New USGA 9 Hole Score Differentials Seem Way Off

[iacas]

11 minutes ago, DaveLeeNC said:

This is not some small thing for 9 hole golfers.

 

Based on a model you made up for a single golfer. And you seemed to use my "14.2 becomes a 14.4 or something" and got… a 16.7, so that alone should tell you something in your models may have been off, if you were trying to use my 14.2 guy.

 

Also, a model is not what I said the next step was. I specifically said what I said about using actual real-world data from actual golfers because… well, let's put it this way: you don't think the USGA not only "modeled" it but used actual data from actual golfers to come up with this? You don't think they did what I suggested the "next step" for those complaining about it would be before they made this change, to see what they could expect? You don't think they tested the edge cases of "guys who play like 80% nine-hole rounds" and deemed the tradeoffs to be worthwhile?

 

20 hours ago, iacas said:

What would the next step be? To compare a few hundred golfers who play nine-hole rounds (from seldom to frequently), and evaluate over about the last 22 to 38 scores what their handicaps would be with both methods.

 

My guess is that a 14.2 might become a 14.4 or something. Until someone does that, it's all just conjecture. And we've had a lot of this conjecture.

 

Your model is just more conjecture, it's not data.

 

Even if it does mess up people who play only nine holes, if something is better for 95% of golfers and worse for 5%, should they still have done it? Or 90/10? Or 85/15? Especially if that 5-15% are not playing in 18-hole competitions very often (i.e. they can have a nine-hole league handicap)?

 

Get back to us when you have hundreds of golfers and thousands of rounds (at a minimum) of actual data, because the USGA has orders of magnitude more than that against which they have already checked this stuff. Until then… we can make "models" that we make up work out just about any way we want them to.

 

[DaveLeeNC]

9 minutes ago, Pnwpingi210 said:

I like this idea of modeling.

 

were the scores you input actual scores for a real golfer?

I just modeled a stream of 9 hole differentials with the statistics that I stated. dave

[Pnwpingi210]

2 minutes ago, DaveLeeNC said:

I just modeled a stream of 9 hole differentials with the statistics that I stated. dave

Do you play 9 hole rounds?  Why not just use real scores.

[DaveLeeNC]

16 minutes ago, Pnwpingi210 said:

Do you play 9 hole rounds?  Why not just use real scores.

I but I just started that so don't have a meaningful sample (which is way more than 20 scores). 

 

dave

[Pnwpingi210]

5 minutes ago, DaveLeeNC said:

I but I just started that so don't have a meaningful sample (which is way more than 20 scores). 

 

dave

That’s fair.  I be curious to see if using real data (your scores) results in different than the theoretical scores used in your model

[DaveLeeNC]

4 hours ago, iacas said:

Get back to us when you have hundreds of golfers and thousands of rounds (at a minimum) of actual data, because the USGA has orders of magnitude more than that against which they have already checked this stuff. Until then… we can make "models" that we make up work out just about any way we want them to.

 

The USGA could have (and probably did) some kind of least squares optimization of the data that they had. This most certainly does not look anything like a result that would come out of that. The algorithm (most suspiciously) seems to use the expected score of the unplayed 9 in its calculation and expected scores are not what determines index. In looking at my current index there are NO differentials in my top 8 that are as high as my mean differential. And I am not a particularly erratic golfer who puts up lots of low and high scores. 

 

So this really looks like some kind of compromise regarding appearance. I believe that we will see another 9 hole posting procedure in the next year or two once this approach settles in and folks see for themselves how this works. 

 

BTW, I did not try to match your 14.2/14.4 statement. I simply used some round numbers that generate an index (for the 18 hole case) that is around 3 strokes lower than the mean (and the numbers made sense to me). 

 

In the meantime the aggregate of 9 hole golfers will see higher differentials than what they would see playing 18 holes and it will be that way for a while. Such is life. 

 

dave

[iacas]

47 minutes ago, DaveLeeNC said:

The USGA could have (and probably did) some kind of least squares optimization of the data that they had.

 

I'm strongly suggesting that the USGA took actual data from real golfers playing real golf and calculated their index "both" ways, and made a determination based on that as to whether it felt right, how different they were, etc.

 

For hundreds of thousands or even millions of scores. That they didn't have to do a sim with some guessed-at constraints.

 

47 minutes ago, DaveLeeNC said:

The algorithm (most suspiciously) seems to use the expected score of the unplayed 9 in its calculation and expected scores are not what determines index.

 

I don't really know what you're trying to say there. Could you clarify?

 

As I understand things, it's basically adding an average nine holes to your posted nine holes. That average nine holes is based on the actual scores (differentials) shot by people with your index.

 

A 7 index averages about a 10 differential.

 

47 minutes ago, DaveLeeNC said:

In looking at my current index there are NO differentials in my top 8 that are as high as my mean differential.

 

Huh? That's math. The average of your lowest eight are generally going to be lower than the average of your 20.

 

Unless your last 20 are something like… 1 1 1 1 1 1 1 4 4 4 4 4 4 4 4 4 4 4 4 4 (index = 1.4, average = 2.95, and one of your counting differentials was a 4)… you're generally going to see that your handicap index [average differential of your best 8] is lower than the average of your last 20.

 

47 minutes ago, DaveLeeNC said:

I believe that we will see another 9 hole posting procedure in the next year or two once this approach settles in and folks see for themselves how this works.

 

I'd wager some money on that. I doubt it.

 

Again, do you genuinely think they didn't take hundreds of thousands if not more of existing scores from real golfers playing real rounds, calculate the indexes using the "old" method and the "new" method, before they decided to go with this new method?

 

47 minutes ago, DaveLeeNC said:

BTW, I did not try to match your 14.2/14.4 statement.

 

Okay.

 

47 minutes ago, DaveLeeNC said:

In the meantime the aggregate of 9 hole golfers will see higher differentials than what they would see playing 18 holes and it will be that way for a while. Such is life.

 

Generally, but not necessarily.

 

And the truly important thing is whether their handicap index is affected, not just a single differential.

 

Edited June 23 by iacas
got rid of 8 ) being converted to an emoji

[DaveLeeNC]

12 hours ago, iacas said:

I don't really know what you're trying to say there. Could you clarify?

 

As I understand things, it's basically adding an average nine holes to your posted nine holes. That average nine holes is based on the actual scores (differentials) shot by people with your index.

 

A 7 index averages about a 10 differential.

 

The algorithm doesn't look like something that would come out of some analytical optimization process. It looks like something that was chosen for either appearance or to be consistent with the new way of handling unplayed holes. If you wanted to use average scores for the unplayed 9 holes in your postings, then the new algorithm seems to do exactly that.

 

12 hours ago, iacas said:

Generally, but not necessarily.

 

And the truly important thing is whether their handicap index is affected, not just a single differential.

 

When you have a process that generally raises your differentials it is a process that will also generally raise your index. If you play one hole and then (round after round) post your average hole scores to generate your posted 18 hole score, your index will go up. If you play two holes and then (round after round) post your average holes scores your index will go up. And the same thing can be said for 9 holes.

 

It is interesting to note that the old way of handling unplayed holes used to be based on shooting your index, not your average. The philosophy seems to have changed for this case as well (possibly so they would be consistent as they are different degrees of the same problem). 

 

dave

[Newby]

1 hour ago, DaveLeeNC said:

It is interesting to note that the old way of handling unplayed holes used to be based on shooting your index, not your average.

 

I'm not sure what you mean by 'your average' but Erik has it right. The unplayed 9 are recorded as the average scores of all other players with the same Index as you.

 

WHS will use an “expected score” formula to take nine-hole scores and adjust them to 18-hole score differentials. Mining the data gathered from the 100 million scores posted yearly around the world under the WHS, a model scoring formula for every handicap index for males and females has been developed. The appropriate one will be applied to your index and create an expected score on any remaining holes you had for handicap purposes, taking the place of applying a score of net par for any missing holes.

The new formula is built to account for a standard golf course, so the calculations are no longer course dependent compared to the math applied in the past.

 

 

Edited June 24 by Newby

[sui generis]

14 minutes ago, Newby said:

Mining the data gathered from the 100 million scores posted yearly around the world under the WHS,

 

Uh, oh. There's the problem . . . not enough data. Sure enough, Joe Six Pack's pulled-out-of-his butt solution is where the ruling bodies should have looked. 😉

[nsxguy]

2 hours ago, DaveLeeNC said:

It is interesting to note that the old way of handling unplayed holes used to be based on shooting your index, not your average

 

"Shooting your index" happens roughly 2-3 times per 20 rounds, so 10-15% of the time.

 

Seems to me "average" would be a better assumption, a more likely occurrence, for the 9 holes not played. 

[Pnwpingi210]

5 minutes ago, nsxguy said:

 

"Shooting your index" happens roughly 2-3 times per 20 rounds, so 10-15% of the time.

 

Seems to me "average" would be a better assumption, a more likely occurrence, for the 9 holes not played. 

Also I’m not sure a statistical model captures the impact of fatigue, focus, blood sugar, alcohol consumption, body aches and pains in an 18 hole round vs 9.  This is why real golfers scores(such as the millions of scores the USGA uses) is better than a model that assumes none of those factors are different on hole 1 vs hole 18.

Edited June 24 by Pnwpingi210

[DaveLeeNC]

7 hours ago, nsxguy said:

 

"Shooting your index" happens roughly 2-3 times per 20 rounds, so 10-15% of the time.

 

Seems to me "average" would be a better assumption, a more likely occurrence, for the 9 holes not played. 

It depends on what "better" means. For the occasional one off unplayed handful of holes, it is probably a pretty good solution. Since your average is (in most cases) outside of the scoring range that is going to be used in calculating your index, a whole bunch of those averages thrown in is maybe not such a good idea (as it will generally raise your index). 

 

BTW, this is a very tough problem if you want a single solution that covers all cases from the occasional handful of unplayed holes to half of your 360 holes (the twenty 18 hole rounds that make up your index) being unplayed. 

 

dave

[DaveLeeNC]

8 hours ago, Newby said:

I'm not sure what you mean by 'your average' but Erik has it right. The unplayed 9 are recorded as the average scores of all other players with the same Index as you.

 

WHS will use an “expected score” formula to take nine-hole scores and adjust them to 18-hole score differentials. Mining the data gathered from the 100 million scores posted yearly around the world under the WHS, a model scoring formula for every handicap index for males and females has been developed. The appropriate one will be applied to your index and create an expected score on any remaining holes you had for handicap purposes, taking the place of applying a score of net par for any missing holes.

The new formula is built to account for a standard golf course, so the calculations are no longer course dependent compared to the math applied in the past.

 

 

FWIW, mean (or average) and expected values are the same thing for typical, well-behaved distributions. 

 

And thanks for posting the quote. I have read a bunch on this and it looks familiar but I don't recall where I saw it - did you save a link? 

 

dave

[iacas]

10 hours ago, DaveLeeNC said:

The algorithm doesn't look like something that would come out of some analytical optimization process. It looks like something that was chosen for either appearance or to be consistent with the new way of handling unplayed holes.

 

I don't know what makes you say that. I can see the logic behind adding an average nine holes to get an 18-hole score. Makes plenty of sense to me.

 

10 hours ago, DaveLeeNC said:

When you have a process that generally raises your differentials it is a process that will also generally raise your index.

 

You missed the point. If you play nine holes occasionally, it may not affect your index at all. All it would take is for those rounds to fall in the higher 12 differentials.

 

Like I said, until we see actual, real-world results (and not a sim with made-up constraints, and not tiny sample sizes like one golfer), if this results in the average handicap going up 0.1 or 0.2 strokes… this feels like much ado about nothing, with people focusing on the minority of rounds that create your index (the best 8, ignoring the other 12).

 

10 hours ago, DaveLeeNC said:

It is interesting to note that the old way of handling unplayed holes used to be based on shooting your index, not your average.

 

No, not necessarily. It was based on shooting net par, but if the course rating was 68.3… you wouldn't arrive at your index.

 

@DaveLeeNC, please answer the question I've asked you a few times now. Here it is again:

 

23 hours ago, iacas said:

Again, do you genuinely think they didn't take hundreds of thousands if not more of existing scores from real golfers playing real rounds, calculate the indexes using the "old" method and the "new" method, before they decided to go with this new method?

 

Thank you.

[Newby]

26 minutes ago, DaveLeeNC said:

 

And thanks for posting the quote. I have read a bunch on this and it looks familiar but I don't recall where I saw it - did you save a link? 

 

dave

Sorry but I didn't. I'll see if I can find it.

 

I assume you have seen this from the USGA

 https://www.usga.org/content/dam/usga/pdf/2024-revision/Detailed-Infographic-2024-9-hole-scores.pdf

An expected score is the score a player is expected to achieve over a specified number of holes on a course of standard difficulty. It is based on the average Score Differential of players with a given Handicap Index (ie the same) and a normal distribution of scores.

[nsxguy]

33 minutes ago, DaveLeeNC said:

FWIW, mean (or average) and expected values are the same thing for typical, well-behaved distributions. 

 

And thanks for posting the quote. I have read a bunch on this and it looks familiar but I don't recall where I saw it - did you save a link? 

 

dave

 

https://www.golfdigest.com/story/world-handicap-system-changes-2024-more-reliable-timely-nine-hole-scores-short-courses

 

 

[DaveLeeNC]

@nsxguy @Newby  Got it - thanks. Quite frankly I had forgotten that they had explicitly said that they were using averages/expected values. I had gotten hung up on the process of figuring that out from the data that we had from posted results.

 

dave

[DaveLeeNC]

3 hours ago, iacas said:

Again, do you genuinely think they didn't take hundreds of thousands if not more of existing scores from real golfers playing real rounds, calculate the indexes using the "old" method and the "new" method, before they decided to go with this new method?

 

Here - Detailed-Infographic-2024-9-hole-scores.pdf (usga.org) - the USGA has said exactly what they did. They came up with the average (more explicitly the expected) differential for a given golfer's unplayed 9 holes and use that to calculate an 18 hole differential. That is what they said they did. It sounds very sensible and when 9 hole scores dominate your posting history this approach will (in aggregate) increase your index for most golfers. I believe that they knew that when they chose this approach (it is pretty obvious). And there may well be valid reasons for having made this choice anyway - I have no way of knowing. 

 

dave 

 

 

Edited June 25 by DaveLeeNC

[iacas]

15 minutes ago, DaveLeeNC said:

Here - Detailed-Infographic-2024-9-hole-scores.pdf (usga.org)

 

Thanks. We've been aware of this for months.

 

15 minutes ago, DaveLeeNC said:

I believe that they knew that when they chose this approach (it is pretty obvious). And there may well be valid reasons for having made this choice anyway - I have no way of knowing.

 

That does not answer the question I've asked you several times.

 

You seem to think they didn't take their millions of existing rounds, calculate the indexes of the golfers in their database (i.e. real golfers who played real rounds) BOTH ways to see what effect this would have.

 

Without the quotes, again:

 

Do you genuinely think the USGA didn't take hundreds of thousands or millions of existing scores from real golfers playing real rounds, calculate the indexes using the "old" method and the "new" method to compare the differences, before they decided to go with this new method?

[nsxguy]

[DaveLeeNC]

13 hours ago, iacas said:

Without the quotes, again:

 

Do you genuinely think the USGA didn't take hundreds of thousands or millions of existing scores from real golfers playing real rounds, calculate the indexes using the "old" method and the "new" method to compare the differences, before they decided to go with this new method?

 

I cannot be more clear here. I will try ONE MORE TIME. 

 

I think the USGA did what they said they did. They took "thousands or millions" of existing scores from real golfers playing real rounds" and calculated the Expected Differentials (by handicap range) and that is what they used in the new formula because that is what they decided was 'sensible'. I don't know if they did a comparative analysis of resulting indexes for 9 hole golfers or not. If they did I believe that the results would show an upward bias for 9 hole golfers. Further I believe that if they had done this and it slowed reasonably convincing results they would have clearly stated that they have done this analysis. I have not seen that statement. 

 

I cannot state this any more clearly, so don't ask (again). I will be happy to clarify although I don't know why anyone would be interested in delving into my opinion of what the USGA did or did not do.

 

dave

 

ps. Using anything but something like 'expected differentials' is going to be a tough sell to the masses. Obviously I am not 'the masses" here and I don't expect my opinion to be catered to by the USGA or R&A. And as I have stated previously if I were in their shoes with their knowledge I might even agree with their decision. 

[Pnwpingi210]

13 hours ago, DaveLeeNC said:

 

Here - Detailed-Infographic-2024-9-hole-scores.pdf (usga.org) - the USGA has said exactly what they did. They came up with the average (more explicitly the expected) differential for a given golfer's unplayed 9 holes and use that to calculate an 18 hole differential. That is what they said they did. It sounds very sensible and when 9 hole scores dominate your posting history this approach will (in aggregate) increase your index for most golfers. I believe that they knew that when they chose this approach (it is pretty obvious). And there may well be valid reasons for having made this choice anyway - I have no way of knowing. 

 

dave 

 

 

Thanks for posting that.

 

 This new method produces a more consistent and comparable Handicap Index for those who post 9-hole scores.
» Prior to 2024, the order in which the 9-hole scores were combined could add volatility to the Handicap Index.
» It was also common for two good 9-hole scores to combine and produce an 18-hole Score Differential which was lower
than any of the Score Differentials based on an 18-hole score in the player’s scoring record - which resulted in a Handicap
Index that may be difficult for the player to play to.

 

I think the bolded statement says they were looking to reduce artificially low scores (two 9 hole rounds combined for an 18 that result in a lower score than the golfer was able to post on an other full 18 hole round).   

Edited June 25 by Pnwpingi210

[iacas]

40 minutes ago, DaveLeeNC said:

I cannot be more clear here. I will try ONE MORE TIME.

 

That's the first time you actually answered the question. In trying to answer the question, you previously said:

 

14 hours ago, DaveLeeNC said:

Here - Detailed-Infographic-2024-9-hole-scores.pdf (usga.org) - the USGA has said exactly what they did. They came up with the average (more explicitly the expected) differential for a given golfer's unplayed 9 holes and use that to calculate an 18 hole differential. That is what they said they did.

 

That PDF says nothing about whether they used millions of real rounds from real golfers to calculate their indexes both the old and the new way, which is what the question asks.

 

While I appreciate that you were frustrated enough to yell, the seven others I asked to read this discussion all said the same thing: "he hasn't answered the question yet."

 

40 minutes ago, DaveLeeNC said:

They took "thousands or millions" of existing scores from real golfers playing real rounds" and calculated the Expected Differentials (by handicap range) and that is what they used in the new formula because that is what they decided was 'sensible'.

 

Good. So now my next question is: what do you think they saw as a result for 90 to 95% of golfers in comparing their indexes calculated under the "old" method and the "new" method?

 

40 minutes ago, DaveLeeNC said:

Further I believe that if they had done this and it slowed reasonably convincing results they would have clearly stated that they have done this analysis. I have not seen that statement.

 

OMG. 😳 Of course they did this (compare real results with both methods). They have a database of millions of hole scores. They're the sole handicapping body for the US. Of course they did this, test the new method against the old… to see what happened.

 

Another question: do you think they'd have pushed this change through if a good chunk of handicaps rose by a good amount when golfers shot the same scores? I'll answer it for you: of course not.

 

Too many people are freaking out and over-reacting to this without having any actual real-world data.

 

On 6/23/2024 at 5:56 PM, DaveLeeNC said:

I believe that we will see another 9 hole posting procedure in the next year or two once this approach settles in and folks see for themselves how this works.

 

How much would you be willing to wager on that one?

[DaveLeeNC]

@iacas  No further comment

[sui generis]

About time this was put to bed? 😉

[st1800e]

Maybe I missed it , but is the treatment/algorithim of not playing,  say, one, two, three holes etc. the same as  having only played nine holes? 

[davep043]

56 minutes ago, st1800e said:

Maybe I missed it , but is the treatment/algorithim of not playing,  say, one, two, three holes etc. the same as  having only played nine holes? 

Its similar, from the few results I've seen.  Simplifying a lot, a player with a HI of 15 would have an "expected differential" for 9 holes somewhere around 9.0 (1/2*HI+1.5).  His "expected differential" for 3 holes not played would be about 3.0.  

[DaveLeeNC]

22 minutes ago, davep043 said:

Its similar, from the few results I've seen.  Simplifying a lot, a player with a HI of 15 would have an "expected differential" for 9 holes somewhere around 9.0 (1/2*HI+1.5).  His "expected differential" for 3 holes not played would be about 3.0.  

To expand on this all of the statements that I have seen refer to (in some form) "expected differentials" for unplayed holes or unplayed sides. So I would also expect the results for (as an example) the differential for playing/posting holes 1-9 vs. 1-10 to differ by a very small amount.

 

In principle the processes here could be identical as the 'expected score' (based on CR/Slope) is generated on a by hole basis. But I doubt that data is available to the GHIN software. 

 

dave

Edited Saturday at 01:52 PM by DaveLeeNC

[Newby]

32 minutes ago, DaveLeeNC said:

 

 

In principle the processes here could be identical as the 'expected score' (based on CR/Slope) is generated on a by hole basis. 

 

dave

I have seen somewhere that the USGA has said they are the same. (ie for each unplayed hole)

Edited Saturday at 02:24 PM by Newby