Quote:
Originally Posted by CohibaBehike
I believe the single most important factor in PGA is course history. Not even sure how you could put that in a model.
Quote:
Originally Posted by CalledDownLight
I've got a few ideas and not sure which would be the best predictor, but probably something like Projected Points = A(WRh-AvgFinish) + ...
A would be some coefficient, WRh is historical world rank going into the event, and AvgFinish is the Average Finish in that event. Thus, that portion of the equation would add projected points to players with finishes above their World Ranking and decrease totals for players below it.
Quote:
Originally Posted by ntanygd760
I would think the best way would be to group all the holes they play and group them into like holes. Similar shot shape required, fairway width hazards ect. It would be a ton of work but if you made the groups just generic enough you could predict the scores you would expect everyone to get on each hole ect. No idea how you factor in people yipping 4 footers but there is a lot of shotlink data out there. All this seems like way too much work but I think it is the direction I would go to make a predictive model for the PGA
been thinking this over for the past few days and I might consider doing something like this:
player's actual course history to get a baseline. obviously a low sample but it's the best immediate baseline.
then rate courses on a similarity scale (length, slope, rating, grass type, fairway width, hazards, fades/draws required, green speed etc.)
take scores from another course and project a range for the current course. the more similar, the tighter the range.