Tuesday, November 04, 2008

Ratings and Depression

On Wednesday April 16, 2008 at 7:00 pm Mark Glickman, Chairman of USCF Rating Committee and Associate Professor of Health Policy and Management in the Boston University School of Public Health, gave a lecture about the USCF rating system. The rating is key when used in paring the first round of a swiss tournament, and allows grouping in Quad tournaments. Essentially when two players face each other the difference in their ratings creates a winning probability percentage for each player.

The exact details of how ratings are calculated are quite complex and vary depending on your strength. Players can get bonus points if they play very well in a tournament, which is one of the mechanisms to try and give newcomers/aka children higher ratings faster. This is because children improve at a rate faster than the system can compensate them so, most kids win at a greater rate than their rating indicates. This causes their opponents to lose more points than they should when losing.

Consequently over the past 10 years this phenomenon has caused a deflation of about rating 100 points. Meaning at 1998 ratings levels, of most players below masters, would have had ratings about 100 points higher than they are today even if their actual strength is the same. This is called "rating deflation" but I call it "rating depression".


Mark said that the USCF is trying to build in a new way to compensate for this deflation. Glickman said that the true underlying effect is caused by the fact that most players improve at a faster rate during their first six years of play. And that it appears that chess teaching, coaching, access to internet information, etc. are ramping up this improvement rate. He said that the rate of improvement has accelerated to a 6% increase some years when compared to the previous year.


I worked with a very talented manufacturing engineer, Bob Thomas, who demonstrated to me that the very difficult process of assembling electrical train car connector heads got better and better every year with the improvement forming a natural log curve; aka the learning curve. And this improvement occurred with no organized effort by the company: the folks on the factory floor just got better as time went on.


The key trick is to BEAT the learning curve and it appears that kids, coaches, and parents are doing this. The number of kids rated mid 1800's and beyond, relative to their age at the BCF, is going up all the time.


Another interesting thing Mark Glickman mentioned was that most chess players start showing a slight decline in the ratings beginning at the age of 63 on average for most players.


What are your feeling about ratings and rating deflation?


Please Comment. Mike Griffin 11/05/2008


http://en.wikipedia.org/wiki/Learning_curve


http://math.bu.edu/people/mg/ratings.html

No comments: