Round Ratings

[timg]

Some amazing DGCR members helped to come up with the Scratch Scoring Estimate (SSE) a while back so you could get an idea of what a 1000 rated player *might* get on your local course. Thanks to that, it has enabled us to now rate your rounds similar to the PDGA round ratings you get for tournaments.

The process is simple, just enter your score as you have always done and if you played all the holes at a course and have chosen a tee the rating will be immediately calculated as soon as you submit your card. Of course, if the course in question doesn't have an SSE, you skipped some holes, played a doubles/custom round, etc. then a rating can't be calculated for that particular round.

I'm hoping to expand on this in the future to give you a DGCR Player Rating and I plan on adding the rating to the sort feature in the various areas where you can view people's scores.

Special thanks to jeverett for the round rating formula!

 

[Disc Fiend]

I'm hoping to expand on this in the future to give you a DGCR Player Rating and I plan on adding the rating to the sort feature in the various areas where you can view people's scores.

^This would be awesome! Thanks for the new feature Tim, I think I am going to finally start recording my rounds on here.

 

[chrishysell]

I just added the 1024 rated round I shot Sunday morning at the LoCo and it was rated 1024. Sooo, what's different?

 

[bradharris]

Out of curiosity, has anyone compared the course SSEs to actual results from PDGA play to see how closely they match?

 

[mashnut]

I just added the 1024 rated round I shot Sunday morning at the LoCo and it was rated 1024. Sooo, what's different?

It's a system based solely on the lengths, terrain and foliage info here on this site, but it's designed to approximate SSA and the ratings you would get from the PDGA system. Looks like your example is one where the two systems are in agreement.

 

[jeverett]

It's a system based solely on the lengths, terrain and foliage info here on this site, but it's designed to approximate SSA and the ratings you would get from the PDGA system. Looks like your example is one where the two systems are in agreement.

Indeed, that's the goal of the DGCR scratch score 'estimate' formulas, and in this case it sounds like they were fantastically accurate! In general, as long as the 'course foliage' field (i.e. lightly, moderately, or heavily wooded) roughly matches how the course layout actually plays in terms of technicality, the SSE formula seems to be accurate within +/- 1.5 of a 'real' SSA.. or a round rating accurate to within +/- 15 rating points.

 

[mashnut]

Which is well within the normal variations you see in the PDGA's SSA simply due to weather and other tournament day factors. :thmbup:

 

[Disc Fiend]

I am excited for the DGCR player ratings. It could be displayed in the right hand corner with the years playing, courses played etc.

 

[jeverett]

The feature looks good! I did notice that historical rounds are getting ratings too, however the ratings calculations seem to use the current course layout rather than the historical one that was recorded with the round itself. e.g. if the current course layout is really short, but the historical round was recorded with much longer hole lengths, the round is being rated assuming the current course layout instead (and thus rating in really really low).

 

[Jax11]

This sounds great, I will be testing it soon!

 

[jeverett]

Out of curiosity, has anyone compared the course SSEs to actual results from PDGA play to see how closely they match?

Dave242 was the one who calibrated the SSE formulas to (as closely as possible) match up with 'real' SSA data. He used a very large sample of PDGA data (I remember seeing it), but of course mileage may vary. The biggest source of error is when the course foliage field doesn't really line up well with how a course actually plays (e.g. the course uses lots of OB or forced layups, making it play much tougher than the foliage value would suggest, or the specific hole parameters are more open than the overall foliage on the property itself would suggest).

 

[New013]

for courses where the pins rotate? yeah i could see that being a problem if they move a lot and that doesn't get updated. though if you know the course well enough to post a lot of scores on the course you should know to adjust.

 

[jeverett]

for courses where the pins rotate? yeah i could see that being a problem if they move a lot and that doesn't get updated. though if you know the course well enough to post a lot of scores on the course you should know to adjust.

Yes, getting accurate (as accurate as possible) values out of the round rating system is going to require users updating the course pin position values whenever they change (before you record your round). Also, as I mentioned above, currently existing scorebook round ratings aren't taking into account the hole lengths actually listed for that round, and are instead using the 'current' course layout length value.

 

[timg]

Yes, getting accurate (as accurate as possible) values out of the round rating system is going to require users updating the course pin position values whenever they change (before you record your round). Also, as I mentioned above, currently existing scorebook round ratings aren't taking into account the hole lengths actually listed for that round, and are instead using the 'current' course layout length value.

I wrote a general function for it so historical rounds might be a little out of whack. I can go back and adjust them tomorrow to base the SSA off the distances on the card.

 

[jeverett]

I wrote a general function for it so historical rounds might be a little out of whack. I can go back and adjust them tomorrow to base the SSA off the distances on the card.

Excellent. Yeah, I would guess that in most cases using the current layout to evaluate historical rounds will still provide reasonable accuracy, it's just that for the specific example course I looked at, the course layout fluctuates between a 4300ft. course and a 5800ft. course (roughly a 5 throw difference, or a whopping ~50 rating point difference)! It would be great if the round rating formula could grab the course length value associated with each historical round.

 

[Dave242]

Dave242 was the one who calibrated the SSE formulas to (as closely as possible) match up with 'real' SSA data. He used a very large sample of PDGA data (I remember seeing it), but of course mileage may vary.

Yes - I did do a bunch of analysis on probably 12-15 courses where I knew the exact layout, knew the course (wooded lightly, moderately, and heavily), and knew the weather conditions rounds were played in. Not really that big of a sample set. I have no idea how timg implemented the formulas that I found to match the "gold standard" of PDGA Scratch Scoring Averages. There was discussion about also factoring in how hilly a course is set to be in the course page at the time he implemented them and also several other formula proposals out there. Heck, I'm not even sure what was implemented was anything I worked on or contributed to.

Thank you for providing the slope/curve for correlating SSA/SSE to a point-per-stroke value. Good stuff.

And......huge ups to timg for doing this. This is fun! It might get me to start entering my scores.

The biggest source of error is when the course foliage field doesn't really line up well with how a course actually plays (e.g. the course uses lots of OB or forced layups, making it play much tougher than the foliage value would suggest, or the specific hole parameters are more open than the overall foliage on the property itself would suggest).

Another huge area where SSE fails is for very short courses. For them, SSE is often a score that can only be reached via 1 or more aces (and deuces everywhere else). It should be noted that the PDGA does not generate SSAs for these types of courses (SSA<44 if I remember correctly).

 

[jeverett]

Another huge area where SSE fails is for very short courses. For them, SSE is often a score that can only be reached via 1 or more aces (and deuces everywhere else). It should be noted that the PDGA does not generate SSAs for these types of courses (SSA<44 if I remember correctly).

I'm pretty sure that the PDGA will allow very 'easy' courses, just not course layouts with less than 13 holes.. Chuck would know better, though. But yeah, in retrospect we maybe should have altered the length-based formulas to either use some kind of floor (e.g. the SSE can't ever be better than a birdie on every hole) or even a non-linear function. I don't think we really had enough data (or accuracy with the particular function inputs being only foliage type and course length) to do much better, though.

 

[burdphil]

Yay, a new toy! Sweet...

 

[mashnut]

I'm pretty sure that the PDGA will allow very 'easy' courses, just not course layouts with less than 13 holes.. Chuck would know better, though. But yeah, in retrospect we maybe should have altered the length-based formulas to either use some kind of floor (e.g. the SSE can't ever be better than a birdie on every hole) or even a non-linear function. I don't think we really had enough data (or accuracy with the particular function inputs being only foliage type and course length) to do much better, though.

In an extreme case, if you had a 1000 rated player play a course that was 18 holes all 100', he'd probably average better than a 36 after several rounds. I'm not sure it's a problem that the ratings come out that way, it's a fringe case but it's not totally unreasonable that a 1000 rated round might have to include an ace.

 

[timg]

I'm pretty sure that the PDGA will allow very 'easy' courses, just not course layouts with less than 13 holes.. Chuck would know better, though. But yeah, in retrospect we maybe should have altered the length-based formulas to either use some kind of floor (e.g. the SSE can't ever be better than a birdie on every hole) or even a non-linear function. I don't think we really had enough data (or accuracy with the particular function inputs being only foliage type and course length) to do much better, though.

I can just have it not do it for courses under a certain length but I don't see the harm in leaving it as is.