CSS Farce

[Backache]

It occurred to me that we have a few days on which both men's and women's competitions take place so a comparison of each CSS might be informative. If CSS was a fair and accurate representation of difficulty then surely the difference between SSS and CSS for both men and women would tend to match?

From the results still accessible to me on the website, there were 8 such days in the last year. This image shows the analysis, I haven't drawn any conclusions from it but it seems things worked out the same only 3 times out of the 8. Interesting but fairly inconclusive without a much larger sample set, I think.

Hmm In no cases did it move in an opposite direction and in no cases was the womens CSS adjusted downwards which you seem to be worried about.

 

[FairwayDodger]

Hmm In no cases did it move in an opposite direction and in no cases was the womens CSS adjusted downwards which you seem to be worried about.

Indeed, although it's a three shot difference as often as it stays the same. Not enough comps for a serious comparison though.

Obviously the OP was because I was peeved at what happened on Monday and it's more annoying when it goes down than up but I think the discussion has moved into a more interesting debate about the accuracy or otherwise of CSS.

I thought there might be more of an an obvious correlation between CSS change and number of entries but there doesn't seem to be. The most interesting result is the day where almost the same number of players played in the men's and women's comps but one CSS didn't move and the other went reductions only.

 

[Backache]

Indeed, although it's a three shot difference as often as it stays the same.

To be pedantic it was the same three times , a three shot difference (for reductions only otherwise the same twice), a two shot difference twice, and a one shot difference once.
At no point would you have been disadvantaged by the CSS for playing better than SSS by the CSS.

 

[FairwayDodger]

To be pedantic it was the same three times , a three shot difference (for reductions only otherwise the same twice), a two shot difference twice, and a one shot difference once.
At no point would you have been disadvantaged by the CSS for playing better than SSS by the CSS.

You are correct although I didn't put those figures up to argue about them, just to try and give a comparison of how CSS can produce different results under the same conditions where the only variable factor is the human one - different players performing differently.

Remember - this is only 8 comps played over a period of a year, a small subset. And don't get too hung up on whether CSS goes up or down, I've got over Monday's disappointment and am only debating the accuracy or otherwise of the calculation.

 

[FairwayDodger]

I've been reading up on the CSS calculation a bit and have learned a couple of things.

1) It's not playing to buffer, but rather to within 2 shots of SSS, so that mitigates the impact of cat 3 and 4 variability slightly.
2) At my club the relative lack of cat 1 and cat 2 players means that the chance of CSS going down is much larger (if I'm reading the CSS table correctly) and, as the only cat 1 player my presence is usually ignored (cat 1s rounded to 0% of the field) for that lookup.

The more we discuss this, and the more I read, the more convinced I become that the process just doesn't meet its objective.

 

[Backache]

You are correct although I didn't put those figures up to argue about them, just to try and give a comparison of how CSS can produce different results under the same conditions where the only variable factor is the human one - different players performing differently.

Remember - this is only a 8 comps played over a period of a year, a small subset. And don't get too hung up on whether CSS goes up or down, I've got over Monday's disappointment and am only debating the accuracy or otherwise of the calculation.

This is golf apparently identical shots can produce very different outcomes dependant on a fraction of a difference of where a ball lands. If you play enough qualifiers the vagaries should even out. The important question is, should we attempt to make allowances for conditions when calculating handicaps? I believe it is sensible and CSS is about as objective as it comes, though imperfect and on any one individual day will be subject to inconsistencies.

 

[FairwayDodger]

This is golf apparently identical shots can produce very different outcomes dependant on a fraction of a difference of where a ball lands. If you play enough qualifiers the vagaries should even out. The important question is, should we attempt to make allowances for conditions when calculating handicaps? I believe it is sensible and CSS is about as objective as it comes, though imperfect and on any one individual day will be subject to inconsistencies.

Which is where we disagree - I think CSS is a lottery and have so far seen nothing to convince me otherwise.

 

[Backache]

Which is where we disagree - I think CSS is a lottery and have so far seen nothing to convince me otherwise.

The problem is when things are subject to random variation people tend to remember the things that work against them or at least support a certain hypothesis. The actual figures published both by yourself and Congu suggests that though there is random variation it is not a lottery and CSS generally goes up in difficult conditions and comes down in easier conditions though the exceptions are more memorable and stick in the mind.

 

[FairwayDodger]

The problem is when things are subject to random variation people tend to remember the things that work against them or at least support a certain hypothesis. The actual figures published both by yourself and Congu suggests that though there is random variation it is not a lottery and CSS generally goes up in difficult conditions and comes down in easier conditions though the exceptions are more memorable and stick in the mind.

Well, you can claim CONGU's figures prove your hypothesis if you like, I know they make that claim. But I've already highlighted the fundamental flaws in these which are that they are only based on men's competitions (so no cat 4 and quite probably a different split of players across handicap categories) and make no reference to what the actual conditions were.

My own figures are just the only ones I have available for the comparison I was trying to make. I'm making no great claims that they prove my case but they absolutely do not show what you purport they do. Sadly I didn't play in all those comps and couldn't accurately remember the conditions even if I did.

Even if there was some trend towards CSS going up in difficult conditions and down in easy ones it would still be the anomalous results that should trigger further refinements in the algorithm.

What I would really like to see is a longer term analysis of men's and women's comps on the same course on the same day with the breakdown of #players in each handicap category and a description of ground, green and weather conditions and difficulty of pin placements on the day.

 

[DCB]

Got to empathise with Fairwaydodger here. My wife has similar concerns at times when the CSS seems unrealistic given the scores returned. It can be a lottery depending on when you play in a Ladies monthly medal. Often the main medal day can have a CSS 1 or 2 higher than the Alt medal day, purely down to the make up of the field and the scores they return. No easy answer to it though.
Some of the better events are held on a single day, no Alt day. This then gives a larger field and appears to be a more realistic CSS that is calculated as a result. If you look at the high handicap ladies scores though it can be really scary !

 

[duncan mackie]

I've been reading up on the CSS calculation a bit and have learned a couple of things.

1) It's not playing to buffer, but rather to within 2 shots of SSS, so that mitigates the impact of cat 3 and 4 variability slightly.
2) At my club the relative lack of cat 1 and cat 2 players means that the chance of CSS going down is much larger (if I'm reading the CSS table correctly) and, as the only cat 1 player my presence is usually ignored (cat 1s rounded to 0% of the field) for that lookup.

The more we discuss this, and the more I read, the more convinced I become that the process just doesn't meet its objective.

1. Where is this set out?
2. You wouldn't be ignored! It's actually a safety situation for exactly what you have presented. The matrix tightens the parameters as the proportion of lower handicaps in the field increases. If the majority of the field is Cat 4 with a lone cat 1 it wouldn't be appropriate to have the parameters for cat 1 and give increased significance to scores. In a field of 16 eligible players you will get fully (over) represented at 10%.

edit - here's the relevant section to your point 1 from the SGU CONGU manual 2012-2015

Enter in Box E the number of such competitors in Categories 1 , 2, 3 [and 4] who have returned
nett scores, including those corrected under the provision of Clause 17.1 (c), in their Handicap
Category Buffer Zone and better before the application of Clause 19.

17.1c deals with dq and eligible scores for handicapping. 19 with stableford adjustments.

 

[FairwayDodger]

1. Where is this set out?
2. You wouldn't be ignored! It's actually a safety situation for exactly what you have presented. The matrix tightens the parameters as the proportion of lower handicaps in the field increases. If the majority of the field is Cat 4 with a lone cat 1 it wouldn't be appropriate to have the parameters for cat 1 and give increased significance to scores. In a field of 16 eligible players you will get fully (over) represented at 10%.

1. Based on this link, Duncan, maybe I made the mistake of going for a simplified description rather than CONGU. Is it wrong?

http://www.dallingtongolf.com/pdf/calculation-of-css.pdf

2. Yes and agreed, but in a field of more than 20 (as most are) I'm not represented at all at 0%.

I think it's the problem of trying to make a statistical process that sort of (maybe) works for men's competitions stretch to a completely different demographic that plays in women's competitions.

I wonder if CSS in women's comps has been subjected to the same level of scrutiny as it has in the men's game?

 

[FairwayDodger]

edit - here's the relevant section to your point 1 from the SGU CONGU manual 2012-2015

Enter in Box E the number of such competitors in Categories 1 , 2, 3 [and 4] who have returned
nett scores, including those corrected under the provision of Clause 17.1 (c), in their Handicap
Category Buffer Zone and better before the application of Clause 19.

17.1c deals with dq and eligible scores for handicapping. 19 with stableford adjustments.

Just saw your edit and that backs up the way I originally thought it works, so much for my attempt at being conciliatory!

 

[FairwayDodger]

Can I just reiterate that this isn't sour grapes, I'm genuinely interested in understanding why this does or doesn't work.

I don't need to go through my old scores, as Louise did, to know that my handicap would be a fair bit higher if all my comps had been measured to SSS but I still think that would be fairer.

I do like the concept of allowing some leeway for slightly higher scores in poor conditions and I don't know how you would achieve that in a non-subjective manner without reference to other players performance or adding a huge level of complexity such as by referring out to Met Office data, for example.

What I don't like is that there's just too much human variability in the current system. Especially in smaller fields with a high proportion of cat 3 and 4 players.

All in my opinion, of course!

 

[Matty2803]

Just a quick question - How is the SSS created/calculated?

 

[DCB]

Just a quick question - How is the SSS created/calculated?

SSS is created when the course is rated by the local Union. If the course changes significantly this may well change the SSS at the next rating. It isn't a once and for all rating.

 

[duncan mackie]

1. Based on this link, Duncan, maybe I made the mistake of going for a simplified description rather than CONGU. Is it wrong?

http://www.dallingtongolf.com/pdf/calculation-of-css.pdf

2. Yes and agreed, but in a field of more than 20 (as most are) I'm not represented at all at 0%.

I think it's the problem of trying to make a statistical process that sort of (maybe) works for men's competitions stretch to a completely different demographic that plays in women's competitions.

I wonder if CSS in women's comps has been subjected to the same level of scrutiny as it has in the men's game?

1. It's a simplification of an out of date version of the system (including the table which has changed)

2. You are no more ignored than anyone else - you would be added to cat 2 numbers and, if that total still didn't make 10%, added to the next etc. I tried to explain why that was the safer route than any percentage being rounded to 10% in my previous post. I practice it might increase the lower parameter by 1 and the upper by 2 ie 18-36 is the 0/30/70 range for CSS=SSS whilst 19-38 is the 10/20/70 (assume 70 represents cat 3+4).

Statistically it makes sense - the main issue is that sometimes it's impossible to reconcile the apparent difficulty of the course and conditions with the performance of a field (and you make some good points about the difference in impact of certain elements on different handicaps).

As to how much thought went into the consideration of the ladies relevant to the men - I think the various changes and enhancements implemented clearly show that it is a lot - some might even suggest it's disproportionate in favour of the ladies but I believe every section should have the necessary effort expended rather than any arbitrary, or other, distribution of labour!

 

[Matty2803]

SSS is created when the course is rated by the local Union. If the course changes significantly this may well change the SSS at the next rating. It isn't a once and for all rating.

Thanks. How do the local unions rate the courses?

 

[FairwayDodger]

2. You are no more ignored than anyone else - you would be added to cat 2 numbers and, if that total still didn't make 10%, added to the next etc. I tried to explain why that was the safer route than any percentage being rounded to 10% in my previous post. I practice it might increase the lower parameter by 1 and the upper by 2 ie 18-36 is the 0/30/70 range for CSS=SSS whilst 19-38 is the 10/20/70 (assume 70 represents cat 3+4).

Ah, thanks for that, I'd missed that I'd get added to the cat 2 %age and that makes sense.

Looks like I'll need to bite the bullet and read CONGU if I want to get into the nitty-gritty but I'm still of the opinion that it doesn't work very well, especially for women's competitions at club level.

Of course, I reserve the right to be completely wrong and come back with a new opinion in the future!

 

[duncan mackie]

Thanks. How do the local unions rate the courses?

They use rating manuals and sheets.

Factors include length, hazards, fairway size, nature of rough , size and slopes and speeds of the greens etc etc

He USGA rating system is now being used and all courses are being re rated over the next 5 years if they haven't been recently rated to it. The manual is online in the handicapping section of the USGA website.