Mean reversion is a trading strategy based on the observation that prices tend to fluctuate around a long-term average and that extreme deviations from this average are temporary. When a price moves significantly above its historical mean, a mean reversion strategy sells (expecting the price to fall back). When a price moves significantly below its mean, the strategy buys (expecting a recovery). The concept is often summarized as "what goes up must come down" and vice versa, though the reality is more nuanced.
The statistical basis
Mean reversion is grounded in the statistical concept of regression to the mean. If a process has a stable long-term average, extreme observations are likely to be followed by observations closer to the average. In financial markets, this manifests as price oscillations around a central tendency, whether that tendency is a simple moving average, a fair value estimate, or a statistical equilibrium.
The key challenge is distinguishing between temporary deviations (which will revert) and permanent regime changes (which will not). A stock that drops 20% might be temporarily oversold and due for a bounce, or it might be adjusting to legitimately bad news that permanently reduces its value. Mean reversion strategies must incorporate mechanisms to differentiate between these scenarios.
Common mean reversion approaches
Pairs trading is a popular mean reversion strategy that identifies two historically correlated instruments. When the price ratio between them deviates from its historical average, the strategy buys the relatively cheap instrument and shorts the relatively expensive one, profiting when the ratio reverts. This market-neutral approach reduces exposure to overall market direction.
Statistical arbitrage extends pairs trading to large portfolios of instruments, using statistical models to identify and exploit temporary mispricings across dozens or hundreds of assets simultaneously. These strategies typically hold positions for hours to days and rely on diversification across many small bets.
Bollinger Band strategies use price deviation from a moving average, measured in standard deviations, to identify oversold (buy) and overbought (sell) conditions. When the price touches or crosses the lower band, the strategy buys. When it reaches the upper band, the strategy sells.
Mean reversion versus momentum
Mean reversion and momentum are conceptually opposite strategies that tend to perform well in different market environments. Momentum profits from trend continuation. Mean reversion profits from trend reversal. Interestingly, both effects are well-documented in financial markets, but they operate on different time scales. Very short-term returns (seconds to days) often exhibit mean reversion. Intermediate-term returns (months) exhibit momentum. Very long-term returns (years) revert again.
Practical example
A pairs trading strategy monitors two oil companies, A and B, whose stock prices have historically maintained a ratio of 1.5. When the ratio expands to 1.8 (A is relatively expensive), the strategy shorts A and buys B. When the ratio contracts back to 1.5 over the next two weeks, both positions are closed for a profit. The strategy does not depend on overall market direction because the long and short positions offset each other.
How Tektii helps
Tektii's tick-level data and realistic execution modeling are particularly valuable for mean reversion strategies, which often trade frequently and are sensitive to execution costs. The platform supports multi-instrument backtesting, making it possible to test pairs trading and statistical arbitrage strategies that require simultaneous positions across multiple assets. Tektii's slippage modeling ensures that the narrow edges typical of mean reversion strategies are evaluated realistically rather than inflated by frictionless execution assumptions.