5. Mean reversion: the price returns to its average

Level 5: Momentum, Mean-Reversion and Pairs Trading Strategies

So far, trend and momentum share the same underlying bet: that recent movement continues. Mean reversion bets on exactly the opposite: that prices which have drifted a long way from their historical average tend, sooner or later, to come back toward it.

There are several ways to apply this idea. Single-asset reversion looks at the historical average price of a particular share, and buys when the price is abnormally low, sells when it is abnormally high. Pairs trading, which you are going to program in the rest of this level, applies the same logic but to the difference between two related assets, rather than to the absolute price of a single one. Sector reversion extends that same idea to a whole sector against the broad market. Currency reversion applies the concept to forex pairs that have drifted from their historical exchange rate. And there are also options-based variants, which allow versions with downside protection or extra leverage.

The tools for spotting these deviations are the ones you already know: moving averages to define what "the average" is, Bollinger Bands to visualize how many standard deviations the price has drifted, the RSI to detect overbought or oversold conditions, and the z-score — which you are going to use in the next module — to quantify exactly how many standard deviations that distance amounts to.

Mean reversion does not come free: it fails in strongly trending markets, where "what goes up keeps going up" for far longer than expected, and buying "the cheap thing" turns into buying something that keeps falling. It also fails if the asset's fundamentals have genuinely changed — the historical average stops being a valid reference point if the company (or the sector, or the country) is no longer the same. And, like any statistical strategy, it works with probabilities, not certainties: there is no guarantee that a price far from its average will return to it within the time horizon you care about.