Mean Reversion Strategies

Among the many trading strategies that exist, two families dominate: mean reversion and momentum. They rest on opposite assumptions about how prices behave, and it helps to understand both. This lesson examines mean reversion, the idea that prices tend to return toward an average after straying from it, which connects directly to the stationarity and autocorrelation concepts of Unit 5.

The core idea

Mean reversion is the principle that prices, or the relationships between prices, tend to revert toward an average or equilibrium level after deviating from it. The strategy that follows is intuitive: when a price moves unusually far below its typical level, a mean-reversion trader buys, expecting it to rise back toward the average, and when it moves unusually far above, the trader sells or shorts, expecting it to fall back. The bet is on the deviation being temporary and the price snapping back to where it usually sits.

The connection to statistics

Mean reversion is the practical expression of statistical ideas from Unit 5. A mean-reverting series exhibits negative autocorrelation, where above-average moves tend to be followed by below-average ones and vice versa, pulling the series back toward its center. Mean reversion also requires the series to be stationary, with stable statistical properties over time, so that there is a genuine, persistent average for the price to revert toward. This is why mean-reversion strategies are often applied not to raw prices, which tend to trend and are non-stationary, but to constructed quantities like the spread between two related assets, which can be more reliably stationary, an idea developed in the pairs-trading lesson.

Measuring the deviation with a z-score

To trade mean reversion systematically, you need to measure how far the price has strayed from its average, and the natural tool is the z-score from Unit 5. You compute a rolling mean and standard deviation of the price over a recent window, then express the current price as a z-score: how many standard deviations it sits from the mean. A common rule enters when the z-score is extreme, buying when it is very negative (unusually cheap) and selling or shorting when it is very positive (unusually expensive), then exiting as the z-score returns toward zero.

Where it appears

Mean reversion underlies many specific techniques you have already encountered. The oscillators from Unit 4, such as the relative strength index, attempt to identify overbought and oversold conditions, which is fundamentally a mean-reversion concept. Bollinger Bands, which mark how far a price has strayed from its moving average, are frequently used to trade reversals back toward the mean. Statistical arbitrage and pairs trading, covered in the next lesson, apply mean reversion to the spread between assets. Across all of these, the common thread is buying what has fallen too far and selling what has risen too far, in expectation of reversion.

Mean reversion buys the dip and sells the spike, betting the deviation is temporary. The danger is a price that is cheap because something is genuinely broken.

When it works and when it fails

Mean reversion tends to work in range-bound, stable markets, where prices oscillate around a steady level and deviations really are temporary. It fails, sometimes catastrophically, in trending markets, where a price that looks cheap keeps getting cheaper and one that looks expensive keeps climbing. In a strong trend, the mean-reversion trader is repeatedly on the wrong side, buying declines that continue and shorting rises that persist. This opposite behavior to momentum is why the two families suit different market conditions, and why neither works all the time.

The critical risk

The biggest danger in mean reversion is mistaking a permanent change for a temporary deviation. A price may have fallen far below its historical average not because of a fleeting dislocation that will reverse, but because something has fundamentally and permanently changed, a deteriorating business, a structural shift, a regime change. Buying such a price in expectation of reversion is the classic error of catching a falling knife, holding a losing position as it keeps falling because the assumed equilibrium no longer exists. The series has ceased to be stationary, and the average the trader is betting on has moved or vanished. This is why mean-reversion strategies demand strict risk management, the stop-losses and position sizing from Unit 6, to limit the damage when a deviation turns out not to be temporary. The honest caveat is that mean reversion assumes the past equilibrium still holds, and structural breaks, which are not always visible in advance, can destroy the strategy.