What it is
RSI mean reversion buys "oversold" and sells "overbought". RSI(14) measures the ratio of recent up-closes to down-closes on a 0 to 100 scale; readings below 30 mean sellers have dominated the last two weeks of bars, above 70 means buyers have. The method bets that such stretches snap back. It is the single most common beginner strategy in crypto, and on BTC and ETH daily charts it is also one of the worst, because trends in crypto routinely hold RSI above 70 for weeks.
We publish it with full numbers because the losing rows are the lesson.
Rules
- Market / timeframe: BTC or ETH, 1-hour, 4-hour or daily candles.
- Entry long: RSI(14) crosses below 30. Entry short: RSI(14) crosses above 70. Enter at the close of the crossing bar.
- Exit: RSI returns through 50, or 10 bars elapsed, whichever comes first. No price stop, which is how the method is usually taught.
- Position size: the table uses 100% of equity, unleveraged, so the worst-trade column shows exactly what no stop does.
Chart example
Evidence
Our own backtest on Coinbase candles. Daily rows cover 2023-10-20 to 2026-10-19 (three years). 4h and 1h rows cover 2025-10-19 to 2026-10-19 (one year). No fees, no slippage, unleveraged.
| Market / TF | Trades | Win rate | Avg trade | Profit factor | Worst single trade | Total return (1x) | Max DD | Buy & hold |
|---|---|---|---|---|---|---|---|---|
| BTC 1D | 21 | 38.1% | −3.0% | 0.36 | −18.5% | −51.1% | 53.4% | 178.0% |
| BTC 4h | 44 | 47.7% | −0.5% | 0.68 | −10.2% | −20.5% | 23.6% | −24.5% |
| BTC 1h | 200 | 51.5% | 0.1% | 1.16 | −8.1% | 15.6% | 21.1% | −23.9% |
| ETH 1D | 22 | 40.9% | −3.7% | 0.29 | −21.5% | −60.1% | 61.9% | 51.1% |
| ETH 4h | 46 | 58.7% | 0.2% | 1.16 | −12.1% | 7.7% | 24.1% | −39.4% |
| ETH 1h | 194 | 56.7% | 0.0% | 1.05 | −13.9% | 1.5% | 27.0% | −39.2% |
Split by side, daily chart: BTC longs (RSI below 30) 8 trades, 50% winners, −0.2% average; BTC shorts (RSI above 70) 13 trades, 31% winners, −4.8% average. ETH shorts averaged −3.4%, ETH longs −4.2%.
How to read it: on daily candles the rule lost half the account in three years, mostly by shorting a market that kept going up after RSI crossed 70. On hourly candles the win rate rises above 50% but the average trade is zero; the rule is a coin flip with a 14% worst case. The higher win rate that attracts people to mean reversion is real, and it is paid for by the occasional trade that keeps going against you.
When it fails
- Trends. RSI above 70 is what a bull market looks like. Shorting it is shorting the trend. The 2024 BTC run held daily RSI above 70 for most of February and March.
- No stop. The method's worst single trades (−18% and −21% on the daily rows) are what happens when "it has to come back" meets a market that does not. At 5x leverage those are liquidations.
- Low timeframes with fees. 200 trades a year at a 0.1% average edge is wiped out by a 0.05% taker fee each way.
- It is a filter, not a system. RSI extremes work better as a reason not to chase (do not buy a 15-minute RSI of 85) than as a reason to fade.
RektChat's chart has RSI built in. Try fading 70 on a paper account through one real trend and you will not need this table to remember the lesson.