As my $5 Patrons know, the foundation of Stable SPR is simple: different player stats stabilize at different rates. Which makes sense: it's easier to have a hot (or cold) streak in 3-point shooting percentage than it is to out of nowhere start grabbing 10 rebounds per game. Different stats have more inherent variance!
Stable SPR has used 20 player seasons and found how long it takes each statistic to stabilize. The photos on this post show how long player stats (and therefore Stable SPR) take to stabilize, as a % versus average. For example:
-TOs per minute take 114% longer to stabilize than the average rate stat.
-3P% takes 90% longer to stabilize than average among shooting stats.
Since it's been asked - how does this formula work for Stable SPR? It's pretty simple actually:
Weight = Samples/(Samples + MeanRegression)
Expected(Stat) = Current(Stat) x (Weight) + BaselineEstimate(Stat) x (1-Weight)
Where MeanRegression = # of samples of mean regression best predict the next game. This is either attempts (i.e. Free throw attempts for FT%) or Minutes played for rate stats.
The "MeanRegression" value is what is implied when referring to "Stabilization."