The golden cross, the death cross, reclaiming the 50 or the 200 day line: moving-average crosses are the most-watched patterns in technical analysis. Tested across decades, almost none of them beat an ordinary day, and the single exception is easy to misread.
Few things on a chart get more attention than a moving-average cross. Price reclaiming its 50-day line, its 200-day line, the 50 crossing above the 200 (the golden cross) or falling below it (the death cross): each is treated as a change of regime, a moment to get long or step aside. We wanted to know whether crossing a moving average actually predicts anything, or whether it only describes where price has already been.
We ran the tests on SPX, the S&P 500 index; QQQ, the Nasdaq 100; SMH, the semiconductors; and two heavyweight single names, NVDA and META. For each one we flagged the day price closed back above its moving averages, measured the close-to-close return over the following days, weeks, and months, and compared it to what the same instrument did after an ordinary bar.
The tidiest bullish picture a chartist can ask for is price vaulting above all of its moving averages at once, the 5, 10, 20, 50, and 200 day lines together. It turns out that almost never happens. To clear the entire stack in one session, price has to leap from below the fastest line all the way above the slowest, a move so large it only occurs on a violent reversal.
| Lines cleared in one day | SPX | QQQ | SMH |
|---|---|---|---|
| 5 and 10 | 630 | 347 | 337 |
| 5, 10, 20 | 222 | 124 | 132 |
| 5, 10, 20, 50 | 53 | 16 | 36 |
| all five (through the 200) | 1 | 0 | 5 |
Clearing just the two fastest lines is common enough, a few hundred times across the decades. Add the 20, then the 50, and the count collapses. Clearing all five in one day has happened once on the S&P in 56 years, never on the Nasdaq, and five times on the more volatile semis. And when we measured what followed the common two- and three-line crosses, the forward returns were indistinguishable from an average day. The fast crosses describe a bounce that already happened. They do not forecast the next one.
One cross does beat an ordinary day by enough to matter: the day price closes back above its 200-day moving average after having fallen below it. The 200-day is the line that separates a correction from a trend change, and reclaiming it clears an ordinary month on the S&P and, dramatically, on NVDA. On the Nasdaq and the semis it does the opposite.
| Pattern | Next month (avg / % up) | Next quarter (avg / % up) |
|---|---|---|
| SPX baseline | +0.78% / 62% | +2.31% / 67% |
| 200-day reclaim (186) | +1.13% / 67% | +3.25% / 69% |
| QQQ baseline | +0.89% / 62% | +2.55% / 67% |
| 200-day reclaim (86) | +0.65% / 63% | +0.81% / 61% |
| SMH baseline | +1.48% / 60% | +4.23% / 66% |
| 200-day reclaim (107) | +0.64% / 52% | +2.09% / 54% |
| NVDA baseline | +3.92% / 59% | +12.42% / 67% |
| 200-day reclaim (60) | +10.60% / 72% | +23.32% / 75% |
| META baseline | +2.50% / 63% | +8.12% / 69% |
| 200-day reclaim (34) | +4.64% / 71% | +8.38% / 67% |
Read it by comparing each reclaim row to the muted baseline above it. On the S&P the reclaim does help: 67% green over the next month against a 62% baseline, and it holds through the quarter. On QQQ and the semis it is a drag, running below baseline on both. The reason is the baseline itself. A 200-day reclaim only happens after a real decline, and coming out of a decline the high-flying funds have to clear an enormous ordinary drift (the semis average more than +4% a quarter after any bar) that a recovery cannot keep up with. On the steadier index, where the baseline is a gentler +2% a quarter, the recovery clears it comfortably. The 50-day reclaim, for what it is worth, does nothing on any of them.
The number that jumps off the table is NVDA: a 200-day reclaim followed by +10.6% over the next month against a +3.9% baseline, and +23% over the quarter. That is a real within-NVDA result, measured against NVDA’s own already-huge drift. But it rests on 60 occurrences on the single best-performing megacap of the era. A stock that went on to multiply many times over will make almost any “it recovered” marker look prophetic, because on that stock recovering and continuing were the same event. META, another big winner, shows a milder version of the same result on 34 cases. The diversified fund that holds names like them, SMH, shows none at all. Pick a survivor and the 200-day reclaim looks like a machine. Pick the basket and it is gone.
This is the quiet problem with every backtest run on a handful of winners. What you are measuring then is not the moving-average cross so much as the fact that you already know which names kept going up. Run the same rule across a broad, unbiased set of stocks, winners and losers together, and the spectacular single-name result shrinks toward the modest index one.
A moving-average cross tells you that price and its own average just changed places. For almost every version of it, that is the whole of the message: the forward returns look like an ordinary day. The one cross with a real advantage, reclaiming the 200-day line, works on the broad index, inverts on diversified funds, and on the single names where it looks strongest is mostly the survivor’s trend showing through. The cross marks where price has been. It does not tell you where price goes next.
Testing popular ideas without cherry-picking
Browse Reality Check
Comments
Loading comments…