Its edge over holding came from being out of the fund for most days: UVXY was invested only 14.8% of the time and SOXS 17.9%.
The pattern is consistent. The rule beat holding mainly where holding was negative, and it trailed holding in every group where holding earned a strong return. A momentum rule with a 63-day lookback pays for downside protection with a lag on the way back up. The compare page against the RSI(2) snapback shows the same trade against a faster mean-reversion rule that has a median CAGR of 4.74%.
Year by year, median across all ETFs
| Year | 3-month momentum | Buy & hold | ETFs with a gain |
|---|---|---|---|
| 2021 | 0.0% | 4.1% | 27 of 59 |
| 2022 | −14.7% | −12.7% | 8 of 59 |
| 2023 | 0.0% | 8.9% | 29 of 59 |
| 2024 | 0.0% | 9.7% | 24 of 59 |
| 2025 | 1.1% | 11.1% | 31 of 59 |
| 2026 | 0.0% | 3.7% | 22 of 59 |
Year by year across the 59 funds
The year-by-year medians show a rule that did not help in the one year where help was most needed. In 2021 the median fund returned 0% under the rule against 4.1% for holding, with 27 funds positive. In 2022 the median was negative 14.7%, against negative 12.7% for holding, and only 8 funds were positive. That is the lowest count of any year and the one year where the rule's median did worse than holding.
The cause is the lag. A 63-day lookback that turns negative only after a quarter of decline leaves the position open through the first part of a bear market. The rule then exits late and, when prices bounce, stays out until the 63-day return clears 5% again. In a year with repeated bear-market rallies, that sequence repeats and loses on each cycle. The entry buffer of 5% makes the re-entry later still.
In 2023 the median was 0% against 8.9% for holding, with 29 funds positive. In 2024 it was 0% against 9.7%, with 24 positive. In 2025 it was 1.1% against 11.1%, with 31 positive. The median of exactly 0% in four of the six years comes from the many funds that sat in cash for an entire calendar year, since the median across 59 funds lands on a fund with no trades. The partial year 2026 shows 0% against 3.7% with 22 funds positive.
The yearly medians say that most of the 59 funds were out of the market in a given year more often than they were in it. The median exposure over the whole window is 44.7%. Funds that did well under the rule are the ones whose trends lasted more than a quarter, and the table of funds shows where that happened: broad index funds and semiconductor funds in particular.
The medians cannot show how a single fund moved inside a year. A fund that rose 20% in three months and fell 15% in the next three counts as one trade pair, and the median can hide it. The ETF pages linked from the results table list the year-by-year returns for each fund.
Changing the parameters
| Version | Median CAGR | Median max DD | Median round trips |
|---|---|---|---|
| Published rules | 0.0% | −26.2% | 12 |
| Enter above 0% | 1.0% | −29.9% | 35 |
| Enter above 10% | 0.0% | −21.3% | 7 |
| Enter above 15% | 0.0% | −20.1% | 4 |
What changing the entry threshold does
Three variants change only the entry threshold and keep the exit at 0%. The facts report the median across all 59 funds for each.
Entering above 0% removes the buffer. The median CAGR is 0.96%, the median drawdown is 29.86%, and the median fund made 35 round trips. That is far more trades than the base rule at 12, and the drawdown is deeper than the base median of 26.23%. With the entry and exit at the same level, the rule flips every time the 63-day return crosses zero, and each flip is a chance to buy a small bounce that fades.
Entering above 10% gives a median CAGR of 0%, a median drawdown of 21.27%, and 7 round trips. Entering above 15% gives a median CAGR of 0%, a median drawdown of 20.15%, and 4 round trips. The higher thresholds trade less and draw down less, and the median return stays at 0% because the median fund barely traded. A 15% quarterly gain is a rare event on most of these funds, so the rule sat in cash on many of them.
The base 5% setting sits between the two extremes: 12 round trips and a 26.23% median drawdown. It is not obviously better than the others on the median CAGR, which is 0% for the base rule and for the two higher thresholds. Only the 0% entry shows a positive median, and it takes a deeper drawdown to get it.
These medians do not rank the settings on a single fund. A fund with a strong trend, such as SOXL, may do better at a higher entry threshold because each signal is a stronger trend. A choppy fund such as IWM can lose on all of them. The 63-day lookback itself is also a parameter, and the facts do not include a sweep of it. The monthly cycle rule and the weekly 7% target rule have median CAGRs of 5.46% and 6.61%, which is higher than any momentum setting here.
Frequently asked questions
What is the 3-month momentum strategy?
Hold while the trailing 3-month return is positive (above +5% to enter, below 0% to exit). Time-series momentum on a quarterly lookback, the horizon much of the academic momentum literature uses. The template enters after a +5% three-month run and exits when the same measure turns negative. The gap between entry (+5%) and exit (0%) is a buffer against flip-flopping around a single threshold.
Does 3-month momentum beat buy-and-hold?
Across 59 ETFs backtested 2021-01-04 to 2026-10-02, it beat same-ETF buy-and-hold on 20 of 59 (34%). Median CAGR was 0.0% with a median max drawdown of 26.2%. Per-ETF results vary widely; the table lists every one.
Why 63 days?
63 trading days is about one quarter, a common momentum lookback. You can sweep it in DeployQuant to see how the horizon changes results.
Does 3-month momentum beat buy-and-hold on ETFs?
In this test it beat buy-and-hold in 20 of 59 funds, and none of the 12 broad index funds. The median CAGR was 0%, and the median max drawdown was 26.23%. It had a shallower drawdown than holding in 54 of 59 cases.
What are the rules of the 3-month momentum switch?
It buys at the next open when the 63-day return is above 5% and sells everything at the next open when that return falls below 0%. It uses 98% of the sleeve when it buys. There is no stop-loss and no profit target.
Which ETFs did the momentum switch work on best?
SOXL returned 16.43% a year, SOXX 14.3% and CLSE 11.46%. CLSE did it with a 7.95% max drawdown. SPY and QQQ each returned 8.21%, against 14.56% and 16.7% for holding.
How did the strategy do in 2022?
The median fund lost 14.7% under the rule against 12.7% for holding, and only 8 funds were positive. This was the only year the median did worse than holding, because the 63-day lookback exits after the decline has started.
What happens if I enter at 0%, 10% or 15% instead of 5%?
Entering above 0% gave a median CAGR of 0.96% with 35 round trips and a 29.86% median drawdown. Entering above 10% gave 0% with 7 round trips, and above 15% gave 0% with 4. Higher thresholds traded less and drew down less.
Why is the median CAGR 0%?
Many funds produced no trades or very few, so the middle of the ranking sits at zero. SGOV never entered at all. Funds with long trends, such as semiconductors and broad index funds, produced the positive returns.
Compare with other strategies
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.