Mean reversion trading. Fade the extremes.

Mean reversion is the opposite of trend following. Buy oversold expecting a bounce. Sell overbought expecting a pullback. The thesis is that prices tend to return to an average over time, and extreme deviations from that average are likely to reverse.

The strategy works in specific conditions and fails badly in others. The trader who applies mean reversion in trending markets catches falling knives and gets buried. The trader who applies it in range bound conditions captures consistent small wins.

What mean reversion actually is

The core idea is statistical. Over many observations, prices oscillate around an average. When price moves significantly above the average, it tends to revert lower. When it moves significantly below, it tends to revert higher.

The "mean" in mean reversion can be a moving average, a VWAP, a Bollinger Band midline, or any other measure of central tendency. The "reversion" is the expected snap back to that mean.

The strategy generates many small winning trades when the underlying is range bound. It generates large losing trades when the underlying is in a strong trend that does not revert.

The typical setup

Entry signal: extreme deviation from mean.

Common measures of extreme deviation include.

RSI(2) below 5 for long entries, above 95 for short entries. Larry Connors popularized this in The 2 Period RSI Stock Strategy.

Bollinger Band touches. Price touching the lower 2 standard deviation band signals oversold. Touching the upper band signals overbought.

Distance from the 200 day moving average. More than 10 percent below the 200 EMA on a normally stable stock often signals oversold.

Z-score calculations. Price two or more standard deviations from a rolling 20 or 50 day mean.

Entry execution.

Buy at the extreme deviation. Some traders wait for a reversal candle (hammer, bullish engulfing) at the extreme for additional confirmation. Others buy mechanically when the signal triggers.

Exit signal.

Return to the mean (the moving average, the VWAP, the Bollinger midline). Or RSI(2) crossing back to neutral (above 50 for long exits). Or a fixed time exit (close after 3 to 5 days regardless).

Stop loss.

Below the most recent swing low (for long entries). Or a fixed dollar percentage. Most mean reversion systems use wider stops than trend following because the trade is going against the immediate momentum.

The conditions where mean reversion works

Range bound markets.

Stock chopping between clear support and resistance for weeks or months. Mean reversion buys at the support edge and sells at the resistance edge. Each cycle produces a small win.

Intraday on liquid stocks.

Intraday moves often reverse, especially around clear technical levels. The opening gap that fades to fill. The afternoon pullback to VWAP. These are mean reversion setups even on stocks in larger trends.

Overextended stocks at extreme deviations.

A normally stable stock that has dropped 15 to 20 percent in a few days on no fundamental change often bounces sharply. The deeper the deviation from the recent mean, the higher the probability of a reversion bounce.

Post panic markets.

After capitulation declines (VIX above 40), the broader market often produces sharp mean reversion bounces over the following days and weeks.

The conditions where mean reversion fails

Strong trending markets.

The most overbought condition stays overbought as the trend continues. The mean reversion trader who shorts the overbought signal gets stopped out repeatedly as the stock keeps making new highs.

Stocks with fundamental news.

A stock that has dropped 20 percent on a guidance cut is not oversold technically. It has been repriced to reflect the new fundamentals. Mean reversion buying on the technical signal catches a falling knife.

Low volume conditions.

Mean reversion signals on thin volume often fail because the move that produced the extreme can extend further on continued thin volume.

During regime changes.

Markets going from low volatility to high volatility (or vice versa) often produce extended trends that overwhelm mean reversion systems. The new regime needs to settle before mean reversion can work again.

The filter that separates the conditions

The single biggest improvement to mean reversion systems is filtering by the higher timeframe trend.

Mean reversion only on stocks above the 200 day moving average for long entries. Only on stocks below the 200 day for short entries.

The 200 day filter cuts out the falling knife trades (stocks in serious downtrends that keep falling) and the parabolic top fades (stocks in strong uptrends that keep extending).

Adding this filter to a basic RSI(2) mean reversion system improves win rate and reduces tail losses meaningfully across most US equity universes.

The position sizing

Mean reversion trades have lower win rates than expected when they fail (the failures are often large). Position size has to account for the wider stops.

Risk 0.5 to 1 percent of account per trade. Standard discipline.

Avoid pyramiding (adding to losing positions hoping for the bounce). Mean reversion entries are based on a single signal. Adding to losing positions doubles the loss when the reversion does not come.

The famous practitioners

Larry Connors. Author of multiple books on short term mean reversion strategies. The RSI(2) system is his most popular contribution.

James Simons and Renaissance Technologies. Use sophisticated statistical arbitrage and mean reversion strategies on short timeframes. Among the most successful hedge funds ever.

D.E. Shaw and Two Sigma. Quantitative hedge funds that include mean reversion as one of multiple strategies.

Most pairs trading and statistical arbitrage at hedge funds is mean reversion at its core. The strategies bet on temporary deviations between related assets returning to their historical relationship.

The setup for retail mean reversion

Universe. Liquid large caps and broad index ETFs (SPY, QQQ, IWM). Avoid small caps and recent IPOs where the math does not work cleanly.

Timeframe. Daily for swing mean reversion. 5 to 15 minute charts for intraday mean reversion.

Entry signal. RSI(2) below 5 on a stock above its 200 day moving average. Or Bollinger Band lower touch on a stock in a range.

Entry execution. Buy at the close of the signal day or the open of the next day.

Stop loss. Below the most recent swing low or 5 to 8 percent below entry, whichever is tighter.

Profit target. RSI(2) crossing above 50, or price returning to the 20 day moving average, or 3 to 5 day time exit.

Win rate target. 65 to 75 percent. Mean reversion has higher win rates than trend following. The trade off is the occasional large loss on a trade that fails to revert.

Where the audit fits

The audit reads the actual mean reversion trades and identifies whether the higher timeframe filter was applied or whether the trader was buying falling knives. For most retail mean reversion traders the pattern is taking signals on stocks in strong downtrends, which produces the large losses that wreck the strategy. The plan locks the filter rules. Five to seven pages.