Do moving average crossovers actually work?
Some do, some do not, and the deciding factor is not the indicator, it is how often the cross fires. We walk-forward tested 9 moving average crossover strategies across 28 tickers and ran each through a 100-run random-entry gauntlet. Eight held a real edge on at least one ticker. The most famous one, the 50/200 Golden Cross, held an edge on none.
Based on the same process behind every Library report. How we test.
What counts as a moving average crossover
Two kinds show up in this test. The first is a cross of two moving averages, a faster one crossing a slower one, like the 20-day crossing the 50-day, or the famous 50/200 Golden Cross. The second is price crossing a single moving average, like price reclaiming its 50-day line. We tested both. Every version also required price to be above its 200-day average, so these are trend-following entries, not counter-trend bets.
The crossovers that held a broad edge
| Setup | Ticker | Return | Sharpe | Win rate | Edge held on |
|---|---|---|---|---|---|
| SMA 20/50 cross | SPY | +103.4% | 0.83 | 61.9% | 9 / 28 |
| EMA 8/21 trend rider | SPY | +52.2% | 0.75 | 44.7% | 9 / 28 |
| EMA 20/50 trend rider | QQQ | +110.6% | 0.91 | 51.9% | 7 / 28 |
| MACD cross above 200 SMA | AAPL | +304.3% | 1.36 | 49.2% | 5 / 28 |
These held up on five or more of the 28 tickers, so the edge is not resting on one lucky name. Return shown is a representative index or large-cap, not the best single ticker. Breadth is the number to trust.
The crossovers with only a narrow edge
| Setup | Ticker | Return | Sharpe | Win rate | Edge held on |
|---|---|---|---|---|---|
| 50-day moving average cross | GLD | +146.2% | 0.91 | 26.9% | 4 / 28 |
| Price / VWMA cross | AAPL | +395.7% | 1.30 | 40.5% | 4 / 28 |
| Hull MA cross | MSTR | +821.9% | 0.87 | 38.2% | 2 / 28 |
| Weinstein Stage 2 breakout | XOM | +65.9% | 0.44 | 25.5% | 2 / 28 |
These cleared the bar on just one to four tickers, usually carried by a single powerful trend. The returns look enormous because the best ticker was a volatile mover, so read them as 'worked here, not everywhere.' Return shown is the best ticker.
The famous one that failed: the Golden Cross
The 50/200 SMA cross is the most preached signal in all of trading, and it is the clearest example of why a good-looking backtest can lie. On its best ticker, QQQ, it beat 99 of 100 random-entry runs, which sounds airtight. Then you count the trades: four in a decade. A signal that fires four times gives you no way to separate skill from luck, so a 99 out of 100 score is really a coin flip that landed heads. Across all 28 tickers it cleared our bar on exactly zero. The problem is not the idea of a trend filter, it is the speed: the 200-day average moves so slowly that the cross almost never happens.
The pattern: how often it fires decides everything
Line the survivors up against the bust and one variable explains the whole table: trade frequency. The 8/21 and 20/50 crosses fire often enough over a decade to build a sample you can trust. The 50/200 barely trades. Faster averages react sooner, so they generate more signals and a measurable edge, at the cost of more whipsaws. Slower averages whipsaw less but fire so rarely you can never prove they work. The takeaway is blunt: when you judge any crossover, look at the trade count before you look at the return.
How to use this
- Prefer faster crossovers like the 20/50 or 8/21 over the 50/200 if you want enough trades to trust the result.
- Check the trade count first. Under about 30 trades, treat any backtest as a hint, not evidence.
- Watch breadth. A setup that only works on one hot ticker is fragile; the 20/50 held up on 9 of the 28 tickers we ran.
- A high return with a low win rate means a few big trends carried it. That can be fine if you can stomach the drawdowns, but it is not a broad, reliable edge.
- Whatever crossover you use, backtest your exact version on your own tickers before trusting it. Small parameter changes matter.
Common questions
Do moving average crossovers actually work?
Some do. Faster crosses like the 20/50 and 8/21 that trade regularly held a real, walk-forward-tested edge across multiple tickers. The slow 50/200 Golden Cross did not, because it fires too rarely to trust the result.
Does the Golden Cross work?
Not in our test. The 50/200 cross beat 99 of 100 random entries on its best ticker but only traded four times in a decade, and it cleared our bar on zero of 28 tickers. A near-perfect score on four trades is statistical noise.
What is the best moving average crossover?
By breadth, the 20/50 SMA cross and the 8/21 EMA cross held an edge on the most tickers (9 of 28 each), which makes them more trustworthy than variants that only worked on one volatile name.
Why does a faster crossover beat the Golden Cross?
Sample size. Faster averages cross more often, so they generate enough trades to measure a real edge. The 50/200 Golden Cross fires so rarely that even a great score is based on a handful of trades, which is a coin flip.
Are moving average crossovers profitable?
Several were in our backtest, especially with a 200-day trend filter, but a positive backtest is not a guarantee. Costs, slippage, and a low win rate on trend-following entries all matter, so test your own version before you trade it.
We take each setup's rules as commonly stated. Your parameters may differ, so run your own on your own account and watch the walk-forward result before you trust it.