What flaws do you think there are in the handicap system?

[iacas]

Kevin, this is my last response on this, and your last response to me.

4 hours ago, natureboy said:

Slope is all about portability. Some players who play home courses that have tougher that average obstacles for bogey golfers may get a bump more correctly reflecting their skill relative to the general population of bogey golfers. And a few players primarily playing very easy courses will get dinged slightly relative to most golfers at their HCP who play average difficulty courses.

Slope is not really about portability, no. Yes, the documentation may say that, but that's simply because of how it came about. It's about accurately measuring the difficulty of a golf course for players who aren't scratch golfers. They had course ratings, but found that a 74 course and a 72 course might not play two shots or a proportionate number of shots differently for a 23 handicapper. Maybe the 74 was wide open and the 72 was tight, so the 72 played as difficult as the 74 (or more so) for the 23 guy.

It's simply about attempting to somewhat accurately - while maintaining the simplicity of just having two numbers - determine the difficulty along a wide range of abilities. They only say "portability" because you just divide and multiply the slope to find your course handicap.

4 hours ago, natureboy said:

The average Index number for each handicap won't shift, because the average players make up the bulk at any given HCP and adjustments get made both above and below the mean. It will just increase the accuracy and therefore portability for some golfers whose home courses mask slightly better skill and some whose home courses slightly exaggerate their skill. I expect making the corrections would even slightly address a built-in opportunity for a savvy sandbagger.

Oh brother.

Look, so far as I can tell, your entire argument is this: the "bogey golfer" has a larger shot dispersion pattern than the USGA says. You're using one source of data for this and giving no credit to the possibility that the USGA has continued to monitor their own standards and data.

You've given the example that trees 50+ yards (IIRC) from the fairway need to be considered. That would only increase the slope rating, and thus drop the handicaps of ALL players. Remember, only two points define the line. Whatever change you make will, yes, "increase the accuracy" of the system for some players. It will also decrease the accuracy for others.

And here's the other thing you've yet to demonstrate… that there's a problem that needs fixing. I don't know of any bunch of 20 handicap players out there bemoaning that when they play their friend's course, they get creamed all the time because he has thick trees 55+ yards away from the fairway and your course does not.

The other thing you've failed to consider: handicap is more a measure of potential, not your average. If you hit it 50+ yards offline in one round, you're probably not counting that round among your best 10. There will be rounds where you hit it better, and those are the ones that end up being counted.

4 hours ago, natureboy said:

The chart is labeled "Accuracy Table" - Men and [Women] (Dimensions of Expected Landing Area 2/3 of the Time - in Yards). The area between one standard deviation above and below the mean has a cumulative probability of ~ 68% of the expected results, which is basically equivalent to the USGA 2/3 = 67%.

I know what one standard deviation is, thank you…

4 hours ago, natureboy said:

Broadie's 'Degrees Offline' is the number of degrees representing one standard deviation of shot dispersion off centerline. To convert it to a width you just use a little trig for the lateral deviation offline at the expected average distance and double it (to reflect deviation above and below the mean / left & right of center) to get an expected landing area width. You can do the inverse to convert the USGA landing area into a right triangle degrees off line. They are fundamentally measuring the same thing, one with an angle, one with a width.

They're not fundamentally measuring the same thing. You don't get that. One measures actual performance, and the other measures what we should care about. Both exclude things outside a standard deviation because… you have to draw the line somewhere.

At 200 yards, the USGA will consider anything within 33 yards to be a significant factor for a bogey golfer. That's 9.4° accuracy, or 4.7° on both sides of center. They get 37 yards of "depth" to consider something a significant factor, too, but that's beside the point.

Look, as plain as I can say it, you're not evaluating the course rating system properly if you think we don't consider things farther away than 16.5 yards away from the center of the fairway (the worst golfers at the longest range we consider). Or 20.5 yards (scratch golfers get 41 yards width at 250 yards). We absolutely consider things outside of 16.5 yards away, or 20.5 yards away. They're simply not significant factors. They're considered, but not significantly.

On March 11, 2016 at 0:22 PM, natureboy said:

That means roughly 1 in 5 drives for a 12 HCP player would end up outside the 60 yard wide 'landing zone' (30 yards on either side of the hole center line) outside of which the USGA doesn't take trees into account for rating purposes.

That's as wrong now as it was when you typed it. Never mind the fact that we don't consider "a 12 Handicap player" per se. As you know, it's two points that make up a line. I told you then: "Plus, trees > 30 yards away from the centerline aren't ignored, they're just very minimally weighted."

4 hours ago, natureboy said:

That's a straw man argument. Just like when you build your shot zones you ignore the extreme outliers and focus on the expected average around which most of the values fall.

I don't think you know what a straw man argument is. This is your argument.

You're arguing for the consideration of huge areas of land to be considered and weighted. It's your argument, buddy.

4 hours ago, natureboy said:

The issue is not that the USGA isn't properly looking at all possible shots by all possible golfers. The shortcoming in the model is that the 'expected landing area' specifically chosen to reflect where 67% of the bogey golfer's shots are expected to fall does not accurately reflect the real-world average bogey golfer population. It understates the expected landing area width by ~ 50% off the tee. Even for a 90 shooter on a 'typical course' - which would be equivalent to a 16 HCP - the USGA width is too narrow by ~ 8%.

You've continually missed two points. Maybe if I put them in a bullet list you'll see them. Then, when this comes up in another conversation, you can post in that one, too:

• Trees outside of the defined areas are considered. They're simply not significant factors.
• If you considered more heavily trees, rough, hazards, etc. well beyond the bounds of what the USGA normally considers, slopes will simply increase. "There aren't any trees on this hole at all, but 50 yards away, there are, and the bogey golfer is in them 20% of the time…" will only increase the slope, thus reducing their index.

For the last time… we do consider trees and obstacles outside the "accuracy table."

4 hours ago, natureboy said:

That's interesting. I haven't yet seen a large data set with that kind of trend. Is that from Shots To Hole?

Ah, all's not as it's cracked up to be, eh Kevin?  Shorter hitters are often more accurate if you measure by things like "fairways hit."

Old guys hit a ton of fairways… because they don't swing fast enough to hit it too far offline nor to put much curve on the ball.

4 hours ago, natureboy said:

I'm only considering the parameter or expected shot envelope the USGA deliberately chose. The choice of a landing area that would contain 2/3 of expected shots certainly seems to have been intentional in capturing the shot probability between one standard deviation above and below the mean.

If the USGA was focused strictly on a width measurement, why reference the expectation of it containing 2/3 of the shots or any fraction of shots at all?

Because it's the area in which obstacles are considered significant. They're weighted for obstacles outside that area, but on a sliding scale. Is the sliding scale codified? No, because everyone understands it pretty well and it's a bit of a "feel" thing anyway. That's why ratings teams are made of teams. I think some clump of trees are a 5 and another thinks they're a 7 and we discuss. We don't consider the tree limbs that only enter the "landing zone" defined in the book. They're all considered. If the trees in the landing zone are a 4 but they get really thick outside, they end up a 5 or a 6.

I've said variations of this now several times, including dating back to the blog entry where you originally started this whole line of discussion.

4 hours ago, natureboy said:

Your 250 yards example is the same straw man argument. We are talking about one standard deviation above and below the average for the large general population of bogey golfers, not the rare shot or even the average for an individual. That's an extremely rare outlier. Even typical 113 shooters keep almost all shots within ~ +/- 25 degrees offline and 100 yards +/- centerline. You discard consideration of that low likelihood just like when you build your shot zones. 

No it isn't. It's still your example. It's simply, in this case, a bit of hyperbole… in the hopes that you would begin to understand how ridiculous you continue to be on this. We already do consider obstacles outside the accuracy table.

Your entire argument is that "the numbers don't match up." Cool.

So I'll end this discussion with you again with something I said  about a month ago:

On March 20, 2016 at 10:28 PM, iacas said:

The point that you seem to be missing is that you haven't proven or even really begun to demonstrate knowledge of how the rating/slope accounts for this now or begun to make a case as to why it should be changed.

Cheers.

[iacas]

@natureboy, though I'm not at all resuming the conversation, I just thought I'd let you know that in 2016-2017, we're only considering shots within 50 yards of the centerline of the fairway. This is down from 60 yards.

The entire method for calculating how penal "trees" are on a hole was re-done, too, to match more closely the way we've always done it locally: to look at the trees and assign a rating to them. The old way involved a whole lot more measuring.

It also now dawns on me that you think the 60 yards (which it used to be) was the diameter. It was the radius. We accounted for all trees within 60 yards of the center of the fairway, or an area 120 yards across. Now it's "only" 100 yards across.

I apologize for overlooking that. I didn't think someone would make that type of mistake, but I think now that you did. I stand by my comments about how we give trees to the outer edges less weight, because that's still true. A pine tree five feet off the fairway is worse than a pine tree 48 yards away from the center of the fairway.

[No Mulligans]

This makes no sense (same score, easier slope, results in a better differential):

[Golfingdad]

7 minutes ago, No Mulligans said:

This makes no sense (same score, easier slope, results in a better differential):

What's the alternative?

[No Mulligans]

4 minutes ago, Golfingdad said:

What's the alternative?

Eliminate the slope adjustment when the score minus the rating is less than zero.  A simple "if, then" program/calculation change.

[DaveP043]

Maybe use the inverse of the normal slope adjustment when scores are below the CR, that is, multiply the difference by slope/113.  Greater slope will mean greater negative differential, as it should be.

[Golfingdad]

5 minutes ago, No Mulligans said:

Eliminate the slope adjustment when the score minus the rating is less than zero.  

I fully get what you're saying about the comparable differentials, but just from a math perspective, I don't think that makes a whole lot of sense either.  Like I mentioned in the other thread - the "zero" line is random and arbitrary, so instead of changing the formula when it crosses an imaginary barrier, maybe you could just take the rating and lower it by 15 or something and base everything off that.

[No Mulligans]

4 minutes ago, Golfingdad said:

I fully get what you're saying about the comparable differentials, but just from a math perspective, I don't think that makes a whole lot of sense either.  Like I mentioned in the other thread - the "zero" line is random and arbitrary, so instead of changing the formula when it crosses an imaginary barrier, maybe you could just take the rating and lower it by 15 or something and base everything off that.

Something like that would fix it too.  Add 15 to the score, do a calc and then move the result back down by 15.  The formula would have to be tweaked a bit.  This adds some complexity that might be hard to explain/understand/remember by many.

[Golfingdad]

1 minute ago, No Mulligans said:

Something like that would fix it too.  Add 15 to the score, do a calc and then move the result back down by 15.  The formula would have to be tweaked a bit.  This adds some complexity that might be hard to explain/understand/remember by many.

Also, I'm not sure it's worth it.  I wanted to show @DaveP043 that his idea wouldn't make sense either, so I did the math on a couple of made up scores right on either side of the course rating of an imaginary 72.5/150 course.  But I found that it actually DOES make just as much sense as our other ideas beucase the difference in any of the methods is really, really minor.  Using Dave's idea, a 73 is a 0.38 and a 72 is a +0.66.  Using yours, it's 0.38 to +0.5, and using mine it's 11.68 to 10.92.  So the delta is 1, 0.88 and 0.76.

In reality, people below scratch aren't playing a lot of net games, so it's probably something that doesn't really matter to enough people anyways. 

[No Mulligans]

If we consider the definition of slope:

From wikipedia:

Quote

The slope rating of a golf course is a measure of its difficulty for bogey golfers. The term comes from the likelihood that when playing on more difficult courses, players' scores will rise more quickly than their handicaps would predict.

And consider as a score less rating approaches zero, the slope adjustment has little to no affect on the differential.

Then, imo, it seems consistent to just eliminate the slope calculation for scores less rating below zero. 

Yes it affects very few players, but why not?

[saevel25]

 

2 minutes ago, Golfingdad said:

In reality, people below scratch aren't playing a lot of net games, so it's probably something that doesn't really matter to enough people anyways. 

In most cases it's rare to see a sub 120 slope with a rating of 72. I looked at Ohio's golf courses, http://www.ohiogolfguide.com/page.php?page=courselist.php&rtype=select&selecttype=Public&selectholes=18 There is only two courses, with a rating within 0.2 of 72, that have a slope sub 120. If you take the non-outlier courses out, we are talking about 0.4 difference in differential between a guy who plays a 72/124 and a 72/137. 

I don't think it's that big of a deal to worry about. Though I get the logic that why should a guy playing a harder course, with regard to slope, have a worse differential? 

Maybe there needs to be an adjustment factor associated with handicap that takes into account that slope is less influencing on better players.

[iacas]

3 hours ago, No Mulligans said:

This makes no sense (same score, easier slope, results in a better differential):

It makes perfect sense.

A golf course plays more difficult for worse golfers. A 0 is a worse golfer than a +6.

 

The slope of the line doesn't change.

[Hardspoon]

4 hours ago, No Mulligans said:

This makes no sense (same score, easier slope, results in a better differential):

Even though @iacas just beat me to this, I'll elaborate anyway...basic difference is that lower slope is not "easier" slope, it's "flatter"...in other words, "allows scoring that is more similar to a scratch golfer".

So, on the 100-slope course, a +5 is expected to score closer to scratch, which in his case, would be worse...by shooting the same score, he performed better relative to that expectation.

3 hours ago, No Mulligans said:

If we consider the definition of slope:

And consider as a score less rating approaches zero, the slope adjustment has little to no affect on the differential.

Also important: that Wikipedia definition is not really correct...it is not only related to bogey golfers.  The USGA definition:

A Slope Rating is the USGA® mark that indicates the measurement of the relative playing difficulty of a course for players who are not scratch golfers, compared to scratch golfers.

So, as the score less rating gets FURTHER from zero (on the other side), the slope adjustment once again has a greater effect.

[SG11118]

The handicap system has a massive database of continually posted scores from golfers with known handicaps.  Why couldn't this database of scores be utilized with statistics to more accurately and cost effectively rate golf courses?

Is it simply that the golf industry doesn't want to put a bunch of course raters out of work? 

[Groucho Valentine]

In would like to see the handicap system take into account course conditions that can make a course play more difficult or easy on a particular day. But i cant see how you could include it into the system. It seems like it would be too complicated. 

[DaveP043]

40 minutes ago, Groucho Valentine said:

In would like to see the handicap system take into account course conditions that can make a course play more difficult or easy on a particular day. But i cant see how you could include it into the system. It seems like it would be too complicated. 

Under the CONGU system used in England and a few other countries, they DO use a Competition Scratch Score as the baseline.  As I understand it, the CSS is based on the actual scores recorded for a specific competition, as compared against the players' handicaps.  Its a fairly complicated process that gets done after all scores are recorded for a specific competition, but its intended to take into account the weather and course set-up and other variables that may make the course tougher (or easier) than "normal".   This is a link to an article that discusses the use of CSS, and some of the issues with it:

http://www.golf-monthly.co.uk/features/the-game/is-competition-scratch-score-css-fair-80093

We've talked about the eventual unification of handicap systems worldwide sometime relatively soon, so its realistic to wonder if something like the CSS is in the cards for US golfers.

[Groucho Valentine]

19 minutes ago, DaveP043 said:

Under the CONGU system used in England and a few other countries, they DO use a Competition Scratch Score as the baseline.  As I understand it, the CSS is based on the actual scores recorded for a specific competition, as compared against the players' handicaps.  Its a fairly complicated process that gets done after all scores are recorded for a specific competition, but its intended to take into account the weather and course set-up and other variables that may make the course tougher (or easier) than "normal".   This is a link to an article that discusses the use of CSS, and some of the issues with it:

http://www.golf-monthly.co.uk/features/the-game/is-competition-scratch-score-css-fair-80093

We've talked about the eventual unification of handicap systems worldwide sometime relatively soon, so its realistic to wonder if something like the CSS is in the cards for US golfers.

Thats interesting. It also made my head hurt...

[DaveP043]

7 minutes ago, Groucho Valentine said:

Thats interesting. It also made my head hurt...

I agree!  In researching stuff for this website, I've read through a bit of the CONGU Handicap Manual, and it makes the USGA manual seem crystal-clear by comparison.  But some of the things that are done under CONGU, like counting only competition scores (with very limited exceptions) and taking day-to-day difficulty into account do make sense to me.