The Relative Strength Index (RSI) is one of the most widely used oscillators for identifying overbought and oversold conditions. In mean reversion strategies, RSI helps time entries when prices have deviated significantly from their mean.
RSI Basics
Developed by J. Welles Wilder, RSI measures the speed and magnitude of recent price changes on a scale of 0 to 100.
Multi-Period Analysis
QuanterLab tests RSI across multiple periods to find the most effective settings for each security:
- RSI(2) - Very short-term, highly sensitive
- RSI(7) - Short-term balance
- RSI(14) - Traditional period, smoother signals
Different securities respond better to different periods based on their volatility and trading characteristics.
Win Rate Testing
Rather than using fixed thresholds (e.g., always buy when RSI < 30), the platform backtests various entry thresholds and measures historical win rates:
- RSI(2) entry at 10: What percentage of trades were profitable?
- RSI(7) entry at 20: What percentage of trades were profitable?
- RSI(14) entry at 30: What percentage of trades were profitable?
The threshold and period combination with the highest historical win rate is used for scoring and signals.
Entry and Exit
Typical mean reversion RSI usage:
- Entry - RSI falls below the optimal threshold (e.g., 10-30 depending on period)
- Exit - RSI rises above a higher threshold (e.g., 60-70)
An RSI reading of 25 might be attractive, but only if the security has already been identified as mean-reverting via Hurst Exponent and Variance Ratio. RSI alone does not indicate mean reversion—it only measures how extended the current move is.
Platform Scoring
RSI Win Rate receives significant weight (25% daily, 15% hourly) in the composite score. This reflects its importance as an entry timing mechanism once mean-reverting candidates are identified.
Combining with Other Indicators
RSI works best in conjunction with:
- Bollinger Bands - Price below lower band confirms RSI oversold reading
- Volume - Low relative volume on the decline suggests selling exhaustion
- ADX - Low ADX confirms ranging market suitable for mean reversion