What mean reversion actually is

Start with the word. The mean is just a typical level a price tends to sit near — a moving average, a fair value, a long-run relationship between two assets. A mean-reversion strategy assumes that when a price stretches unusually far from that level, the gap is more likely to close than to widen further. Buy what has fallen too far, sell what has risen too far, and collect the return as the price comes home. That is the whole premise; everything technical is a way of deciding what “too far” means and when to act.

Why prices revert at all

Reversion is not magic and it is not guaranteed. It shows up because a lot of real market behaviour is temporary: a forced seller finishes selling, a panic exhausts itself, a spread between two related assets gets arbitraged back. When the cause of a stretch is transient, the price tends to return once the cause passes. When the cause is permanent — a genuine change in value — it does not, and that is the failure mode every honest version of the strategy has to respect.

The two numbers behind “too far”

In practice, “stretched too far” is measured against the price's own recent variability. A move of a given size means one thing in a calm market and another in a wild one, so the strategy scales the distance by how much the price normally wanders — its volatility. A stretch of two or three times the usual range is a different signal from a stretch of half of it. This is why two reversion traders looking at the same chart can disagree: they are using different windows for the mean and different thresholds for the stretch.

Where the idea breaks

The strategy's great weakness is the trend that does not revert. A price falling for a real reason can keep falling well past any historical “too far”, and a reversion trader who keeps buying the dip is, in that case, just funding someone else's correct view. The discipline that separates a method from a hope is the stop: a level at which you concede the stretch was not temporary. A reversion strategy without a hard rule for being wrong is not a strategy, it is a martingale waiting to blow up.