Mean reversion is a catch-all for many different strategies that all share the common trait of fading deviations from a historical mean.
That mean is frequently price, but it can also be a measure of valuation or similar fundamental indicators.
Directional Mean Reversion
Directional mean reversion strategies involve projecting a directional price move in a specific security.
For example, if Apple stock falls two standard deviations from it’s 90-day linear regression, we might expect the net returns in those situations to lean positive.
There’s many different ways to devise a mean reversion system, whether it’s consecutive closes, a deviation from an average, or an indicator value. All of them essentially boil down to “buy when a stock drops too much” or “sell when a stock has increased too much.”
Relationship Mean Reversion
Most professional mean reversion traders don’t trade directionally. Instead, they trade the relationships between securities. A simple example is statistical arbitrage between two share classes.
In most cases, both shares have nearly identical intrinsic value and should be trading at the same value.
When dislocations in these share prices occur, pairs-traders will buy the cheaper share and sell the more expensive share, betting that the delta between the two will tighten.
Here’s a simple example of a correlation chart between two closely related stocks: Pepsi and Coca-Cola:
Traders might devise a system to buy the cheaper and sell the more expensive stock when the delta between the two stocks considerably widens.
To keep things simple, though, we’re going to focus on directional mean reversion strategies.