Sunday, July 26, 2009

Rethinking Go Rating

I find that some tweaks are in order . . . so I’m re-working some of the numbers. This has caused a delay in the big tables of Go Ratings (that we’ll use for frames of reference down the line).

  • I’m rethinking the issue of using 77% (numerically, .770) as the baseline for Free Throws. Over the last 20 years the league average has not been 77%, and it normalizes more towards 75% (.750). As a result I’m probably going to change the formula accordingly.
  • Similarly, I think that my baseline for Assist:Turnovers really hurts non-passing specialists too much. Not everyone in the league is an awesome ball handler . . . yet they are still capable of making good enough passes to keep the defense honest. The point of Go was to see how much pressure a single player put on the defense. Turnovers are already another step of the equation as it is. Putting the base line from 1.48 down towards 1.25 isn’t going to shake things up too much. Good passers will still be good – but now more forwards will not be HURT for attempting to pass out of the post.
  • The number for Offensive Rebounds (0.2673) was taken from the current league average, though this is like the FT% number . . . and does not stand through the test of time. Heck, the “proper” number for the 1984-1985 season was closer to 30%, which would cause havoc with today’s style of play. This number needs more work to find balance.
  • Lastly, with Shooting I don’t think the last part of the equation should be “total points / 10”, that’s not good enough as points is a total crapshoot sometimes – obviously the guys who shoot the ball more or play more minutes will look better because of this.

So . . . hold tight stat-a-roos, I’m going back to the work book to figure this one out.


Ivan Bezdomny said...

All seems reasonable...

Except that the value of assists is still way too high. You do (total assists * 2.23) * [assist to TO ratio above league average]. The 2.23 is based on FG vs 3FG balance. So you give full credit to the assister for the points from the shot. What about the shooter? Isn't this double counting in bias of passing guards?