Changes to Handicapping System for Next Year

[Sakuraba]

The highest possible score a 10 handicap could post under the "old" system was 108 (on a par 72) now its 126, and that could be on a par 70. I think its a mistake.

[BC MIST]

The highest possible score a 10 handicap could post under the "old" system was 108 (on a par 72) now its 126, and that could be on a par 70. I think its a mistake.

Whether the player's scores were 108 or 126, neither would count in the calculation of the actual handicap factor, so using his extremely poor scores is not an indicator of his true potential, nor should they be used to criticize the change in the system.

More accurate would be to take actual scores that count in the factor calculation, apply the two ESC formulas and note the difference. Using actual hole by hole scores of a 13.2, 2011 RCGA factor, his new 2012 factor would be 14.4, a difference in course handicap of only 1.

While I can understand some golfers not appreciating seeing their factors go up by 1 or perhaps two strokes, fixing one flaw in the Handicap System is progress.

[jvincent]

While I can understand some golfers not appreciating seeing their factors go up by 1 or perhaps two strokes, fixing one flaw in the Handicap System is progress.

I'm trying to understand what flaw is being fixed by the change.

[Sakuraba]

Whether the player's scores were 108 or 126, neither would count in the calculation of the actual handicap factor, so using his extremely poor scores is not an indicator of his true potential, nor should they be used to criticize the change in the system.

More accurate would be to take actual scores that count in the factor calculation, apply the two ESC formulas and note the difference. Using actual hole by hole scores of a 13.2, 2011 RCGA factor, his new 2012 factor would be 14.4, a difference in course handicap of only 1.

While I can understand some golfers not appreciating seeing their factors go up by 1 or perhaps two strokes, fixing one flaw in the Handicap System is progress.

I see your point, hopefully a 10 handicap would not have a score higher than 108 counting, however it is conceivable. I guess he wouldn't be a 10 for very long.

[Kilroy]

I'm trying to understand what flaw is being fixed by the change.

Must be the one where single digit players needed to give up an extra stroke or two to the 18 (err.. make that 20) cappers.

[jvincent]

Put simply, it did not make great sense to treat a 1 handicap the same as an 18 handicap (maximum hole score of 2 over par) when their abilities are so different. By reducing the size of handicap ranges (they will become increments of 9) that problem is mitigated.

This is from the statement that Matt McKay made, and I actually agree with this. I just disagree with the way it has been implemented. If they wanted narrow ranges it was very easy for them to do that. Here, I'll do it for them.

Course handicap of 5 or better: Max = bogey.
Course handicap of 6 to 14: Max = double bogey.
Course handicap of 15 to 23 : Max = triple bogey.
Course handicap of 24 to 32: Max = quadruple bogey.
Course handicap of 33 or higher: Max = quintuple bogey.

See how easy that was and how simple it is to remember?

[orangeTANG]

Curious how this would affect two friends that play match play.

One consistently struggles on a Par 3, the other struggles on a Par 5. Under the previous system they would both take a max Double Bogey on their tough holes (<18HCP), but in the future calculations over time the advantage would definitely be to the person struggling on Par 3s since his overall handicap factor would be higher than the other player.

The difference of 7 on a Par3, compared to 7 on a Par 5 would mean that an additional 2 strokes would need to be given during the match if scoring was consistent among the last 20 rounds.

[BC MIST]

I'm trying to understand what flaw is being fixed by the change.

-Currently, in a head to head match, the lower handicap player has a slightly better than 50% chance of winning. The new ESC will bring that closer to 50/50.

[jvincent]

-Currently, in a head to head match, the lower handicap player has a slightly better than 50% chance of winning. The new ESC will bring that closer to 50/50.

Based on a study done using scores from ONE golf course. Yeah, good statistical sample there.

[Kilroy]

What about the famous claim that "The high cap can play 5 under his cap more easily than I can and that puts me at a disadvantage" You hear that all the time. Now we will hear it even louder. It's fine by me if that's the reality but good luck convincing most 5 cappers that it's fair to give up another stroke to Mr 18.

[Golfbum]

All I have to say is "Let the SANDBAGGING begin"

[BC MIST]

Based on a study done using scores from ONE golf course. Yeah, good statistical sample there.

For a long time I have been of the opinion that higher handicappers have a greater "margin of improvement" than single digit ones, and therefore have an advantage in match play. However, Matt's point above is based, at least, on some statistical evidence, regardless of the small sampling, and should carry a little more weight than our biased opinions. I would guess that the USGA has conducted studies on the issue and have adjusted the ESC that we are about to adopt, to the results.

[BC MIST]

All I have to say is "Let the SANDBAGGING begin"

Do you not mean continue? When you have examined the scores of members at a club, you will realize that the problem of players NOT posting scores is far more significant than players posting erroneous ones.

[Libbing]

If this handicapping formula is the same as the U.S., better handicappers have an advantage in a one-on-one match for two reasons:
1) there is a 4% reduction factor in the formula (so you compute the differentials based on slope/rating and then reduce it by 4%; so a higher handicapper loses a small amount there
2) the wider range of deviation in scores only helps a low handicapper when they are playing their absolute best, in the other situations they lose out.

On the second point, keep in mind that the handicap (in the US at least) is based on the best 50% of the last 20 rounds. It is not based on the average of the last 20 or all rounds over a time period. For simplicity of my example, consider then two players playing at one of three different performance levels:
a) their "best" performance (this is the level at which the handicap is computed - 25th percentile where 1% is lowest score)
b) average (at the 50 percentile)
c) worst (average of the worst 50% of scores = 75 percentile)

The system is setup so that if both players play at the best performance, the high handicapper is disadvantaged by 4%. However, if you assume that the higher handicapper also has greater deviation in scores in absolute strokes (e.g. if both have a 10% deviation, the high handicapper will have higher deviation in absolute number of strokes), then when both players play at average, the high handicapper would lose because the absolute (not %) differential in strokes between then best and average would be greater for the high handicapper. This would also apply if both players play their worst rounds.

So now consider 9 outcomes of a round where you have 3 outcomes for each players performance

Play in a given round (Best = best performance / where HC is computed)
----------------------
Low HC High HC Outcome
--------- ---------- ----------------
Best Best Low HC wins by 4%
Best Avg Low HC wins
Best Worst Low HC wins

Avg Best High HC wins
Avg Avg Low HC wins (absolute differential dings the high HC, plus the 4%)
Avg Worst Low HC wins

Worst Best High HC wins
Worst Avg High HC wins
Worst Worst Low HC wins

In that simple grid, the lower handicap wins 6 out of 9 times. This analysis might break down when its not a 1:1 match. Haven't thought enough about that one, but I think it is the same. As a 5-6 handicapper playing in a lot of outings with higher handicaps (15 - 30), I am in the top of the bracket a lot with solid chances to win. While I might not win that much more frequently than others, I am in position to win a lot more frequently (or, said another way, finish in the top 25% a lot more often).

I think for the system to be equal on average, you'd have to take out the 4% factor and you'd have to use the average of all scores. They don't do that because it makes it impossible for a low handicapper to beat a high handicapper in many situations. Consider a low HC with an average of 10 with range of +/-5. Then a high HC with a HC of 25 but a range of +/-10. When the high HC is playing their best (10 under average) it would be almost impossible for the low HC (best = 5 under average) to beat that. Thus, they only look at the best rounds in order to negate that issue.

Before anyone says it, I know I'm thinking about it too much - we had a big debate about this at work this year so (of course, between a low and high handicapper during a complete sandbagger contest for betting!)

[BC MIST]

If this handicapping formula is the same as the U.S., better handicappers have an advantage in a one-on-one match for two reasons.... Thus, they only look at the best rounds in order to negate that issue.

Very informative post and certainly supports what Matt McKay has posted above and what he has written in private emails.

[BC MIST]

Based on a study done using scores from ONE golf course. Yeah, good statistical sample there.

FWIW: Based on a discussion that we had yesterday with an RCGA director, the scores that were used by the prof at Simon Fraser University numbered in the "hundreds of thousands," and clearly showed that the low cap player has an advantage over the high one, hence the 2012 move to the current USGA, ESC system.

[jvincent]

I don't have an issue with the number of rounds, but the the fact that they were taken at exactly one course.

It is almost certain that the scores (and resulting conclusions from them) are in some way biased because of the layout of the course used. Does it have lots of wide open tee shots? Is there a lot of water in play? Are the par 3's relatively short but dangerous? Etc.

All of those could result in the lower handicap having an advantage at that course. Run the same study at a course with a different layout and you in all likelihood get a different result.

Sorry, but that study is simply bad statistics. D- on the methodology from me.

[Colby]

Because of the handicapping changes, we went over the handicaps in the National Capital Golf Tour, basically applying the USGA ESC to everyone to see what the difference would be. The result was that the majority of golfers had their handicaps go up by less than 1, one person went up just over 2 points and some actually went down slightly (because the rounds selected as their best 10 changed due to ESC changes)

[LobWedge]

I don't have an issue with the number of rounds, but the the fact that they were taken at exactly one course.

It is almost certain that the scores (and resulting conclusions from them) are in some way biased because of the layout of the course used. Does it have lots of wide open tee shots? Is there a lot of water in play? Are the par 3's relatively short but dangerous? Etc.

All of those could result in the lower handicap having an advantage at that course. Run the same study at a course with a different layout and you in all likelihood get a different result.

Sorry, but that study is simply bad statistics. D- on the methodology from me.

A course is a course is a course, John. Those hundreds of thousands of scores were taken from play over a number of years, in varying conditions, by countless numbers of players. Your argument only holds water if that multitude of rounds were played by the same person.

[LobWedge]

All I have to say is "Let the SANDBAGGING begin"

Do tell?

[jvincent]

A course is a course is a course, John. Those hundreds of thousands of scores were taken from play over a number of years, in varying conditions, by countless numbers of players. Your argument only holds water if that multitude of rounds were played by the same person.

Disagree.

Use Loch March, Kanata Lakes, and Greensmere Premier all from the gold tees as an example. The slopes and ratings are virtually identical. dHowever, there are two par 3s at Loch March which are MUCH more difficult for an 15-18 handicapper than any of the par 3s at Greensmere. Those holes alone can skew the scoring average significantly under the new system since it now changes from a max of 5 to a max of 7.

Yes, I know that only the best 10 scores count but even over time players are going to score higher on those two holes than normal. I have first hand knowledge of this.

My point is that from a statistical validity point of view, a study based on a single course only validates the "fairness" of the new system on that course. I have no idea what the layout of the course they used is like but we all know that some courses and holes simply play harder than their ratings would indicate and the results of the study can be biased by that.

To draw conclusions from a study at a single course is simply bad math. I know that math isn't supposed to be the RCGA's area of expertise but there are people out there who know better.

[BC MIST]

Disagree.

If 1000 scores from 1000 courses were used in the study, it is still likely that the same conclusion, that the lower handicap has a slight advantage over the high one, would result.

In post 44, Libbing outlines in his point 1, how a high handicap player ends up with a slightly lower handicap, because of the 4% reduction. If your differential total is 20, your factor is 20/10*.96 = 1.9 or 2. If my differentials are 10 times higher, or 200, I would have a factor of 200/10*.96 = 19.2 or 19. My factor is not 10 times higher than yours.

I would be interested in hearing your comments on his points outlined in his point 2, where he shows that of the 9 possible outcomes, 6 favour the low handicapped player, partly because of the 4% reduction and partly because of the greater range of scores for the high handicapped.

[jvincent]

If 1000 scores from 1000 courses were used in the study, it is still likely that the same conclusion, that the lower handicap has a slight advantage over the high one, would result.

If the study had used even 10 courses, I would have no issues with the results. Extrapolating from one data point is not valid. Me saying it is not likely is just as valid as others saying it is.

In post 44, Libbing outlines in his point 1, how a high handicap player ends up with a slightly lower handicap, because of the 4% reduction. If your differential total is 20, your factor is 20/10*.96 = 1.9 or 2. If my differentials are 10 times higher, or 200, I would have a factor of 200/10*.96 = 19.2 or 19. My factor is not 10 times higher than yours.

I would be interested in hearing your comments on his points outlined in his point 2, where he shows that of the 9 possible outcomes, 6 favour the low handicapped player, partly because of the 4% reduction and partly because of the greater range of scores for the high handicapped.

I need to noodle on his post a little more. I think there may be a difference if we are talking about match play vs. net stroke play.

[jvincent]

I have noodled on Libbings post and I think his table is oversimplifying things. I need to crunch some numbers and come back with a better analysis.

[BC MIST]

I have noodled on Libbings post and I think his table is oversimplifying things. I need to crunch some numbers and come back with a better analysis.

I think that his model makes sense for stroke play but would be less valid for match.

Interestingly, my 2.3 factor has a range of scores on Premiere of 67 to 79 for this year, quite close to his predicted range of plus or minus 5 for the lower handicapped player. I don't know what the range of a 22 handicapper would be but his + or - 10 sounds reasonable, although the minus would likely be a lot higher. Regardless, it seems that the move of changing the ESC so that some handicaps rise a little, will increase the fairness of matches. When we have our singles play downs next season, I will have a close look to see if the results are balanced or not.

[jvincent]

There's definitely a difference between stroke and match play. I don't have an easy (read automated) way of doing that right now so I'm just looking at stroke play for now.

As part of my number crunching I'm using the "Exceptional score" probability to limit the low end of the scores for the data I generate. Oddly enough my distribution function matches your data pretty well.

Once I get all the numbers crunched I'll summarize my stats/methodology.

[Libbing]

Actually, I think the handicap system is setup to handle match play and not stroke play. This is due to 1) the maximum score on a given hole and 2) the intended application of handicap strokes to the most difficult holes first.

On the first point, using the handicap system in stroke play favors the better golfer even more. If the system was designed for stroke play, a maximum score by hole would not make sense. You'd look to cap out the total score since that is the only score that matters at the end of the day. Consider a high handicapper that hits a 10 on a hole but they can only record a 6 or 7. In match play, the extra couple of strokes have a lower chance of mattering because the odds are higher that, for example, they had already lost a hole by the time the snowman 8 showed up. But in stroke play, every stroke is the same for your final score. I also think that a higher handicapper is more likely to get strokes shaved off on a hole by hole basis than a lower handicap. I know I shave fewer strokes off my score than my higher handicap players do (no stats on that, but seems to be the case).

On the second point, handicap strokes are designed to be given out on the most difficult holes first. So if you have a 4 HC point differential with your playing partner, the 4 strokes are given on the 4 hardest holes. With stroke play, that doesn't even apply since its only applied at the total score.

I think the handicapping works best for a 1 on 1 match play. With more than 2 competitors it starts to break down a little bit in match play because the application of the strokes to the hardest holes doesn't work out as well. For example, if you had three people with a 0 handicap, a 2 and a 5. The 2 handicap gets a stroke on the first and second hardest hole. The 5 gets on the hardest 5. That works fine for a competition against the 0 handicap. But consider he bet between the 2 and the 5. In that match, the 5 ends up getting a stroke advantage over the 2 on the 3rd, 4th, and 5th hardest and he gets no advantage on the two hardest holes. That's a quirk.

I think the handicap system work least effectively on stroke play. Before I got into single digits handicap, I would only bet low handicappers in match play so that a double digit score on a hole didn't kill the round for me (it only killed that hole).

[Libbing]

On the subject of the validity of using 1000s of rounds on one course vs being spread out over several / 100s of courses, I agree that it is probably not statistically valid, but it is probably close for the purposes for which the study was used (that is, to give a general idea of the impact).

By design, course slopes / ratings are setup to figure out the difficulty that the golfers of differing abilities will have on a given course. It is essentially an Y=MX+B line graph (Y=handicapped score for the course, M=slope/average slope of 113, X=your handicap index, course rating=B ) that plots estimated scores on a given course based on skills of the golfer. They use course design features such as number of hazards, length, tightness of fairways, green difficulty, etc. etc. to determine the rating/slope. By definition, that rating should plot the impact of those design features across a broad spectrum of players to derive that MX+B formula. However, I seriously doubt that every course is ranked perfectly and that every course and every player fits exactly on an MX+B line. It's like an economics theory that relies on "all else being equal" or "with perfect access to information" - that's great if it weren't for that fact that neither condition ever exists in reality.

I think that if you did that same test on several different courses you'd come up with a slightly different answer. But I don't think it would be off even by as much as a stroke unless the ranking and slopes of that particular course had just been botched in the first place (that is a pretty big assumption, which is the major reason in my mind why the validity is a bit in question - if the course was ranked incorrectly, then the whole thing is irrelevant; more course samples averages those issues out).

[jvincent]

OK, I spent some time with excel and the random number function so I have some numbers for consideration. First, here is an explanation of what I did.

I generated two sets of 20 differentials, one for a low handicap golfer, one for a high handicap golfer using the random number generator. These were used to generate a handicap factor according to the RCGA formula and a course handicap for a course with a rating of 71.5 and slope of 128.

For the low handicap golfer I limited the range of the differentials to be from 1 to 11. The idea here was to get a target handicap of around 5 and a minimum differential of 1 the lowest differential a 5 handicap can expect to post without entering the "career round" zone according to the Popeofslope data. Based on the random number generator the actual range of course handicaps generated was from 2 to 6.

Similarly, for the high handicap golfer I limited the range of differentials to be from 12 to 30. The target handicap in this case was 18 and the same same rationale was used for the ranges. Note that this range is based on the "old" ESC system since the max differential is lower for that system than the new system. I did not set the max to 36 because that would have caused the target handicap to drift higher than 18. The range of course handicaps generated for the high handicap golfer was 15 to 21.

For each set of 20 differentials the low and high golfers NET scores were compared to see who one that game. I did this for 100 sets of differentials, so what I effectively ended up doing was comparing the scores for 100 golfers across 2000 stroke play competitions and counted the number of times the low handicap "won".

Problems with the above methodology:
1. Since I started with differentials, which are by definition ESC adjusted scores, the actual "tournament" scores would be higher on average for both golfers.
2. I gave ties to the low handicap golfer under the assumption that he would have fewer ESC adjusted scores so in all likelihood would have won the ties in a stroke play competition.
3. Scores are assumed to be randomly distributed. In practice I doubt this is really the case for most golfers.

The results: The low handicap golfer won 55% of the matches. This number jives with what Matt MacKay posted earlier.

I then re-ran the calculation with a wider range of differentials for the high handicap golfer to account for the new ESC system. The new range is 12 to 34 and the generated handicap ended up being 16 to 23. Again, this is what was expected for the new system, i.e. a slight increase in handicaps at the high end.

The results (part 2): The low handicap golfer won 58% of the matches. This is NOT what the RCGA was expecting although the numbers are close enough that it may not be statistically significant.

So, my take on this is that for stroke play, the lower handicap golfers are at a slight advantage no matter which system you use. This is not a big surprise since they are more consistent.

If I can figure out a good way to do this for match play I will post that.

[jvincent]

I think that if you did that same test on several different courses you'd come up with a slightly different answer. But I don't think it would be off even by as much as a stroke unless the ranking and slopes of that particular course had just been botched in the first place (that is a pretty big assumption, which is the major reason in my mind why the validity is a bit in question - if the course was ranked incorrectly, then the whole thing is irrelevant; more course samples averages those issues out).

Definitely agree. I have always thought that there is too much subjectivity in the rating/ranking system.

As an example, the following courses are essentially equivalent from a course handicap perspective when played from the GOLD tees: Kanata Lakes, Loch March, Greensmere Premiere. I don't think anyone on this forum would agree with that.