Why the mechanism matters
If objects are selected for an unusually high result, the selection may include both persistently strong participants and those receiving temporary random help. On repeat measurement, the random component may disappear, so the selected group’s average result moves closer to normal. This is a statistical mechanism, not a promise that a price will return to an individual purchase cost.
Different questions
A margin recovering after a one-off order, a quoted price recovering and a negative relationship between future returns and past returns are different hypotheses. One table cannot confirm all of them. Historical studies of stock-market returns show how much the conclusion depends on the horizon, period and test power.[1][2]
Typical error and limitation
Treating a fall as sufficient reason to buy is an error. It may reflect a lasting loss of earning power rather than temporary noise. Name the variable, the stable baseline and the mechanism that is supposed to produce recovery. The author’s example and the distinction between a temporary deviation and a regime change appear in the [guide to reversion and extrapolation](/guides/mean-reversion-extrapolation/). Even an observed statistical tendency does not determine timing or guarantee the survival of a specific asset.
[1] Poterba, Summers, Mean Reversion in Stock Prices. [2] Kim, Nelson, Startz, A Reappraisal of the Empirical Evidence. Primary materials: [Mean Reversion in Stock Prices](https://www.nber.org/papers/w2343) and [A Reappraisal of the Empirical Evidence](https://www.nber.org/papers/w2795).