Something that's been bugging me.

[Shot Clock]

Many stats are based on possession data. Possession data that is calculated generally uses the .44 modifier to take 12% of all free throws out to account for And-1's, flagrants, technicals and 3 pt free throws (3).

That modifier has stayed constant throughout league history regardless of rule changes that had to have an impact.

You would get FT's for illegal D calls. Three shots to make 2 and 2 to make 1 wasn't eliminated until 1982. It wasn't until 1995 that players got 3 shots for getting fouled shooting a 3.

The college game decided the .475 multiplier better fit their game meaning only 5% of FTA's were excluded. It's not an insignificant variable. If I took data from 1987 (.538 TS%) and used the college variable the league average is now .532. If it went the other way to even something like .430 the league average is now .540. That's a pretty big swing.

I read something recently that said they compared actual possessions in a recent season to calculated ones and they were very close. Even if that is the case there's almost no chance the same variable is going to apply to a league in the past before numerous changes.

Since so many stats are based on calculated possession data this really makes me question the use of them outside of the season they are measured. And for TS% every player is treated the same regardless of their ability to get an and-1. I'll elaborate with a really crude example

A player that makes 2264 pts on 1538 FGA and 814 FTA will has a TS% of .597. If this player never went up and tried to complete a basket when he heard the whistle and every single FT was the result of a 2 shot foul because lets say he was a pro flopper his actual TS% would only be .580 based on a .5 multiplier that eliminates no FTA's and counts 2 as one possession.(I say only just based on the difference not the number) We'll never see that extreme but this was to show that there are big variations in player circumstances for getting to the line that aren't reflected in that .44.

[blabla]

Very true. TS% is imperfect, and there's no reason for it to be this way in the age of PlayByPlay data.
ESPN's Real +/- works with true "points per shot" instead of TS%. I believe, nbawowy.com does, as well

[Shot Clock]

I agree. I just think that there is so much importance put on TS% and yet it's so flawed. Yet the same flaw applies to pretty much all possession based stats.

[Blackmill]

I was thinking about the 0.44 moddifier when watching some of Shaq's 2000 finals highlights. He had four and-1's in the first game and I believe two in the second. I don't remember the values, but I calculated his actual TS% and it was somewhere between 10 and 15% higher then his estimated TS%.