Definition
Mean reversion is the tendency of a time series — a price, a spread, a ratio, a volatility level — to drift back toward its long-run average after deviating from it. The deviation is sometimes random noise that decays on its own and sometimes the product of a structural anchor (an arbitrage relationship, a fundamentally-bound ratio, a regulated band) that pulls the series back. The opposite tendency is momentum, where deviations persist or extend.
Mean reversion is the basis of an entire family of trading strategies: pairs trading on the cointegrated spread between two related securities; contrarian equity strategies that fade short-horizon overreactions; volatility-selling strategies that short implied vol when it spikes far above its average. Each rests on the same wager — the deviation from equilibrium will close before it grows lethal.
Why it matters
How it works
A mean-reversion strategy starts by identifying a series with statistical evidence of stationarity — a stationary residual from a cointegrating regression, an oscillating ratio with a clear band, a metric whose autocorrelation goes negative at short horizons. The strategy then defines entry triggers some number of standard deviations from the mean, an exit trigger near the mean, and a stop level beyond which the assumption of stationarity is judged broken.
The hard part is not the entry logic; it is the wait. A spread that has moved two standard deviations from its mean often moves further before it reverts. A trader who sized too large or used too tight a stop is forced out at the worst possible moment — taking a loss right before the trade would have worked. This pattern is why mean-reversion strategies are so sensitive to position sizing and to the stability of the underlying relationship. The famous failure of Long-Term Capital Management was not that its convergence trades were wrong about the long-run direction; it was that the divergences grew large enough to exhaust the firm's capital before reversion arrived.