#51  
Old 05-01-2014, 09:20 AM
grodney's Avatar
grodney grodney is offline
* Ace Member *
 
Join Date: Nov 2008
Location: Charlotte, NC
Years Playing: 38.1
Courses Played: 119
Posts: 3,265
Niced 8 Times in 5 Posts
Default

Quote:
Originally Posted by mashnut View Post
Also, I don't know if he would share, but Chuck has posted several times some numbers he's run based on ratings and standard deviations of how often players at specific ratings will beat one another.
Back when I had full access to the database (ratings and st-devs), I did some "odds" for 2004 Worlds.

In the Adv Master division, Rhett Stroh had a 917 average round, with an insanely low standard deviation of 16. This meant he was a 100% lock to make the semi-finals, but had a 0% chance of making the Top 4.

Compare that to John Dorn who had a 911 average round, but a wild standard deviation of 42. This gave him only a 79% chance of making the semi-finals, yet he had a 2% chance of making the Top 4.

Rhett ended up making the semis and getting 14th, while Dorn finished 71st and out of the semis.

(Based on the ratings and standard deviations, I gave Doug Saulter a 98% chance of making the Top 4, and an 83% chance of winning. He did.)

Fun with numbers.
Sponsored Links
Reply With Quote
  #52  
Old 05-01-2014, 01:17 PM
mashnut's Avatar
mashnut mashnut is offline
*Super Moderator*
 
Join Date: Apr 2009
Location: Escondido, CA
Years Playing: 17.9
Courses Played: 831
Throwing Style: RHBH
Posts: 14,396
Niced 31 Times in 19 Posts
Default

Quote:
Originally Posted by MountainGoat View Post
I don't see why course difficulty should have anything to do with how a rating is determined. It should be based on finishing position compared to projected finishing position based on previous rounds played and the rank of the players finishing above and below said player. The quantity of points awarded or subtracted for the match should be based on sample size, and should be capped based on sample size, this keeps prestigious events special and prevents local leagues from developing the best ranked players on Earth just through league play.

If the worst player wins 1v1 with the highest rated player (provided it's a sanctioned round), he/she should loose a maximum amount of points based on the sample size. So if this 1v1 event had 700 participants, the maximum number of points that could be won or lost in a single round would be set to a certain amount. Higher rated events (Pro Tour stuff, Worlds etc.) would have intentionally inflated ratings both to demand a higher level of play and to demand that the highest ranked players show up or risk losing a big chance at holding their positions. Random events (like local leagues) would carry a base amount and as the sample size increased so would the event rating and therefore the ability to gain/loose rank. This serves multiple purposes; this makes a well promoted event or a growing league a very rewarding play due to the fact that you can really gain some points. Poorly run events would see the opposite effect. This would allow people who want to try, try, try, to be able to get into the ranks by just grinding events and would allow the top pros to get very near the top just by showing up to events and throwing good plastic.

The bottom line is the point system does not help promote PDGA memberships or play, and IMO, that should be it's primary function.
I think you may be mixing up the points system and the ratings system a bit. IF ratings were higher for top level events and lower for C tiers it would pretty much invalidate the entire system. You could be the equivalent of today's 1000 rated player and just stick around cleaning up in intermediate as long as you didn't play any NT events, meanwhile someone lower skilled who had the money to go to a bunch of tour events would be pushed out of the division their skills would typically qualify for.

What you're talking about happens to some extent with the world rankings. If you don't show up to the big events, you can't stay in those top rankings, but that's a totally separate system to the ratings or the points systems.
Reply With Quote
  #53  
Old 05-01-2014, 05:45 PM
DirtyMeathook DirtyMeathook is offline
Birdie Member
 
Join Date: Oct 2013
Courses Played: 17
Posts: 421
Niced 3 Times in 3 Posts
Default

Can we remove all of OP's posts from this thread to tidy it up a bit? Mash and Jeverett provide some great content.
Reply With Quote
  #54  
Old 03-12-2018, 05:15 PM
Halcón's Avatar
Halcón Halcón is offline
Free at Last
 
Join Date: Jul 2012
Years Playing: 8.7
Courses Played: 1
Throwing Style: RHBH
Posts: 12,006
Niced 1,190 Times in 605 Posts
Default

Quote:
Originally Posted by bradharris View Post
The formula is actually pretty simple. I had it figured out last year. It's basically a matter of averaging the scores and ratings of all propagators.
I've always wondered why the ratings at the Memorial were so high. I believe this event might be the most stacked, outside of Worlds, with top rated players. Is that the simple reason, because the field's average rating is just higher? Or do the courses also play a factor.
Reply With Quote
  #55  
Old 03-12-2018, 05:32 PM
Cgkdisc's Avatar
Cgkdisc Cgkdisc is online now
.:Hall of Fame Member:.
 
Join Date: Jul 2009
Location: Twin Cities
Years Playing: 30.8
Courses Played: 692
Throwing Style: RHBH
Posts: 12,341
Niced 1,726 Times in 789 Posts
Default

https://www.pdga.com/why-so-many-1100s-at-memorial

Niced: (2)
Reply With Quote
  #56  
Old 03-12-2018, 07:13 PM
bwgort bwgort is offline
Birdie Member
 
Join Date: Dec 2009
Location: Sandy Springs, GA
Years Playing: 12.7
Courses Played: 83
Posts: 413
Niced 157 Times in 86 Posts
Default

Quote:
Originally Posted by Cgkdisc View Post
I thought of an interesting question:

What's the easiest course that has produced an 1100 rated round? I think I know the answer (hint: it's a course in Georgia), but would love to see chucks answer.
Reply With Quote
  #57  
Old 03-12-2018, 07:16 PM
bwgort bwgort is offline
Birdie Member
 
Join Date: Dec 2009
Location: Sandy Springs, GA
Years Playing: 12.7
Courses Played: 83
Posts: 413
Niced 157 Times in 86 Posts
Default

Quote:
Originally Posted by mashnut View Post
I think you may be mixing up the points system and the ratings system a bit. IF ratings were higher for top level events and lower for C tiers it would pretty much invalidate the entire system. You could be the equivalent of today's 1000 rated player and just stick around cleaning up in intermediate as long as you didn't play any NT events, meanwhile someone lower skilled who had the money to go to a bunch of tour events would be pushed out of the division their skills would typically qualify for.

What you're talking about happens to some extent with the world rankings. If you don't show up to the big events, you can't stay in those top rankings, but that's a totally separate system to the ratings or the points systems.
How do you determine when "ratings are lower"? I think it's demonstrably true that ratings for C-tiers are on average lower than ratings for bigger events. Chuck will say, more local players play C tiers (and leagues) and there's less pressure at those events which accounts for the numerical difference. I believe and chuck can correct me if I am wrong, that he does see numerical differences in the ratings from different tiered events at the same course.
Reply With Quote
  #58  
Old 03-12-2018, 07:41 PM
biscoe biscoe is offline
* Ace Member *
 
Join Date: Dec 2007
Location: spotsylvania, va
Years Playing: 24.7
Courses Played: 94
Posts: 6,812
Niced 2,335 Times in 952 Posts
Default

More INT and REC (actually more newer players in general) players proportionately will drive down ratings imo. It is my belief that newer players improve more rapidly than their rating and thereby depress ratings at events. It would be interesting to see ratings calculated using only players with a low standard deviation as propagators.
Reply With Quote
  #59  
Old 03-12-2018, 08:05 PM
Cgkdisc's Avatar
Cgkdisc Cgkdisc is online now
.:Hall of Fame Member:.
 
Join Date: Jul 2009
Location: Twin Cities
Years Playing: 30.8
Courses Played: 692
Throwing Style: RHBH
Posts: 12,341
Niced 1,726 Times in 789 Posts
Default

Quote:
Originally Posted by biscoe View Post
More INT and REC (actually more newer players in general) players proportionately will drive down ratings imo. It is my belief that newer players improve more rapidly than their rating and thereby depress ratings at events. It would be interesting to see ratings calculated using only players with a low standard deviation as propagators.
In 2017, we introduced an adjustment for propagators in their early years of propagating based on assessing their average advancement after evaluating five years of data on new props. On average, the total pool of new props does not improve as fast as you might think but the small adjustment is now in the calcs.

We also have done the experiment using just Super Props with lower standard deviations. It turns out that there's no benefit. In other words, if you use the 25 out of 50 props with the lowest standard deviation, you get essentially the same variance using all 50. Having more props with slightly higher SDs is as good or better than a smaller pool of super props. The thing is, everyone plays better or worse than their rating at roughly the same frequency. So better to have more players in the mix playing better or worse than a potentially skewed smaller sample.
Reply With Quote
 

  #60  
Old 03-12-2018, 08:47 PM
biscoe biscoe is offline
* Ace Member *
 
Join Date: Dec 2007
Location: spotsylvania, va
Years Playing: 24.7
Courses Played: 94
Posts: 6,812
Niced 2,335 Times in 952 Posts
Default

In a field of 50 players (calling them all props) for each 10 points any one of the players is under-rated there will be a corresponding drop of .2 points in the overall rating. Let's say one guy is actually 30 points better than his rating, 3 are 20 and 6 are 10. That will drop the ratings of the whole field by 3 points apiece as opposed to a field full of players with stable ratings. This effect then ripples outward to other events. In a small field with say 10 props one player whose rating is lagging his skill level by 20 points will cause a 2 point drop for everyone just by himself.

I agree a laughably small sample size is better than a pitiably small one. Trying to generate statistically meaningful data out of this stuff is tough.
Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
PDGA ratings are up. How'd you do? gvan General Disc Golf Chat 3985 01-14-2020 03:52 PM
PDGA Ratings Changes Ermac010 General Disc Golf Chat 15 04-19-2017 04:07 PM
New PDGA Ratings 3/29/11 BIgNik General Disc Golf Chat 7 03-29-2011 02:37 AM
PDGA Ratings unknown2no1 General Disc Golf Chat 2 11-15-2010 09:10 PM
PDGA Ratings harr0140 Tournaments & Leagues 25 04-24-2009 05:22 PM


All times are GMT -4. The time now is 10:57 PM.


Powered by vBulletin® Version 3.8.10
Copyright ©2000 - 2020, vBulletin Solutions, Inc.