Strategy Detail

RSI Mean Reversion

Buys oversold RSI dips and shorts overbought spikes, exits on a cross back through a neutral threshold. The textbook short-term mean reversion shape, parameterized.

What It Does

RSI Mean Reversion runs on a single contract at one timeframe. On every bar close it computes the current RSI value over the configured period and compares it to four thresholds: an oversold level, an overbought level, a long-exit level, and a short-exit level.

  • If the strategy is flat, the side is long-only or both, and RSI crosses down through the oversold level, it goes long at the close.
  • If the strategy is flat, the side is short-only or both, and RSI crosses up through the overbought level, it goes short at the close.
  • A long position exits when RSI crosses up through the long-exit level.
  • A short position exits when RSI crosses down through the short-exit level.

Crossover detection is symmetric. The strategy tracks the previous bar’s RSI and only fires when the line moves through the threshold; sitting below the threshold for multiple bars produces one entry, not a stream of them.

There are no stops, no profit targets, no trend filters, no position sizing beyond a fixed contract count. The RSI cross is the only exit. Holding period is whatever RSI takes to mean-revert.

Why It Works (Sometimes)

Mean reversion at short horizons is one of the better-documented edges in liquid equity indices. After a sharp move down, the next day’s return is on average slightly positive; after a sharp move up, the next day is on average slightly negative. Larry Connors codified this with RSI-2 in the mid-2000s, and the pattern has held up reasonably well in equity index futures since.

The edge is small, the variance is large, and the strategy can chain a string of winners before a single big trend move erases them. RSI mean reversion strategies tend to look great until they don’t, and the “don’t” usually coincides with a regime change to trending markets. The presets here are the textbook starting points; before relying on any of them, run them through a regime that matches what you actually trade.

Presets

Five presets ship out of the box, covering the canonical variants:

  • Connors’ RSI-2 Long / Short: the original short-term setup. Fires often (typically 10-30 trades per year on a daily contract), short holding periods, high win rate, ugly tail.
  • RSI-14 Oversold Long / Short: the textbook slower variant. Fires rarely, holds longer, less prone to whipsaw.
  • RSI-7 Two-Sided: a middle ground that takes both long and short entries off intermediate RSI extremes.

Use the form to set your own period and thresholds; the presets are starting points, not endpoints.

Best In

  • Mean-reverting regimes. Range-bound markets, post-shock reversion, post-event chop.
  • Liquid equity index futures (NQ, ES) where the mean-reversion edge is best documented.
  • Research where you want a clean, well-known baseline to compare a more sophisticated mean-reversion idea against.

Where It Struggles

  • Strong directional trends. RSI-2 in a bull market will short overbought spikes that keep going up; long-only variants avoid this side of the failure but still get smaller wins.
  • Single-sided regimes that last for months. The strategy has no regime filter; it will keep entering whether the setup is working or not.
  • Markets with persistent skew (commodities in a structural backwardation, for example). The two-sided preset can be especially exposed.

Possible Uses

  • A control strategy when evaluating other short-term reversion ideas: any setup whose returns track RSI-2 long is not adding much.
  • An honest stress test of the “mean reversion” thesis on whatever contract and timeframe you care about. The same parameter set can be sampled across regimes.
  • A starting point for layered strategies. RSI dips combined with a regime filter, or with a volatility filter, or with a position-sizing rule, often produce more usable backtests than the bare strategy does.

What It Does Not Do

  • No stops, no targets, no time exit. If RSI never crosses back through the exit threshold, the position is held until the strategy is shut down.
  • No volatility-adjusted thresholds. The same oversold value applies across calm and chaotic regimes.
  • No re-entry while in position. Once a long is open, additional RSI dips below oversold are ignored.

Test this strategy

Run it on your contracts, timeframes, and parameters.