Crypto signal win rates: we back-tested 3,463 signal occurrences
Golden crosses, death crosses, RSI extremes, MACD crossovers — and the chart patterns our screener detects — re-run mechanically across 47 coins and 5.0 years of daily history. Every single occurrence graded by what price did next. No cherry-picking, no survivor stories — and the numbers update automatically as new history arrives.
Last regenerated Jun 2026 · window Jun 2021 → May 2026
Key findings
- 1
The strongest signal by forward return was the golden cross: an average move of +12.7% in its favour over the following 30 days across 80 occurrences — versus +2.8% for simply holding through a random 30-day window.
- 2
The most consistent signal was the macd bearish cross, moving in its favour 54% of the time at 30 days. For context, a random 30-day window drifted upward 47% of the time — crypto's long-run upward bias is the bar every bullish signal has to clear.
- 3
Regime is the hidden variable: an average coin's random 30-day return was +5.5% while Bitcoin traded above its 200-day average (bull regime) but -1.1% below it (bear regime). The same signal can look brilliant or broken depending purely on the backdrop it fired in.
- 4
Bullish signals as a group hit 50% at 30 days while bearish signals hit 52% — bearish signals fight crypto's structural upward drift, which is why we grade them against a falling-price definition of "win" and still report the asymmetry.
- 5
Among chart patterns, the head and shoulders was the standout: 67% of 434 detected occurrences moved in its favour within 10 days, averaging +3.5% per occurrence.
Win rate by signal — all history
Share of occurrences where price moved in the signal's favour within 30 days (up for bullish signals, down for bearish). The dashed line marks how often a random 30-day window simply drifted up — the honest benchmark for the bullish signals.
The regime split — where edges live and die
The same signal, graded separately depending on whether Bitcoin closed above (bull) or below (bear) its 200-day average on the day it fired. This is the single most important table on this page.
| Signal | Bull: win 30d | Bull: avg 30d | Bear: win 30d | Bear: avg 30d |
|---|---|---|---|---|
| 49% | +8.7% | 49% | +1.0% | |
| 56% | -4.9% | 51% | -0.7% | |
| 48% | -11.9% | 56% | -1.6% | |
| 61% | +18.3% | 51% | +0.1% | |
| 53% | -0.2% | 53% | -1.4% | |
| 55% | +13.5% | 45% | +7.8% |
Baseline 30-day drift: +5.5% in bull regimes vs -1.1% in bear regimes. A bullish signal only has a real edge if it beats that drift within the same regime.
Chart patterns — the numbers nobody publishes
We ran our live pattern detectors across the same history: 11,723 detected occurrences across 28 coins, each graded 10 days forward. Pattern folklore claims 80%+ reliability; here is what a mechanical detector actually finds.
| Pattern | Occurrences | Win rate (10d) | Avg move |
|---|---|---|---|
| 11,279 | 49% | +1.6% | |
| 434 | 67% | +3.5% | |
| 10 | 10% | -6.9% |
Full diagrams, target calculations, and live detection for every pattern in the pattern library.
How this study works (and what it doesn't claim)
Every signal is fired only on its transition bar — the exact day RSI(14) crossed into oversold, the day the 50-day SMA crossed the 200-day, the day the MACD line crossed its signal line — then graded by the close 7 and 30 daily bars later. Patterns are detected by the same pure-JS detectors the live screener runs, deduplicated by completion bar, and graded 10 bars forward. “Win” always means price moved in the signal's stated direction.
The universe is the top 47 coins by market cap with usable daily history, which introduces a survivorship bias we want you to know about: coins that collapsed out of the top ranks aren't in the sample. No fees, slippage, or position sizing are modelled — this measures the signals' statistical tendency, not a strategy's P&L. Past performance does not predict future results.
The full mechanical rules are documented on the methodology page, and the always-current numbers live on the track record. Cite this study freely with a link — the data regenerates daily, so the numbers stay honest.
Trade the signals with the stats attached
The screener shows these exact signals live on every top-100 coin — with this back-tested record a click away, so you always know the base rate behind the badge.
Open the live screener