Kind of looking for our model wizards to answer this as the documentation doesn’t make it fully clear to me…
Is there any time-related ‘decay’ factor to an individual’s Course History performance adjustment?
I.e. would a golfer who played THE PLAYERS in his last 5 years finished 1st, 1st, 3rd, CUT, CUT have THE SAME course history performance bump as the same player finishing in last 5 years CUT, CUT, 3rd, 1st, 1st?
Assuming skill at time of event is flat, field strength, all the other known yada yada.
1 Like
No decay in course history. I guess the logic was that we already struggle to get much predictive power out of course history, we can’t be down-weighting certain data points. But on weeks like this with a long history it’s worth thinking about.
1 Like
Thank you. I’m attempting to use a logarithmic decay factor based on when a player has last played the corollary course… trying to find some predictive power out of it, but admittedly not seeing as much power as I’d wanted.
One thing I would think be useful to decompose in course history is the SG categories. I would assume most of course history operates through tee-to-green, but even if it doesn’t, it’s just much less noisy.