Round ratings for split tournament...

[ScottyLove]

j
You probably don't/can't know this exactly, but how many PDGA "regulars" were there (those are the most likely propagators)?

It's easy to see the propagators... the are in BOLD in the results.

 

[jeverett]

It's easy to see the propagators... the are in BOLD in the results.

Is this definite? I suspected that, but hadn't ever seen it confirmed.

 

[1978]

Say you have a course that has little scoring separation, poorly designed holes I guess. Wouldnt having a higher field average = a much better rating for similar scores across different pools?

What am I missing about the example below?
Field A - Average Rating 1000 average score = 50
Field B - Average rating 950 average score = 52

Field A score 50 = 1000 rated
Field B score 52 = 950 rated the 50 would only be like 980 rated.

So can the design of a course + difference in average ratings make it so different fields can create a different SSA just based on their ratings.

 

[Cgkdisc]

Course design affects the standard deviation of the scores not necessarily the SSA produced by propagators of different skill levels.

 

[jeverett]

What am I missing about the example below?
Field A - Average Rating 1000 average score = 50
Field B - Average rating 950 average score = 52

Field A score 50 = 1000 rated
Field B score 52 = 950 rated the 50 would only be like 980 rated.

Hi 1978,

The missing component there is the relationship between player (initial) rating and rating-points-per-throw. The PDGA uses two linear equations to match SSA up with an (arbitrary) associated rating-points-per-throw. In order for the difference between a 950 rated player shooting an 'average' 52 and a 1000 rated player shooting an 'average' 50 means a rating-points-per-throw of 25.. much higher than either the PDGA linear compression formula or the observed slope of initial rating vs. scoring spread. Realistically, the value is typically in the 5-15 points-per-throw range, both using the PDGA compression formulas or the slope of initial rating vs. scoring spread. Does that help at all?

 

[jeverett]

Just to add a little to the above, here's the 'compression' formula that would apply:

For SSA's below 50.3289725:
rating_points_per_throw = -0.487095 * SSA + 34.5734

So, if the SSA after field A plays is determined to be 50, the rating_points_per_throw increment would be 10.21865. So a 52 would be rated at 979.5627 (rounded to 980). So your field B would be vastly more likely to average ~55 than 52. (actually 54.89301424356).

 

[1978]

Just to add a little to the above, here's the 'compression' formula that would apply:

For SSA's below 50.3289725:
rating_points_per_throw = -0.487095 * SSA + 34.5734

So, if the SSA after field A plays is determined to be 50, the rating_points_per_throw increment would be 10.21865. So a 52 would be rated at 979.5627 (rounded to 980). So your field B would be vastly more likely to average ~55 than 52. (actually 54.89301424356).

Can you explain the 2 tournaments played 2 weeks apart, same exact layout and conditions, basically. Am vs pro

Azalea Am
R1-R4 Score of 60
Average player rating 895, average propigator rating 904
977 980 974 982 (average for a 60, 978)

Azalea Open
R1-R4 Score of 60
Average player rating 963, average propigator rating 963
989 988 994 993 (average for a 60, 991) (13 pts higher)

How is this not mostly because of the higher rated players in the open tournament padding the ratings of the rounds.

 

[Lefty23899]

Can you explain the 2 tournaments played 2 weeks apart, same exact layout and conditions, basically. Am vs pro

Azalea Am
R1-R4 Score of 60
Average player rating 895, average propigator rating 904
977 980 974 982 (average for a 60, 978)

Azalea Open
R1-R4 Score of 60
Average player rating 963, average propigator rating 963
989 988 994 993 (average for a 60, 991) (13 pts higher)

How is this not mostly because of the higher rated players in the open tournament padding the ratings of the rounds.

I've seen the exact same thing on several occasions

My favorite example is from this tournament:
http://www.pdga.com/tournament_results/16071

Round 3: Pro and Adv played with extremely tight OBs and 5 longer tees, INT and lower played on the same course at the exact same time without the OBs and from the regular tees. An INTs 72 was rated higher from the easier layout than the 72s from the tougher layout.

I wish I could find the response when I asked CK about it. It was actually pretty comical, something about having to be more focused to get 2s with the added OB, thus making it easier :doh:

 

[jeverett]

Can you explain the 2 tournaments played 2 weeks apart, same exact layout and conditions, basically. Am vs pro

Azalea Am
R1-R4 Score of 60
Average player rating 895, average propigator rating 904
977 980 974 982 (average for a 60, 978)

Azalea Open
R1-R4 Score of 60
Average player rating 963, average propigator rating 963
989 988 994 993 (average for a 60, 991) (13 pts higher)

How is this not mostly because of the higher rated players in the open tournament padding the ratings of the rounds.

Hi 1978,

Hmm.. do you happen to have the initial ratings of all propagators/players as well as their scores for each round? Without actually graphing it all, it's tough to really see what was happening with ratings/round scores. I watched footage of that event, and the course looked like it had a fair number of really short but really technical holes, and then some longer (but still really technical) holes too. Given the better angular accuracy of gold-level players, I admit that particular 13-point gap looks odd. If anything, I'd expect it to be in the other direction (the Open field 60 would rate lower). How many players were in each event, by the way? If you can get me initial ratings and round scores, I'd be happy to make some charts and run the regression.. it may help explain it.

Edit: running a Pearson Correlation Coefficient for the rounds may prove interesting, too. Heavily wooded courses tend to have low(er) correlation coefficients (the predictive value of initial player rating vs event score).. i.e. they tend to induce a lot more randomness into round scores than less wooded courses).

 

[RogerSmith]

Can you explain the 2 tournaments played 2 weeks apart, same exact layout and conditions, basically. Am vs pro

Azalea Am
R1-R4 Score of 60
Average player rating 895, average propigator rating 904
977 980 974 982 (average for a 60, 978)

Azalea Open
R1-R4 Score of 60
Average player rating 963, average propigator rating 963
989 988 994 993 (average for a 60, 991) (13 pts higher)

How is this not mostly because of the higher rated players in the open tournament padding the ratings of the rounds.

I think I can explain this very easy - without using fancy charts, regression, or even a Pearson Correlation Coefficient: You are looking at unofficial results. Wait until the results are official, and then take another look at these events.

 

[jeverett]

I think I can explain this very easy - without using fancy charts, regression, or even a Pearson Correlation Coefficient: You are looking at unofficial results. Wait until the results are official, and then take another look at these events.

@RogerSmith,

The key difference between the example 1977 gave and one where 'official' ratings might really fix the issue is the two-week gap between the two events. This isn't a combined Pro/Am event, where between unofficial and official rounds all pools that played the same course and layout will get combined (as long as the round results are fairly similar) (thus effectively averaging the SSA). This is a 4-round Pro event, and then two weeks later a 4-round Am event on the same course. The two event dates may not even be in the same rating cutoff (I didn't check).