Strikeouts, Stabilization and Surprising Swings
You’ll hear us talk about statistical stabilization here, and link to pieces like Russell Carleton’s or Derek Carty’s. The basic idea is that there are thresholds at which a stat moves into a decent sample and becomes more meaningful.
Maybe you’ll come away thinking that we’ve said that ‘x stat is stable so that’s what you’ll get the rest of the way,’ and if so, that’s on us. That’s not what stabilization means in this context.
What it means is that the r-squared number that correlates a player’s past stats with his future stats in that category has passed .50. That’s a mouthful, here’s another try: if you were to try and predict future work in a category, you’d regress their current work against the league average. At the stabilization point, you can use half their own number plus half the league average in your calculations.
One more try, in the most colloquial language possible: Stabilization is the point at which a number in a category tells us more about their future work than the league average.