Mean reversion is the trading philosophy that assumes prices oscillate around an average and that extreme deviations from that average are likely to reverse. Buy when oversold (price stretched below the mean), sell when overbought (above), let the snap-back pay you.
Common signals: RSI extremes, Bollinger Band touches, distance from a moving average measured in standard deviations, and statistical z-scores against a rolling baseline. The exact signal matters less than the discipline of defining 'too far' before you enter, then sticking to a stop if the move just keeps going.
The mathematical profile is the inverse of trend-following: high win rate (60-75% is typical), low average R (often under 1.0), and devastating tail losses when a 'mean reversion' trade turns out to be the first leg of a regime change. The strategy gets paid most of the time and then occasionally gives back six months in a week.
Where it works: range-bound markets, calm volatility regimes, and specific instruments with statistical reversion tendencies (pairs trades, sector rotations, futures spreads). Where it fails: trending markets, volatility expansions, and anything where the 'mean' you're betting on shifts mid-trade. The single highest-leverage skill in mean-reversion trading is recognizing when the mean itself has moved — and most traders find that out the hard way through journal review after the fact.