Calling on Ratings Guru's.

[Solbo]

I know that a while back, I had asked a person on here to calculate my rough rating based on what I had shot for a few tournies, in advance of the "official" ratings update. They did this for me and the rating ended up being off by 1-3 points.

Whom was this person? I need to talk to you about the formula, trying to create a ratings system for my regionally based disc golf club.

 

[JHBlader86]

If I did it correctly then I just take the unofficial rounds, add those, then divide by 2000, 3000, or 4000 depending on how many rounds there are, then take that #, plus my rating, divide that by 2000 and it would be my new rating if I didnt have to go by 12 months of past ratings. So, for example in a 4 round tournament...

Round 1 Rating: 956
Round 2 Rating: 937
Round 3 Rating: 982
Round 4 Ratig: 950

956+937+982+950= 3825

3825 divided 4000= 0.95625 or 956 average for the tournament.

To find my new rating, not based off past events, but this one...

956+922 (my current rating)= 1878

1878 divided by 2000= 0.939 or 939 rating.

I'm probably wrong, but thats how I've tried to figure it out.

 

[Mr. Wick]

ahhhh. NEFA ratings.

Wish I could help but math is not my strong pt.

 

[Solbo]

If I did it correctly then I just take the unofficial rounds, add those, then divide by 2000, 3000, or 4000 depending on how many rounds there are, then take that #, plus my rating, divide that by 2000 and it would be my new rating if I didnt have to go by 12 months of past ratings. So, for example in a 4 round tournament...

Round 1 Rating: 956
Round 2 Rating: 937
Round 3 Rating: 982
Round 4 Ratig: 950

956+937+982+950= 3825

3825 divided 4000= 0.95625 or 956 average for the tournament.

To find my new rating, not based off past events, but this one...

956+922 (my current rating)= 1878

1878 divided by 2000= 0.939 or 939 rating.

I'm probably wrong, but thats how I've tried to figure it out.

Sounds reasonable, now how about figuring out what a round should be rated based on how people shot?

 

[Timko]

Pretty sure it would be MDR_3000. He's very good at guessing ratings.

 

[35561]

Sounds reasonable, now how about figuring out what a round should be rated based on how people shot?

This is how I've done it myself: You take all the rated players and calculate an average score of their rounds. Then you calculate their average rating. You can for example end up with the knowledge that a 945.3 rated player would score 62.1 on the course on that round. So now you know that if you play 62 it would get you a rating of 945. The problem with this is that now you need to know what the steps are. For example is 63 then 938, 937 or what? And this naturally multiplies near the ends of the spectrum. So it's not that accurate. If someone knows a better way of doing this I'd be glad to hear. I know Chuck has the official formulas for this but I don't think his allowed to give them forward.

 

[Yehosha]

http://www.pdga.com/files/documents/RatingsGuide.pdf should help a lot. Seems like finding a course's SSA is the hard part. If you know that the rest should be fairly easy to figure out.

 

[Cgkdisc]

If you want a less expensive way to get the equivalent of PDGA Ratings, subscribe to Disc Golf United. Their handicap system uses the same process as PDGA Ratings but converted to handicaps. If DGU calculates a handicap of 6.5 for your round, it's essentially the same as a 935 round rating. It's only $9.95 a year for the service. You can also enter your own personal rounds to maintain a personal handicap for all of your rounds that's separate from the official one generated strictly from tournament or league rounds. There's even a discount for bulk memberships for big clubs. http://www.discgolfunited.com/

 

[bortimus]

If you have several rated players in your club, here's how to get an estimate of the course SSA and the ratings.

Lets say 4 rated players play the same layout.

Player A (rated 984) shoots 57
Player B (rated 950) shoots 61
Player C (rated 938) shoots 66
Player D (rated 879) shoots 71

Take the average of the ratings and set them equal to the average of the scores

(Ratings average) 937.75 = 63.75 (score average)

Let's round up the rating average to 938.

Points per stroke: OK, now we need to do some estimating. At 938 (63.75 score) we are 62 points away from 1000, which is roughly 6.2 strokes (using 10 as the base points per stroke estimate. We will correct for this in a second when we find the actual points per stroke value) Subtract those 6.2 strokes from 63.75 to find the SSA range.

63.75-6.2= 57.55 (rough estimate of SSA, will get closer in a second)

At 57.55, strokes are worth approx 8.5 points per stroke (Chuck can correct me on this)

Now, we know points per stroke (8.5) and distance from 1000 (62)

Divide 62 by 8.5 to find how many strokes to subtract from the known value we found in the beginning (a 63.75 equals 938 rated round) therefore finding the SSA.

62 (distance from 1000)/ 8.5 (points per stroke) = 7.29 (round up to 7.3)

Subtract 7.3 strokes from the known value (63.75) to find SSA.

63.75 - 7.3 = 56.45 SSA

Whew!

Now that we know the SSA and the points per stroke, we can find what each round was rated

Player A shoots 57

56.45 (SSA) - 57 = -.55

-.55 * 8.5 (points per stroke) = -4.675 (round up to -5)

1000 + (-5) = 995 rated round


Player B shoots 61

56.45- 61 = -4.55

-4.55 * 8.5 = -38.675 (round up to -39)

1000 + (-39)= 961 rated round

And so on..


Chuck,

Can you post the SSA ranged with their respective points per stroke values? Thanks

 

[Steady 26542]

Please explain how you determined the points per stroke. (8.5)

 

[bortimus]

Chuck Kennedy had referenced in an old post on the PDGA boards that at a 50.4 SSA course, strokes are worth 10 points each. They increase as the SSA goes down, and decrease as the SSA goes up. By reviewing official tournament rated rounds with the posted course statistics you can find the point per stroke values at various SSAs.

http://www.pdga.com/course-ratings-by-course

Go to any course of your liking, and find a SSA close to what your estimated SSA would be. Then look at the event and see the rated rounds. Find scores 1 stroke apart and the difference will be the points per stroke value. I just happen to use VA courses because I live there. I used Loriella Park round 3 from the 2004 Old Dominion Showdown because the SSA is at 56.65, which is close to where the SSA would fall. Sometimes the points per stroke were 9, sometimes 8. That means it falls at 8.5. Sometimes the numbers are exact, 10 pts, 8 pts, 7 pts whatever. But sometimes 2 scores 1 stroke apart will have ratings 9 points apart, then you find 2 other scores that are 1 stroke apart either higher or lower and they are 8 points apart. This means the points per stroke is 8.5

 

[Steady 26542]

Thanks. I thought you were actually calculating the "points per shot" (PPS) and I was curious to know your formula. I understand how the PPS works in relationship to an SSA of 50 and how it shrinks or explands depending on which side of 50 it's on. Your calculations are decent and within a few points of what the PDGA would have rated them. As always, the more data points the better the data. Well done.

 

[bortimus]

Thanks. Definitely the more scores to use the better.

 

[jcm566]

I'm usually pretty good at estimating ratings with a quick rough estimate looking at the open field. I've always wondered this though:

Say a group of 10 players play a round. They average a 900 rating and their average score was 50. So now a 50 is rated about 900 and you would find variations using the above mentioned method.
Now a second group of 10 plays the next day/weekend. They average a rating of 1000 and they each all shoot an average score of 50. Now the rating for 50 will be a 1000, correct?

I've heard people say that playing NTs doesn't inflate ratings, but I really don't see how it wouldn't. Also the higher the competition at a certian tournament, the higher the ratings will end up

 

[Cgkdisc]

Because it doesn't happen. For an established player to shoot 10 throws better or worse than their rating occurs less than 1 in 100 rounds. For ten players to do it in the same round occurs less than 1 in 10 to the 20th power rounds. Not even sure what the kazillion prefix is for that number, maybe quintillion?

 

[juju1]

I couldn't attach a document but here is a link where you can download a ratings calculator:

http://southernmndga.ning.com/forum/topics/j-j-disc-golf-league

 

[Timko]

Chuck, why is the ratings system linear? I was thinking about this in reference to bowling, where there's a multiplier effect with score. So the closer to a perfect game you bowl, the more your score compounds. Wouldn't it be harder to shoot a super low score because it just takes too many strokes to get through the course? I could think about a potential density function to match this, but I was curious if you had explored a non-linear points-per-stroke formula, and why it was abandoned.

 

[Cgkdisc]

There is a compression effect in the formula that increases the value per throw as the course gets easier to the point where better players can't shoot much better than lesser rated players since they can't get holes-in-ones to keep separation. There's no reason why one throw difference on hole 1 should be worth any different from a one throw difference on hole 5 or hole 9. For example, both Player A & B shoot 52 on a course. They each par 3 the first sixteen holes. Player A aces hole 17 and pars 18. Player B deuces holes 17 & 18. Should they each get different ratings for that score of 52 they both threw?

Theoretically, they could have different ratings if we could analyze more hole data. But our scoring unit for data is a round score, not each hole score, although we can work backwards to determine the estimated SSA value of a hole from the SSA of the course. But that is still a derived figure that may not hold up if we actually had a more detailed way to analyze hole scores.

There's a fairness factor that comes into play. Even if we could get down to some high level math, I don't think we could persuade members that Player A or B should get a better or worse rating for that round. We already have this issue for ratings at an event. If a tournament plays two courses with SSA values maybe 5 throws apart, you'll see players who end up with the same total score getting different average ratings for the event. If Player A shoots two throws better on the easier course than Player B and they throw the opposite scores on the tougher course, Player A will get a point or two better rating overall because one throw is worth more on the easier course.

 

[Timko]

I see what you're saying; it would require some hole by hole rating as opposed to an overall round rating. Something like the math that goes into the hole by hole analysis at the Vibram Open. Of course, that's much easier because it's one event on the same course for each of the rounds.

 

[Timko]

BTW, this is the idea I'm currently throwing around for running a handicap league in KC. Any pointers or pitfalls you've seen would be very helpful.