DeployQuantGolden Cross (SMA 50/200) 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.
Golden cross on SPY is a two-line rule with very few trades. It buys when the 50-day average closes above the 200-day average and sells when the 50-day falls back below. From 2021-01-04 to 2026-10-02 it turned $10,000 into $16,150, a CAGR of 8.71%. Buy-and-hold of the same fund made 14.56% and ended at $21,820. The rule made 2 closed round trips, 1 of them a winner, and held a third position at the end of the data.
What it gave up in return it recovered partly in risk. The maximum drawdown was 18.13%, against 23.99% for holding. The Sharpe ratio was 0.77 against 0.95. Over the six calendar years it beat holding once, in 2022, when it lost 8.4% and the fund lost 17.8%.
That outcome comes almost entirely from three trades. One lost 3.04%, one gained 34.93% over 811 days, and one was still open with a gain of 26.4%. A strategy that trades this rarely has a result that depends on when its few signals arrive. On SPY the rule ranked sixth of the 12 templates, and SPY was seventeenth of 59 funds for golden cross. The test is one window of 5.7 years with no fees in the headline run, so the result describes this stretch of SPY.
Year by year
| Year | golden cross | buy & hold |
|---|---|---|
| 2021 | 6.0% | 29.6% |
| 2022 | −8.4% | −17.8% |
| 2023 | 18.6% | 25.5% |
| 2024 | 23.9% | 24.3% |
| 2025 | 0.2% | 17.4% |
| 2026 | 13.0% | 13.6% |
How each calendar year played out
Golden cross gained 6% in 2021 against 29.6% for holding. That was the widest gap of any year, 23.6 points, and it came from timing. The rule held no position until 2021-10-19, so it missed everything before that date. January to September 2021 show 0.0% in every month of the table. The fund was up 29.6% for the year, and the rule captured the last quarter only, with gains of 2.26% in October and 4.39% in December and a loss of 0.71% in November.
2022 was the year the rule did what it exists to do. It lost 8.4% against 17.8%, a lead of 9.4 points. The whole loss came in the first three months, minus 5.11% in January, minus 2.82% in February and minus 0.68% in March. The sell came on 2022-03-17 and the template then held cash from April to December, which the month table shows as nine months of 0.0%. Holding fell further and bottomed on 2022-10-12 with a drawdown of 23.99%.
2023 returned 18.6% against 25.5%. The rule bought on 2023-01-27 and then held through every month and gained in 2023 despite a minus 4.54% September and a minus 2.05% October. The best month was November at 8.69%.
2024 was nearly a tie, 23.9% against 24.3%. The rule was invested the whole year, and it buys with 98% of the sleeve, which may account for part of the 0.4 point gap.
2025 returned 0.2% against 17.4%, a gap of 17.2 points and the second widest. The sell came on 2025-04-17, after the low of 2025-04-08 and the best day of the data on 2025-04-09. The rule re-entered on 2025-06-30. January gained 2.61% and February lost 1.24%, then March lost 5.42% and April lost 5.33% while the rule was still invested, and May shows 0.0%. From July onward the months are all gains. The next section looks at why that sequence cost so much.
2026 so far shows 13% against 13.6%, close to the fund again because the rule was invested the whole year. April 2026 was the best month at 9.91%, and March the worst at minus 4.72%.
Month by month
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2021 | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 2.3% | −0.7% | 4.4% |
| 2022 | −5.1% | −2.8% | −0.7% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% |
| 2023 | 0.6% | −2.4% | 3.5% | 1.5% | 0.4% | 6.2% | 3.1% | −1.6% | −4.5% | −2.0% | 8.7% | 4.4% |
| 2024 | 1.5% | 5.0% | 3.2% | −3.9% | 4.8% | 3.4% | 1.1% | 2.3% | 2.1% | −0.9% | 5.8% | −2.3% |
| 2025 | 2.6% | −1.2% | −5.4% | −5.3% | 0.0% | 0.1% | 2.1% | 1.9% | 3.4% | 2.3% | 0.2% | 0.1% |
| 2026 | 1.4% | −0.8% | −4.7% | 9.9% | 5.0% | −1.0% | 0.1% | 2.6% | −0.3% | 0.9% | – | – |
What the month table shows
The month-by-month grid reads as a record of when the rule was in the market. The zeros are periods in cash and not months that happened to be flat: nine months at the start of 2021, nine from April to December 2022, and May 2025.
The worst month was March 2025 at minus 5.42%, the worst for holding was September 2022 at minus 9%. The rule's worst months, January 2022 and March 2025, both came while it was invested at the start of a decline, which is the feature of a slow signal. A 50-day average crossing a 200-day average needs weeks of damage before it flips. In January 2022 the damage was still ahead. In March and April 2025 the rule took two consecutive losses of 5.42% and 5.33% before the sell, and the sell came after the low.
The best months were April 2026 at 9.91%, November 2023 at 8.69%, June 2023 at 6.2%, November 2024 at 5.76% and May 2026 at 5.02%.
The invested stretch from 2023 to 2026
After the January 2023 entry the rule held SPY through every later month in the table, so the monthly grid from that point is close to the fund's own record. In 2024 there were three losing months: April at -3.88%, October at -0.92% and December at -2.33%. The best months were November at 5.76%, February at 4.99% and May at 4.85%. The sleeve buys with 98% of the capital, so each of these months runs slightly under the fund.
The second half of 2025 was quiet. July made 2.15%, August 1.95%, September 3.35% and October 2.25%, and November and December added only 0.19% and 0.06%. June 2025 shows 0.1% because the re-entry came on 2025-06-30, the last trading days of the month. In 2026 the losing months were February at -0.77%, March at -4.72% and June at -1.03%, and April and May made 9.91% and 5.02%. The rule's drawdown inside this long hold was the 9.54% episode of 2023 and, at the 2025 peak, the 18.13% decline. Between them it made no trades, so every one of these months is a result of the 2023-01-27 and 2025-06-30 decisions and nothing newer.
Every trade
golden cross on SPY made 2 closed round trips and one position still open at the end of the test, an average hold of 480 days, an average winner of 34.93%, an average loser of −3.04%, a profit factor of 11.00, a longest losing streak of 1. It held a position at the close on 67.7% of trading days.
| Entry | Entry price | Exit | Exit price | Return | Days held |
|---|---|---|---|---|---|
| 2021-10-19 | $419.68 | 2022-03-17 | $406.92 | −3.0% | 149 |
| 2023-01-27 | $385.13 | 2025-04-17 | $519.67 | 34.9% | 811 |
| 2025-06-30 | $608.92 | open | – | 26.4% | – |
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 three trades
The first trade began on 2021-10-19 at an adjusted price of 419.68 and ended on 2022-03-17 at 406.92, a loss of 3.04% over 149 days. It is the only losing trade. The fund's decline began in January 2022, and the rule held through the first months of it.
The second trade began on 2023-01-27 at 385.13 and ended on 2025-04-17 at 519.67, a gain of 34.93% over 811 days. That one trade is most of the result, and it is the reason the profit factor is 11.00 and the average winner is 34.93%. The profit factor describes one winner against one loser and carries no weight as a statistic. The longest hold was also this trade, and the median hold of 811 days is the same trade.
The third trade began on 2025-06-30 at 608.92 and was open at the end with a gain of 26.4%. The exit on 2025-04-17 was at 519.67, so the rule paid 608.92 to get back in after selling at 519.67. That round trip across the April 2025 low is the reason 2025 returned 0.2%. A rule that waits for a 50-day average to cross a 200-day average will sell after the low and buy after the recovery, and SPY's recovery from 2025-04-08 was fast: holding was back to its old high on 2025-06-26.
Overall the rule held a position on 67.7% of sessions. The average hold was 480 days, which includes one trade that lasted more than two years, and the shortest was 149 days. There were 5 fills over the whole window.
Largest drawdowns
| Peak | Low point | Depth | Days to low | Recovered | Days to recover |
|---|---|---|---|---|---|
| 2025-02-19 | 2025-04-08 | −18.1% | 48 | 2026-04-22 | 379 |
| 2022-01-03 | 2023-03-13 | −12.9% | 434 | 2023-06-30 | 109 |
| 2023-07-31 | 2023-10-27 | −9.5% | 88 | 2023-11-30 | 34 |
Buy-and-hold's deepest drawdown ran from 2022-01-03 to 2022-10-12 and reached −24.0%.
Drawdowns
The largest drawdown was 18.13%, from 2025-02-19 to 2025-04-08, and it took 379 days to recover, to 2026-04-22. Holding fell 18.41% over the same peak and trough dates and recovered by 2025-06-26, in 79 days. The rule lost nearly the same amount as the fund in that episode but stayed down for a year, because it sold nine days after the low and bought back higher. This is the one drawdown in which the rule did worse than holding in time, though the depth was nearly the same.
The second was 12.92%, from 2022-01-03 to 2023-03-13, with 434 days to the low and a recovery by 2023-06-30. The peak fell while the rule held its first position. January to March 2022 lost 5.11%, 2.82% and 0.68%, equity was flat in cash for nine months, and the low on 2023-03-13 came after a loss of 2.4% in February 2023, the month after the re-entry. The third was 9.54%, from 2023-07-31 to 2023-10-27, and it recovered in 34 days. Holding had a drawdown of 23.99% from 2022-01-03 to 2022-10-12 that the rule mostly avoided, and it was the main reason the rule's maximum was 5.86 points shallower.
The trade-off is the usual one for a slow trend filter. It sidesteps the long, deep decline and then pays on the sharp one. SPY's 2022 decline was nine months long, which gave a 50/200 cross time to act. The April 2025 decline lasted 48 days from peak to trough, which was too short.
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) | 8.7% | −18.1% | $16,150 | 0.77 |
| 5 basis points | 8.7% | −18.1% | $16,124 | 0.77 |
| 10 basis points | 8.7% | −18.2% | $16,097 | 0.76 |
What costs did
The cost runs change almost nothing. With 5 basis points per fill the CAGR was 8.68% and the final value $16,124. With 10 basis points the CAGR was 8.65% and the final value $16,097. The headline run ended at $16,150. The drawdown moved from 18.13% to 18.14% and 18.16%. Five fills cannot generate much cost, and that is the argument for a slow rule: its result does not depend on execution quality. A template that makes hundreds of trades on SPY would see a larger difference.
Risk-adjusted effect of costs
The Sharpe ratio moves from 0.77 in the headline run to 0.765 at 5 basis points and 0.762 at 10 basis points. The drawdown stays within a few hundredths of a point of 18.13% in all three runs. A rule with 5 fills pays the cost five times, and the total is a small fraction of the ending account value. The cost runs therefore say little about SPY's execution, and a slippage assumption of 10 basis points on a fund with $30,453,859,411 of average daily dollar volume is generous.
Changing the parameters
| Version | CAGR | Max drawdown | Round trips | Win rate | Final value |
|---|---|---|---|---|---|
| Published rules | 8.7% | −18.1% | 2 | 50% | $16,150 |
| SMA 40/200 | 8.6% | −18.2% | 2 | 50% | $16,052 |
| SMA 50/150 | 9.4% | −17.2% | 4 | 75% | $16,785 |
| SMA 60/250 | 8.8% | −18.4% | 2 | 50% | $16,236 |
Changing the averages
Three variants were tested. SMA 40/200 returned 8.59% with a drawdown of 18.16% and 2 trades, 1 win, and ended at $16,052. SMA 60/250 returned 8.81% with a drawdown of 18.41%, 2 trades and $16,236. Both are close to the published rules at 8.71%.
SMA 50/150, which shortens the slow average, did better: 9.44% with a drawdown of 17.16%, a Sharpe ratio of 0.85 and 4 trades with 3 wins, ending at $16,785. The variant is better on return, drawdown and Sharpe together, but it rests on 4 trades in one window, and the other two variants are within a few tenths of the published rules on all three measures.
The pattern across the table is that the choice of averages matters less than the fact of being out in 2022 and invested in 2023 and 2024. Every variant made 2 or 4 trades.
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 behaved
SPY's annualized volatility was 16.41% and its drawdown in its own series was 24.51% from 2022-01-03 to 2022-10-12, recovered on 2023-12-13. It was above its 200-day average on 77.99% of sessions and crossed it 30 times. The rule made 2 closed trades, not 30, because a 50-day average smooths price and crosses the 200-day average much less often than price does. Price crossed its 200-day average 30 times while the rule made only 5 fills, so it sat through the choppy stretches.
SPY rose on 54.4% of days with average up and down days of 0.74%. The average intraday range was 1.15% and the average overnight gap was 0.44%, and 62.46% of the log return came overnight. The fund fell 10% or more from a 20-day high 11 times across 23 days, so large falls were rare and short. The one that mattered for the rule was the 2022 decline, which was slow.
The best day was 2025-04-09 at 9.39% and the worst was 2025-04-04 at minus 5.98%. The rule was still invested on both days and sold on 2025-04-17.
RSI(14) fell under 30 on 18 sessions, with a median forward 5-day return of 3.74% against 0.42% for all days. RSI(2) fell under 10 on 146 sessions, with 0.89% over 5 days and 2.29% over 20. Those are dip-buying numbers and do not bear on a trend rule, but they explain why the RSI(2) template, at 13.17% and a drawdown of 12.69%, ranked second on SPY while golden cross ranked sixth.
Seasonality, correlations and the SPY twins
SPY's average return by calendar month shows the same pattern as most index funds in this window. November averaged 4.01%, July 3.07%, May 3% and October 2.88%, while September averaged -2.21%. Each month has five or six observations. The rule sat in cash for most of the early part of the window, so the seasonal weakness of September only touched it in 2023 and later: September 2023 lost 4.54%, September 2024 made 2.09%, September 2025 made 3.35% and September 2026 lost 0.32%. The mixed record shows the average does not predict a given year.
SPY correlates 0.94 with QQQ, with a beta of 0.69 to it, and 0.08 with TLT. The most correlated funds in the tested set were VOO and VV at 1 and SSO at 1. That explains why golden cross on VOO returned 8.77% and the rule on VV returned 8.43%, both close to the 8.71% on SPY, with drawdowns of 18.06% and 18.62%. The three funds hold nearly the same stocks, so the same two crossovers fire within a day or two of each other and the results differ by small timing and weighting effects.
The lag-1 autocorrelation of daily returns was -0.02 and the weekday averages ranged from 0.01% on Tuesday to 0.14% on Monday. A trend rule has no short-term pattern to use here, and it relies on the slower persistence of multi-month moves such as 2023 and 2024. Whether that persistence continues is outside what this window can show.
The rules
Hold while the 50-day average is above the 200-day; step aside when it crosses below (the death cross).
- WHEN the market opens · IF not invested AND SMA(50) > SMA(200) · THEN buy with 98% of the sleeve
- WHEN the market opens · IF invested AND SMA(50) < SMA(200) · THEN sell the whole position
A widely used trend filter. When the 50-day simple moving average is above the 200-day, the asset is in a long-term uptrend and the strategy holds. When it crosses below, the strategy moves to cash. It trades rarely, with a handful of signals per decade on an index. Its use is skipping the deepest bear markets, and it will lag some rallies.
Good for: long-horizon investors who want to hold trends but sidestep multi-year bear markets.
Watch out: crosses lag at turning points: the strategy gives back the first leg of a crash and misses the first leg of a recovery, and choppy sideways markets whipsaw it.
Golden cross against the other templates on SPY
Weekly 7% target led SPY at 13.88% with a drawdown of 24.1%. RSI(2) snapback followed at 13.17%, then monthly cycle at 12.84% and EMA 12/26 trend at 10.2% with a drawdown of only 11.84%. Momentum breakout made 9.09%, and golden cross made 8.71% in sixth place. Behind it were the 200-day regime filter at 8.3%, 3-month momentum at 8.21% and SMA 10/50 trend at 7.94%. The trend plus trailing stop template made 5.45% with a 31.15% drawdown, RSI mean reversion 4.37% and the dip buyer 2.61%.
The faster trend rules did better on SPY. EMA 12/26 made 10.2% and SMA 10/50 made 7.94% with a drawdown of 12.68%, and both had shallower drawdowns than golden cross at 18.13%. A different window would change the order.
Across other broad index funds the same rule returned 13.82% on QQQM, 13.6% on QQQ, 12.39% on IOO, 8.77% on VOO, 8.43% on VV and 1.05% on IWM, which made 4 round trips and had a 31.01% drawdown. The median for golden cross across all 59 funds was 2.05%, and the median for the broad index category was 8.71%, which is where SPY sits. The full table is on the golden cross page and every template on the fund is on the SPY page.
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 golden cross beat buy-and-hold on SPY?
Over 2021-01-04 to 2026-10-02, golden cross on SPY returned 8.7% annualized vs 14.6% for buy-and-hold: it trailed buy-and-hold by 5.8% per year, with a maximum drawdown 5.9 points shallower than holding (18.1% vs 24.0%).
How many trades did it make?
2 completed round trips over 5.7 years (5 fills), with 50% of round trips closing profitably.
How often does a golden cross happen?
On a broad index, roughly every couple of years. The backtest pages show the exact trade count for each ETF over the 2021 to 2026 window.
Golden cross vs buy and hold: which does better?
It depends on the asset and the window. Every backtest page here shows the same-window buy-and-hold comparison.
How did golden cross do on SPY from 2021 to 2026?
It returned 8.71% a year and turned $10,000 into $16,150. Buy-and-hold made 14.56% and ended at $21,820. The maximum drawdown was 18.13% for the rule and 23.99% for holding.
How many trades did golden cross make on SPY?
It closed 2 round trips and held a third position at the end of the data. The first, from 2021-10-19 to 2022-03-17, lost 3.04%. The second, from 2023-01-27 to 2025-04-17, gained 34.93% over 811 days.
Did golden cross avoid the 2022 decline on SPY?
Partly. It lost 8.4% in 2022 against 17.8% for holding, and all of its loss came in January to March before the sell on 2022-03-17. It held cash from April to December. It was the only calendar year in which it beat holding.
Why did golden cross trail buy-and-hold so much in 2025?
It was invested for the March and April 2025 losses, sold on 2025-04-17 at 519.67 after the 2025-04-08 low, and bought back on 2025-06-30 at 608.92. The year returned 0.2% against 17.4% for holding.
Does a different moving average pair help on SPY?
SMA 50/150 did slightly better, at 9.44% with a 17.16% drawdown and 3 wins in 4 trades. SMA 40/200 gave 8.59% and SMA 60/250 gave 8.81%, so the published 8.71% sits in the middle of the range.
Do trading costs matter for golden cross on SPY?
No. Slippage of 10 basis points per fill cut the CAGR from 8.71% to 8.65% and the final value from $16,150 to $16,097. The rule made only 5 fills.
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.