Best RSI Settings Crypto
The best-tested RSI settings for crypto over 2024-08-15 → 2026-07-05: RSI lookback (bars) 7 · RSI exit / take-profit level 50, at a Sharpe of 1.05 net of fees, slippage and funding. We backtested 4 combinations that actually traded on the same coins the strategy trades — the full list, including the settings that lost money, is below.
RSI settings — the tested grid
precomputed · updates instantlyEvery setting we backtested, ranked. Pick one to load it into the editor; the settings that lost money stay visible too.
Tested settings, not every combination. We tried 8 combinations, but only 4 actually placed trades — the other 4 never triggered an entry, so they're left out. What's shown is every setting that produced real trades.
At $1,000, this setting's worst historical dip was about $96. You'll fork RSI with these settings applied — same engine, your own keys.
Compare all 8 settings
| Settings | Sharpe | Return | Worst DD | Trades |
|---|---|---|---|---|
| RSI lookback (bars) 7 · RSI exit / take-profit level 45 | 1.15 | 25.0% | −7.3% | 80 |
| RSI lookback (bars) 7 · RSI exit / take-profit level 50default | 1.05 | 29.4% | −9.6% | 107 |
| RSI lookback (bars) 7 · RSI exit / take-profit level 55 | 0.89 | 28.9% | −12.9% | 160 |
| RSI lookback (bars) 7 · RSI exit / take-profit level 65 | 0.17 | 2.5% | −26.1% | 265 |
| RSI lookback (bars) 14 · RSI exit / take-profit level 45 | — | 0.0% | −0.0% | 0 |
| RSI lookback (bars) 14 · RSI exit / take-profit level 50 | — | 0.0% | −0.0% | 0 |
| RSI lookback (bars) 14 · RSI exit / take-profit level 55 | — | 0.0% | −0.0% | 0 |
| RSI lookback (bars) 14 · RSI exit / take-profit level 65 | — | 0.0% | −0.0% | 0 |
How this was tested. Everything except the settings you can change was held constant — the same coins, the same costs, the same dates — so the default here lines up with the numbers on our main RSI page.
What this is: a real backtest of 8 RSI settings on Hyperliquid, net of fees, slippage and funding. We show only settings we actually tested and never imply an untested combination works. The default and every losing setting stay visible. Past performance does not predict future results; this is not investment advice.