OT - Did stroke hole allocation method change in the US with WHS?

[Jeff M Johnson]

As the handicap chair at my club, I recently worked with the Washington Golf Association to re handicap our course after some minor remodeling (a few tees moved, and a few bunkers moved).  According to them, most of the stroke - index determination is done by a computer model already, and while they do not specify the handicaps of the individual holes, they do give you a recommendation.  I have years of tournament data, so we used that instead, but the recommendation was only off on one or two holes. 

I think the new allocation of using mostly difficulty (average score over par) to handicap the holes is best for match play if most of the matches played on your course are between golfers of similar handicap. I think that happens a lot more often than scratch v bogey golfer matches. I know it does on our course, as golfers of similar ability tend to play together.

I also feel the usga and the state golf associations are missing an opportunity to even up all the golf clubs by using all the hole by hole data they are gathering in the USGA app. I think it would be pretty easy to figure out which handicaps travel really well, and using stuff like t-square analysis figure out if some clubs are under or over handicapped (ie stroke-index too low or too high.)  Maybe that will get rolled out in a few years.

[Wayne_Kozun]

Thanks Carl, you verified that there have been changes in how this works.

[Wayne_Kozun]

I also feel the usga and the state golf associations are missing an opportunity to even up all the golf clubs by using all the hole by hole data they are gathering in the USGA app. I think it would be pretty easy to figure out which handicaps travel really well, and using stuff like t-square analysis figure out if some clubs are under or over handicapped (ie stroke-index too low or too high.)  Maybe that will get rolled out in a few years.

I agree with this.  Have you ever tried recalculating your handicap based on home and away scores and seeing if there is a significant difference? 

[Matt Schoolfield]

I also feel the usga and the state golf associations are missing an opportunity to even up all the golf clubs by using all the hole by hole data they are gathering in the USGA app. I think it would be pretty easy to figure out which handicaps travel really well, and using stuff like t-square analysis figure out if some clubs are under or over handicapped (ie stroke-index too low or too high.)  Maybe that will get rolled out in a few years.

I agree with this.  Have you ever tried recalculating your handicap based on home and away scores and seeing if there is a significant difference?

So, I'll get a bit philosophical, because I think that this comment has a presupposition that we generally take for granted, but I don't think is actually true.

I cerotainly can't speak for the governing bodies, but when I think of handicaps, I generally think they function they serve is to effectively create an effectively random, performance-based winner based on statistical odds of performance on earlier rounds.

So, for any handicapping system to function well, the following need to be true:

1. Players' results while after a round are, generally, nearly normally distributed.

I think this is generally true, except in the cases of improving rapidly (via, equipment or lessons), which could cause the distribution to skew strongly. So, no worries here.

2. Players' established handicaps generally produce statistically similar net scores across different types of courses.

I don't think this is true at all. Perhaps it is for some, but if a player is very adept at playing, say, windy/linksy courses, they can establish a higher handicap on parkland, and then over-perform on a linksy course.  Granted, taxonomy of golf courses is extremely challenging, but I still think it's a serious shortcoming.

The previous comments get to this, but I think it's a bit more complex. It's not just that different courses are incorrectly handicapped, but that players with different skills can and will perform differently at different types of courses.

3. Players' established handicaps generally produce statistically similar net scores across different types of rounds.

This is my biggest criticism of the current way handicaps are calculated. The vanity handicap and the sandbagged performance can trivially be eliminated if the system created a standard error between different types of rounds. That is to say, in playing a casual round, some people (not naming any names here) tend to perform better than during tournament rounds. Whether that is a legitimate difference (via, say, the use of stroke-and-distance) or an incorrectly calculated handicap is immaterial. Whether it's a vanity handicap or a sandbagger's handicap is also immaterial. By comparing performance during causal rounds and tournaments during that period, we can find the expected delta between the two, and simply apply that to new tournament net scores.

To continue this line of thinking, I also contend that establishing standard errors could be extremely useful in handicapping different types of games. Match play results should differ strongly from stroke play results, so using a standard error between stroke and match play results could more accurately handicap those games.


I do think the current system is generally fine, but it's fairly obvious that the current system is easy to manipulate, and the continuous complaints about sandbaggers means that changes could non-trivially improve the system.

[Michael Felton]

Matt - what you say about different types of players is definitely true, but I don't think there's a simple way to apply that generally to players' handicaps. If one player's game is more suited to links courses and another's to tree-lined courses, how do you handicap the two of them playing against each other in a way that is fair and equitable? Bear in mind both players might play tree-lined courses virtually all the time, so if they can have a close game on those courses, then the first player is going to be hard to beat if you suddenly drop them on a links course. First player would get better and second player would get worse, but you'd have had no way to know that until they play there. Combine that with the general noise of variance of play and insufficient data and you get to a "in an ideal world" situation that is not practically doable in real life.

[Matt Schoolfield]

Matt - what you say about different types of players is definitely true, but I don't think there's a simple way to apply that generally to players' handicaps. If one player's game is more suited to links courses and another's to tree-lined courses, how do you handicap the two of them playing against each other in a way that is fair and equitable? Bear in mind both players might play tree-lined courses virtually all the time, so if they can have a close game on those courses, then the first player is going to be hard to beat if you suddenly drop them on a links course. First player would get better and second player would get worse, but you'd have had no way to know that until they play there. Combine that with the general noise of variance of play and insufficient data and you get to a "in an ideal world" situation that is not practically doable in real life.

So, I'll simply have to respectfully disagree with some of this. All of this should be fairly trivially to calculate if we switched from the context-neutral handicapping system we use now, to one that statistically correlated things. Clustering algorithm could easily group kinds of players by their differences in performance on different courses, again filling in any gaps with collaborative filtering algorithms. Again, I'm not saying it would be perfect, but it would be more effective than what we have now, and would get better over time.

Now, I where I agree is that I don't think people actually want that.

The irony of our handicapping system is that we want winners to be effectively random, but don't want it to feel like the winners are random. We want to see the exceptional play and I think adjusting people's scores for the type of course feels like giving people free strokes for making poor decisions, rather the strokes being base on some less tangible "skill level".