I was wondering if anyone has looked into modeling golfer scoring at the hole level. In particular, has anyone found a way to use DG data to estimate scoring distributions on holes of different length/par? For example, it seems intuitive that a driving distance correlates with par 5 scoring and with eagle and birdie rates on par 5s. It also seems intuitive that approach skill from 150 yards would have an impact on par 3s about that length.
I’ve been thinking of this in the context of how to apportion to strokes gained per round to each hole. For example, for a player expected to gain 1.8 strikes per round, you could:
- Divide it evenly and expect the golfer to gain 0.1 strokes on each hole. But this seems overly simplistic because on an easy par 3 it’s difficult to gain 0.1 strokes on the field and it would be easier on a short par 5 where eagles are possible.
- Apportion the strokes based on the the variance of the hole’s scoring average. A difficult par 5 with eagle and double bogey in play would allow for more strokes to be gained by each golfer. The challenge is that hole scoring is not normally distributed. But I assume some method of scaling and standardization could make this possible.
- Both 1 and 2 assume a golfer will gain strokes “equally” on each hole but I think we all realize that some golfers gain on the field on par 5s and others do so on par 3s. The ideal would be to allocate this based on a golfer’s skill set.
I’m hoping there’s a way to do this with DG data. If not, I think it would require getting hole by hole scoring data. From that, it should be too hard to determine strokes gained as a function of hole length and hole historical scoring. But I haven’t been able to find historical hole by hole scoring data.