2011 Ratings List graphs available on USCF website

All Things Chess
1

In addition to the statistical report on ratings distributions in the 2011 Annual Ratings List
that was sent out yesterday (BINFO 201100570), seven graphs showing the distribution of
ratings on the 2011 Annual Ratings List have been placed on our website.

These graphs were done on 10 point ratings intervals (eg, ratings from 1900 through 1909 are
combined for reporting purposes.)

The first graph is the distribution of established and provisional ratings for adult
members (age 20 and up):

uschess.org/datapage/ratings_2011A.png

The second graph is the distribution of established and provisional ratings for
players younger than 20:

uschess.org/datapage/ratings_2011Y.png

The third graph, which is not that different from the second graph, is the
distribution of established and provisional ratings for players younger
than 16.

uschess.org/datapage/ratings_2011S.png

There are also several graphs showing the distribution of established and provisional
ratings by age group:

Age groups 12/Below, 13-15, 16-19 and 20-24 are shown on these graphs:

uschess.org/datapage/ratings_2011AGE1E.png (established ratings)
uschess.org/datapage/ratings_2011AGE1P.png (provisional ratings)

Age groups 20-24, 25-64 and 65+ are show on these graphs:

uschess.org/datapage/ratings_2011AGE2E.png (established ratings)
uschess.org/datapage/ratings_2011AGE2P.png (provisional ratings)

(Yes, the 20-24 age group is shown on both sets of graphs.)

There are a very large number of players (mostly under 13) with provisional ratings
between 100 and 119, so that portion of the graph is chopped off.

Here are the totals by age group for those ratings:

100-109 110-119 12/Below 2687 673 13-15 238 47 16-19 92 9 20-24 5 0 25-64 15 1 65+ 0 0

There are very few players in any age group with established ratings under 150.

2

Very nice! In the first chart, the jump at 2200 for the floored masters is an eye-catcher.

3

As is the absence of any spike due to players floored at 2100.

There are similar spikes at 1500-2000, but the spikes for lower floors (we go down to 1200 now) are not as noticeable.

4

There are also graphs for female players (rescaled somewhat), see

uschess.org/datapage/ratings_2011AF.png
uschess.org/datapage/ratings_2011YF.png
uschess.org/datapage/ratings_2011SF.png
uschess.org/datapage/ratings_2011AGE1EF.png
uschess.org/datapage/ratings_2011AGE1PF.png
uschess.org/datapage/ratings_2011AGE2EF.png
uschess.org/datapage/ratings_2011AGE2PF.png

5

The lack of a spike at 2100 is likely due to anyone going over 2300 (getting a 2100 floor) probably ended up as a Life Master and has a 2200 floor instead of a 2100 floor.

6

Mike,
I’d be interested in seeing a version of the adult graph using only players active in the last year. I expect that the spikes due to ratings floors will not be prominent.
Also, could we get the data behind the graphs? My eyes seem to see a trend that the upper parts of each 100 point interval are under-populated (ie xx90). It may just be an illusion due to the spikes at xx00 from floored players.
Thanks,
Mike

7

It could be simply that the paucity of xx90-xx99 ratings could be due simply to the existence of the floor just above it. A player floored at 1800 may be more likely to have a “true” rating (the rating he would have without the floor) of 1790-1799 than any other single 10-point range in the 1700’s.

Bill Smythe

8

Seems to me, the existence of rating floors has made true ratings, not only of those playing at their floor, but their regular opponents, problematic.

I suppose the existence of players with a true strength masked by a floor can only accelerate the rating progress of those players actually improving, which may be good.

9

Those graphs were based on the 2011 Annual Ratings List, which is, for the most part, those players who were active in the last year. Some inactive players show up on the list due to rerates of past events.

However, it does not appear that results in a meaningful distortion of the data, since there is no reason why players at their floor would be more likely to see a ratings change due to rerates than players not at their floor. (In fact, players at their floor are probably less likely to see a ratings change.)

Looking at the 456 players on the 2011 Annual List with a 2200 floor, of the 109 players with a current published rating within 9 points of their floor (since the graphs were on 10 point intervals), 103 of them were active in the past year and 6 of them were not.

There is a supplement file available on the website for the 2011 annual list, it should contain the 60,651 records represented in the graphs. It does not, however, contain age information. It also does not have floor information, because that field was not included when the rating supplement file standard was created, some 20 years ago.

Access to member-specific or detailed data with non-public information such as age is restricted under our privacy policy. You would need to submit a data access proposal to the Ratings Committee, which acts as our institutional review board for requests by researchers for access to non-public data. They will review, among other things, what you will be doing with the data and how you will protect the confidentiality of non-public data and make a recommendation to the Executive Director as to whether your data access request should be granted. There may be a fee to generate the data requested.

However, the summary file from which the graphs were drawn has been posted on the website at:
uschess.org/datapage/dist2011.txt.

It should be possible to import it into Excel, it is a fixed field file.

Regarding floors and ‘true’ ratings, if you read the ratings committee reports, archived on Mark Glickman’s website (glicko.net), they should document the committee’s belief that floors are only slightly inflationary, and that is taken into account in their annual review of the bonus formula.

Prof. Glickman has started working on some conceptual tests of a Glicko-2 style rating system, but it is unlikely that the committee will be in the position to make a recommendation to the EB or Delegates in the next year.

10

If you can get the floored players to play, that is! Even when they’re not losing rating points, many of them hate to complete at less than their peak strength.

This is one reason I like the norm/title system…