Occassionaly I write a blog post that could be useful to both future me as well as other people.

Regression towards the mean

regression, data analysis, machine learning

A lot of activities have a non-random component (e.g. skill) and a random component (luck). Observing an extreme case is on average due to a combination of this skill and luck component. When doing a new observation of the same case, it is typically more average, since the luck component is reset.