DeployQuant3-Month Momentum Switch on SPY
SPDR S&P 500 ETF Trust: the most-traded S&P 500 ETF and a common benchmark for US large caps. Backtest 2021-01-04 to 2026-10-02, $10,000 starting capital, computed by the same engine that runs live DeployQuant strategies.
The 3-month momentum switch on SPY turned $10,000 into $15,728 from 2021-01-04 to 2026-10-02, an 8.21% CAGR with an 18.58% maximum drawdown and a Sharpe of 0.86. Buy-and-hold SPY over the same dates ended at $21,820, a 14.56% CAGR, with a 23.99% drawdown and a Sharpe of 0.95. The switch trailed holding by 6.35 points a year and had a drawdown 5.41 points shallower.
The strategy made 9 closed round trips and one position still open at the end. Seven of the 9 won, a 78% win rate, with an average win of 8.27% and an average loss of 6.05%. Its profit factor was 4.95. It was invested 65.7% of the time and made 19 fills in 5.74 years, so it is a slow strategy: one trade every few months, with long holds in between.
It ranked eighth of the 12 templates on SPY by CAGR. The weekly 7% target led at 13.88%, then RSI(2) snapback at 13.17% and the monthly cycle at 12.84%. Across all 59 ETFs the template's median CAGR was 0%, and SPY was its thirteenth best fund. The headline run has no fees or slippage. Everything below describes one window that contains a 2022 bear market and a sharp drop in April 2025.
Year by year
| Year | 3-month momentum | buy & hold |
|---|---|---|
| 2021 | 13.9% | 29.6% |
| 2022 | −16.9% | −17.8% |
| 2023 | 11.0% | 25.5% |
| 2024 | 24.0% | 24.3% |
| 2025 | 13.7% | 17.4% |
| 2026 | 6.2% | 13.6% |
Year by year against holding SPY
The switch beat holding in one year, 2022, and trailed it in the other five.
In 2021 it returned 13.9% against 29.6%. The gap of 15.7 points comes mostly from the months before the first entry, since there was no position in January, February or March. It held from 2021-04-07 to 2021-10-05 for a 6.64% gain, was out for about two weeks, and re-entered on 2021-10-18.
In 2022 it lost 16.9% while holding lost 17.8%, a gap of 0.9 points in its favor. That is almost no advantage in the one year the strategy was designed to protect against. The 2022 results by month are a loss of 6.14% in January, then 0.0% from February to July, a loss of 4.39% in August, 7.34% in September, and 0.06% in December. The January loss came from the position held through the first leg of the decline until the exit on 2022-01-21. The August and September losses came from two re-entries into a market that kept falling. The months between were flat. The switch avoided February to July and gave back much of the benefit in the autumn.
In 2023 it returned 11.0% against 25.5%, a gap of 14.5 points. It entered on 2022-12-27 at 364.98, and the three trades that closed in 2023 all won: 4.46%, 0.77% and 4.41%. It was out in October and November 2023, and the monthly table shows 0.0% for both. November 2023 was a strong month for the market, so the exit cost return there. It re-entered on 2023-12-15.
In 2024 it returned 24.0% against 24.3%, a gap of 0.3 points. The strategy was invested for the whole year in the one position from 2023-12-15 to 2025-02-28, and that is why the year matches holding. The monthly pattern shows it: February 5.01%, March 3.18%, April a loss of 3.90%, May 4.87% and November 5.78%.
In 2025 it returned 13.7% against 17.4%. It sold on 2025-02-28 for a gain of 26.26% on the long trade, and was out for March, April and May, which show 0.0%. Holding's drawdown reached 18.41% by 2025-04-08. The switch re-entered on 2025-06-10.
In 2026, through 2026-10-02, it returned 6.2% against 13.6%. The position from 2025-06-10 was closed on 2026-03-04 at 677.46 for 14.66%. The strategy re-entered on 2026-05-06 at 724.41, and that position was up 6.25% at the end. April 2026 shows 0.0%, and that was holding's best month in the window at 10.33%. The strategy missed it.
The yearly gaps were negative 15.7, plus 0.9, negative 14.5, negative 0.3, negative 3.7 and negative 7.4 points. The two largest gaps came in the years of the first entry delay and of the 2023 rebound.
Month by month
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | 0.0% | 0.0% | 0.0% | 2.6% | 0.6% | 2.1% | 2.3% | 2.8% | −4.4% | 3.6% | −0.7% | 4.4% |
| 2022 | −6.1% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | −4.4% | −7.3% | 0.0% | 0.0% | −0.1% |
| 2023 | 6.0% | −1.6% | 1.3% | 1.6% | −1.9% | 6.3% | 3.1% | −1.6% | −3.6% | 0.0% | 0.0% | 1.2% |
| 2024 | 1.5% | 5.0% | 3.2% | −3.9% | 4.9% | 3.4% | 1.1% | 2.3% | 2.1% | −0.9% | 5.8% | −2.3% |
| 2025 | 2.6% | −2.7% | 0.0% | 0.0% | 0.0% | 3.0% | 2.2% | 2.0% | 3.4% | 2.3% | 0.2% | 0.1% |
| 2026 | 1.4% | −0.8% | −0.5% | 0.0% | 3.8% | −1.1% | 0.1% | 2.6% | −0.3% | 0.9% | – | – |
The months that mattered
The worst month for the switch was September 2022 at negative 7.34%. That month came from the entry on 2022-09-13 at 380.30 and the exit on 2022-09-23 at 351.42, a 10-day trade that lost 7.59%. Holding's worst month was also September 2022, at negative 9.00%.
The best month was June 2023 at 6.25%. The other good months were November 2024 at 5.78%, February 2024 at 5.01%, May 2024 at 4.87% and December 2021 at 4.44%. Holding's best month was April 2026 at 10.33%, which the switch missed completely because it was out until 2026-05-06.
Flat months were common. January to March 2021, February to July 2022, October and November 2022, October and November 2023, and March to May 2025 all show 0.0%, and so does April 2026. The strategy held nothing in those months. The flat months in 2022 and 2025 avoided market declines, and the flat months in 2023 and 2026 missed rises.
The pattern shows how the rule trades off. It sat out most of the 2022 decline, except for two short stretches, and sat out the April 2025 low. It also sat out the rebound that followed each one. The record of the 78% win rate hides this: the strategy won most of its trades and still returned less than holding, because its time out of the market included strong months.
SPY's seasonal averages in the window were highest in November at 4.01%, July at 3.07% and May at 3.00%, and lowest in September at negative 2.21%. With 5 or 6 observations per month, these are weak evidence. The switch's own September record, a loss in 2021 of 4.43% and a loss in 2022 of 7.34%, fits the pattern for those years only.
Every trade
3-month momentum on SPY made 9 closed round trips and one position still open at the end of the test, an average hold of 137 days, an average winner of 8.27%, an average loser of −6.05%, a profit factor of 4.95, a longest losing streak of 2. It held a position at the close on 65.7% of trading days.
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2021-04-07 | $377.30 | 2021-10-05 | $402.35 | 6.6% | 181 |
| 2021-10-18 | $414.99 | 2022-01-21 | $417.72 | 0.7% | 95 |
| 2022-08-11 | $399.91 | 2022-08-30 | $381.86 | −4.5% | 19 |
| 2022-09-13 | $380.30 | 2022-09-23 | $351.42 | −7.6% | 10 |
| 2022-12-27 | $364.98 | 2023-02-27 | $381.26 | 4.5% | 62 |
| 2023-03-30 | $386.64 | 2023-05-04 | $389.62 | 0.8% | 35 |
| 2023-05-23 | $399.14 | 2023-09-22 | $416.74 | 4.4% | 122 |
| 2023-12-15 | $454.45 | 2025-02-28 | $573.80 | 26.3% | 441 |
| 2025-06-10 | $590.82 | 2026-03-04 | $677.46 | 14.7% | 267 |
| 2026-05-06 | $724.41 | open | – | 6.3% | – |
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 trade list
The switch made 10 entries, 9 of them closed. The two trades that carry the result are the long ones. The first was entered on 2023-12-15 at 454.45 and sold on 2025-02-28 at 573.80, a gain of 26.26% over 441 days. The second was entered on 2025-06-10 at 590.82 and sold on 2026-03-04 at 677.46, a gain of 14.66% over 267 days. Those two, plus the first trade at 6.64% over 181 days, account for most of the 57.28% total return.
The smaller wins were 4.46% over 62 days from 2022-12-27 to 2023-02-27, 4.41% over 122 days from 2023-05-23 to 2023-09-22, 0.77% over 35 days from 2023-03-30 to 2023-05-04, and 0.66% over 95 days from 2021-10-18 to 2022-01-21. The two losses came in 2022: 4.51% over 19 days from 2022-08-11 to 2022-08-30 and 7.59% over 10 days from 2022-09-13 to 2022-09-23.
The median trade returned 4.41% and the median hold was 95 days. The longest hold was 441 days and the shortest was 10. The average hold was 136.9 days. Of the exits, 2021 had 1 round trip with 1 win, 2022 had 3 with 1 win, 2023 had 3 with 3 wins, and 2025 and 2026 had 1 each with 1 win each.
The 78% win rate is high and it is accurate, and the two losses are large relative to the small wins. An average loss of 6.05% against an average win of 8.27% means a single loss takes back most of a typical win. The profit factor of 4.95 comes from the two long trades.
The open position at the end was entered on 2026-05-06 at 724.41 and was up 6.25% when the test stopped. Its result counts in the CAGR as a mark to the last close.
Prices in the list are adjusted for splits and dividends, so they sit below the quotes printed at the time.
Largest drawdowns
| Peak | Low point | Depth | Days to low | Recovered | Days to recover |
|---|---|---|---|---|---|
| 2022-01-03 | 2022-12-28 | −18.6% | 359 | 2024-03-20 | 448 |
| 2024-07-16 | 2024-08-05 | −8.1% | 20 | 2024-09-19 | 45 |
| 2024-03-27 | 2024-04-19 | −5.2% | 23 | 2024-05-14 | 25 |
Buy-and-hold's deepest drawdown ran from 2022-01-03 to 2022-10-12 and reached −24.0%.
Drawdowns
The switch's largest drawdown was 18.58%, from 2022-01-03 to 2022-12-28. That is 359 days to the low and 448 days to recover, on 2024-03-20. Holding SPY had a 23.99% drawdown from the same peak on 2022-01-03 to a low on 2022-10-12, and recovered on 2023-12-13.
The switch's low came later than the market's. The market bottomed on 2022-10-12, and the switch's equity bottomed on 2022-12-28, the day after it re-entered at 364.98. The losses in August and September 2022 came after the market had already fallen, and they were what carried the equity down to its low.
The second drawdown was 8.14%, from 2024-07-16 to 2024-08-05, recovered on 2024-09-19 after 45 days. The third was 5.19%, from 2024-03-27 to 2024-04-19, recovered on 2024-05-14. Both were in the position held from December 2023. They are small and are typical of holding SPY through a pullback.
Holding's second drawdown was 18.41%, from 2025-02-19 to 2025-04-08, recovered on 2025-06-26. The switch does not show a drawdown of that size, because it had sold on 2025-02-28. This is where the strategy did what it is built to do. It exited a few days after the peak and was flat for the low, and it re-entered on 2025-06-10 at a price above its exit.
The 5.41 point improvement in drawdown, from 23.99% to 18.58%, was bought with a 6.35 point loss of CAGR. The Sharpe ratio of 0.86 was below holding's 0.95, so the shallower drawdown did not make up for the lower return on that measure.
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) | 8.2% | −18.6% | $15,728 | 0.86 |
| 5 basis points | 8.0% | −18.9% | $15,580 | 0.84 |
| 10 basis points | 7.9% | −19.1% | $15,480 | 0.82 |
Costs
The switch made 19 fills. At 5 basis points per fill the CAGR was 8.03% and the drawdown 18.86%, with a final value of $15,580. At 10 basis points it was 7.91%, with a drawdown of 19.15% and a final value of $15,480. The headline run, with no cost, ended at $15,728 with a CAGR of 8.21%.
The cost of trading is small because the strategy trades rarely. The flat charge does not model a spread, and SPY had an average daily dollar volume of $30,453,859,411 and a median minute volume of 107,870 shares in the data. For a strategy this slow, costs are not what separates its result from holding.
Changing the parameters
| Version | CAGR | Max drawdown | Round trips | Win rate | Final value |
|---|---|---|---|---|---|
| Published rules | 8.2% | −18.6% | 9 | 78% | $15,728 |
| Enter above 0% | 8.7% | −26.2% | 22 | 64% | $16,140 |
| Enter above 10% | 6.7% | −12.5% | 6 | 83% | $14,501 |
| Enter above 15% | 5.9% | −8.0% | 2 | 100% | $13,926 |
Changing the entry threshold
The parameter table changes the entry threshold and leaves the exit at 0%.
Entering above 0% instead of 5% returned 8.70% with a 26.22% drawdown and 22 round trips, 14 of them winners. Removing the buffer produced far more trades, a slightly higher CAGR and a deeper drawdown than the published rules. The extra trades are the whipsaws the buffer was meant to avoid. They did not add much return and they added drawdown.
Entering above 10% returned 6.69% with a 12.47% drawdown, 6 round trips and 5 wins. Entering above 15% returned 5.94% with a 7.99% drawdown and 2 round trips, both winners, with a Sharpe of 0.885. Raising the threshold lowered both the return and the drawdown. A higher bar means fewer entries and later ones, so the strategy stays out of more of the market.
The Sharpe ratios were 0.858 at 0%, 0.86 at the published 5%, 0.799 at 10% and 0.885 at 15%. They are close together, which says the choice of threshold changes how much return and risk the strategy takes more than it changes the quality of the trade-off. The 15% version has 2 trades and its Sharpe rests on them. It should be read as a sample of two.
All four versions returned less than holding's 14.56%. This window offers no threshold at which the switch matched buy-and-hold on return.
How SPY behaved
| Measure | SPY |
|---|---|
| Data in this test | 2021-01-04 to 2026-10-02 (1444 sessions) |
| Total return, buy and hold | 125.3% |
| Annualized volatility | 16.4% |
| Deepest drawdown | −24.5% (2022-01-03 to 2022-10-12) |
| Up days | 54.4% |
| Average daily range | 1.15% |
| Average overnight gap | 0.44% |
| Correlation to QQQ | 0.94 |
| Correlation to TLT | 0.08 |
| Sessions above the 200-day average | 78.0% |
| Crossings of the 200-day average | 30 |
| Falls of 10% or more from a 20-day high | 11 |
How SPY fits a momentum rule
SPY returned 15.2% a year in the profile series, with annualized volatility of 16.41%. It rose on 54.4% of days. It spent 77.99% of sessions above its 200-day average and crossed it 30 times. Its calendar years were 30.46% in 2021, negative 18.16% in 2022, 26.18% in 2023, 24.86% in 2024, 17.72% in 2025 and 13.77% in 2026 through 2026-10-02. The deepest drawdown was 24.51% from 2022-01-03 to 2022-10-12, and the longest lasted 488 sessions.
A momentum rule works when the recent past predicts the near future, and in this window SPY did that in long stretches. Four of the six years returned more than 17%, and the switch was long for all of 2024. The first-lag autocorrelation of daily returns was negative 0.02, so day-to-day moves did not predict each other. The rule uses a 63-day window, which is a different horizon from that statistic.
The peer table shows how much the fund matters. The same rule returned 8.80% on VOOG, 8.43% on IOO, 8.25% on VOO, 8.21% on SPY, 8.21% on QQQ, 8.00% on QQQM and 7.95% on VV. It returned 5.70% on VTV, 2.38% on VOOV, negative 1.59% on IWM, negative 1.62% on EEM and negative 2.86% on QQQE. The large-cap index funds clustered between about 8% and 9%, and the small-cap, emerging-market and equal-weight funds lost money. The category median was 8%, while the median across all 59 funds was 0%.
IWM made 20 round trips and QQQE 19, against 9 on SPY. A fund with choppy 3-month returns generates more entries, and each one costs the gap between the entry and the exit. SPY's cleaner trends gave the rule fewer and longer trades.
The other templates on SPY were the golden cross at 8.71%, the 200-day filter at 8.30% and SMA 10/50 at 7.94%, so the 3-month switch sat in the middle of the trend-following group. EMA 12/26 at 10.20% with an 11.84% drawdown was ahead of it on both measures.
The best days for SPY were 2025-04-09 at 9.39% and 2022-11-10 at 5.48%. The worst were 2025-04-04 at negative 5.98% and 2025-04-03 at negative 4.78%. The switch was out of the market on all four. Of SPY's total return, 62.46% came overnight.
The limits are one window of 5.74 years, daily decisions, and no fees in the headline run. Nine closed trades is a small sample, and two of them carry most of the result.
The rules
Hold while the trailing 3-month return is positive (above +5% to enter, below 0% to exit).
- 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 switch decides
The rule compares today's price with the price 63 trading days ago. If the strategy is out and the 63-day return is above 5%, it buys with 98% of the sleeve at the next open. If the strategy is in and the 63-day return is below 0%, it sells the whole position at the next open. There is no stop and no profit target. The gap between the 5% entry and the 0% exit is a buffer, so a return that hovers near zero does not trigger a buy and a sell on consecutive days.
The lookback is slow by design. A 63-day return only turns negative after SPY has fallen below its level of about three months earlier, and it only exceeds 5% after a gain that large has already happened. Every entry therefore comes after part of a rise, and every exit comes after part of a fall. The template's own caveat says that a three-month lookback can exit near a bottom and re-enter well off the low.
SPY shows both effects. On 2025-02-28 the switch sold at an adjusted 573.80. It was out through the drop to the 2025-04-08 low and bought again on 2025-06-10 at 590.82, a price higher than its exit. It missed the decline and also the first stretch of the rebound, and in this case the re-entry price was above the exit price. The other direction shows in 2022: it bought on 2022-08-11 at 399.91 and on 2022-09-13 at 380.30, both during a bear-market rally, and both trades lost money.
The first entry came on 2021-04-07, and the monthly table shows 0.0% for January to March 2021. The first months fit the need for 63 sessions of history, though the facts do not say how the lookback was seeded. For the whole window the strategy held a position on about two thirds of the sessions.
Decisions are made once a day. The rule does not use intraday data, and fills happen on minute bars with no fees or slippage in the headline run.
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 3-month momentum beat buy-and-hold on SPY?
Over 2021-01-04 to 2026-10-02, 3-month momentum on SPY returned 8.2% annualized vs 14.6% for buy-and-hold: it trailed buy-and-hold by 6.3% per year, with a maximum drawdown 5.4 points shallower than holding (18.6% vs 24.0%).
How many trades did it make?
9 completed round trips over 5.7 years (19 fills), with 78% of round trips closing profitably.
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.
Is a 3-month momentum strategy profitable on SPY?
In this backtest it returned 8.21% a year against 14.56% for buy-and-hold SPY, with an 18.58% drawdown against 23.99%. It won 7 of its 9 closed trades. It beat holding in 2022 only, by 0.9 points.
How many trades does the strategy make?
It made 9 closed round trips and one open position in 5.74 years, with 19 fills. The average hold was 136.9 days, and the longest was 441 days.
Why does the strategy trail buy-and-hold on SPY?
The rule needs a 63-day return above 5% to enter, so every entry comes after part of a rise. It was out of the market for months in 2021, 2022 and 2025. It trailed holding by 15.7 points in 2021 and 14.5 points in 2023.
What happens if the entry threshold changes?
Entering above 0% returned 8.70% with a 26.22% drawdown and 22 round trips. Entering above 10% returned 6.69% with a 12.47% drawdown, and above 15% returned 5.94% with a 7.99% drawdown. None matched holding's 14.56%.
Did the strategy avoid the 2022 bear market?
Only in part. It lost 16.9% in 2022 against 17.8% for holding. It was flat from February to July, but it lost 4.51% and 7.59% on two re-entries in August and September.
How much do trading costs matter here?
Very little. At 10 basis points per fill the CAGR was 7.91% against 8.21%, because the rule made only 19 fills.
How does the same rule do on other funds?
It returned 8.80% on [VOOG](/learn/strategies/quarterly-momentum/voog/) and 8.25% on [VOO](/learn/strategies/quarterly-momentum/voo/), and it lost money on IWM, EEM and QQQE. The median across 59 ETFs was 0%.
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.