In mathematical terms, the financial concept of mean reversion is better known as “regression to the mean,” which refers to the likelihood of a deviated dataset to revert back to the mean.
In other words, if you flip a coin twenty times and all twenty flips result in heads, if you were to do another 10,000 coin flips, it would be extremely likely for the rate of heads occurrences to regress to nearly 50%.
Financial mean reversion is similar but is more ephemeral.
Mean reversion strategies have an underlying assumption that a historical mean of some sort has significance; that considerably deviating from that mean in short order makes it more likely for the security to bounce in the opposite direction.
However, you’ll find that many mean reversion traders don’t actually believe that any historical mean holds significance.
Instead, many view them as convenient reference points. If a stock is three standard deviations away from its 10-day mean, that’s an indicator that it just made a large upside move. And there are potentially profitable signals to be gleaned from that.
As a generality, stocks, and indices in particular, have a strong tendency to mean revert, as opposed to commodities and currencies, which tend to trend more.
But why is this?
The academic literature raises many potential explanations, but I think leverage and margin make the most sense.
When trading stocks, the margin requirements are high.
You typically need to commit at least 25% in margin to the position, making stock trading systems capital intensive.
Compare this to the futures markets, where maintenance margin requirements are routinely less than 10% of the contract’s notional value. Similarly, significant leverage is available in the forex markets.
So why is this significant?
When your stock positions are cash-secured, your exposure levels to individual positions increase significantly when those positions grow in value.
This necessitates rebalancing, which creates selling pressure in the best-performing stocks as they continue to outperform.
Contrast this to commodity and currencies, where even highly leveraged portfolios only utilize a fraction of their total cash. It becomes easier to understand why there’s less friction to the occurrence of trends in these markets.
Mean reversion strategies fade large deviations from historical prices. Buying stocks at 52-week lows is a simple example, while a statistical arbitrage pairs-trading system is more complex.
The 52-week low strategy assumes that the stocks have been sold indiscriminately and are due for a bounce back to some historical mean. The idea here is that the stocks have simply gone down too much.
A pairs-trading strategy, whether it’s based on the relationship between two similar companies, or even between two shares classes of the same stock, assumes that you can profitably trade when there’s a significant deviation from the historical correlation.