The opposite of momentum, and the basis of relative value strategies.
The trap. Genuine mean reversion and a series that has simply not trended yet look identical in a finite sample. A random walk exhibits apparent reversion in-sample purely by chance, so a backtest is weak evidence.
The practical danger. A mean-reversion strategy adds to losing positions by construction, since a bigger deviation looks like a better opportunity. Without a hard stop that behaviour is unbounded, which is how these strategies fail.