ratings

Site hosted by Angelfire.com: Build your free website today!

Ratings

Important Links:
Look up a USCF rating
USCF rating calculator
                       CERN calculator (accurate/we prefer)
               USCF rating supplement for Oregon
USCF ratings by state
                 Washington Scholastic Rating System

            "I'm not in favor of the emphasis on Elo ratings--I believe it creates a strange situation
             in chess tournaments, where chess art is on the backside....We should first think about
              the beauty of chess and the most interesting games."                           ~Anatoly Karpov

The USCF maintains official ratings for its members in three categories: regular tournament chess, quick chess (10-29 minute time controls), and correspondence chess. Regular tournament ratings are the only kind most of us look at. Tournament directors are required to get results to the USCF within 7 days of the event, and USCF subsequently publishes rating updates every two months. USCF members who have played in rated tournaments will see their rating printed on their Chess Life magazine mailing label--see example here. [Note: Internet and chess software ratings bear little, if any, direct correlation to over-the-board tournament ratings.]

In its Introduction for Tournament Directors, the USCF states: "When competing in rated tournaments, players have a right to expect uniformity in procedures and rules of play as explained in The Official Rules of Chess...they also have a right to expect that results will be accurately reported to USCF for rating." According to the USCF, events may be rated if:

1. The sponsoring organization is an affiliate of the U.S. Chess Federation.
2. All players are USCF members in good standing.
3. The event is supervised by a USCF tournament director certified at the proper level.
4. Play is governed by the tournament rules, procedures and policies set forth by USCF in The Official Rules of Chess.
5. The first time control allows each player at least 30 minutes (to qualify for regular ratings).

On the subject of ratings, we ought to state as a matter of perspective that Chess Odyssey de-emphasizes their importance. We consider a rating number to be an insufficient measurement of chess ability or success, especially in kids or any player new to competition. In fact, we'd prefer it if ratings under 800 remained unpublished. However, we do recognize that ratings are a standard tool used in the attempt to pair approximately evenly-matched opponents for tournament play, and they can be helpful in tracking progressive competitive improvement.

Historically speaking, "official" chess ratings are a relatively new thing. From the invention of chess up until the mid-20th century, there was no universal/precise rating system. Elimination or 'knockout' competitions, ranking ladders and pyramids have all been tried by chess clubs over the years, and various independent, often inaccurate, rating systems have been experimented with. Another way of classifying player strength that has been developed is the system of attaining "norms" (a norm is a benchmark based on achieving a predetermined score against a specified field of opponents--a system still used today by FIDE in the process of obtaining international master and grandmaster titles). It wasn't until 1959 that the USCF formed a committee to evaluate rating systems. Chairing the committee was Marquette University physics professor Arpad Elo. Elo "perfected" the statistical algorithm that had originally been created by Ken Harkness for determining an individual player's strength in relation to others he played.

The Elo system was first adopted by the USCF in 1960, with FIDE accepting a modified/separate version in 1970. [Note: FIDE ratings carry more weight--i.e., when a FIDE rating is converted to a USCF rating, it typically jumps approximately 50-150 points.] In the final system, Elo retained the traditions of 200 point divisions between classes, 1500 as the rating for the average adult club player, and 'master' status at 2200 and above (for a list of all the USCF rating classes, see our USCF page). Notably, Kasparov was the first to break the 2800 rating barrier in 1989, and this is a FIDE rating! Regarding the USCF rating system, it should be kept in mind that it has undergone refinements in recent years, such that a USCF rating calculated today, under certain circumstances, could reflect a higher number than it would have 20 years ago using the system as it was then.

Specific calculations used in determining USCF ratings are here. The basic formula for an established player is as follows:

NR = OR + K (x - y), wherein:
NR is the new rating
OR is the old rating
                K is the coefficient, or value of the game, between 10-40*
x is the total score (win=1, draw=0.5)
                 y is the expected score based on OR vs. opponents' ratings

*A lower coefficient is used for older and more established players, giving more weight to previous performance
and adjusting the rating at a slower rate. A higher coefficient is used for younger, less established players, giving
more weight to recent performance, and adjusting the rating more rapidly.

The outcome reflects the difference between a player's expected result (based on their old rating compared to their opponents' ratings) versus their actual result. It is based on the theory that players of equal rating each have a 50% chance of winning a game against each other. An appropriate coefficient is used to modify the speed at which a rating is adjusted. In addition, if a player's performance in a tournament is significantly better than expected per his/her previous rating, he/she may gain bonus rating points as well. Bonus points are incorporated into the USCF rating calculator, and are designed to move the player more quickly to the class of playing strength where he/she belongs.

Previously unrated players will receive what is considered a "provisional" rating for their first 25 games (players ratings usually fluctuate quite a bit during that time), because establishing a player's expected performance requires a certain amount of data. A "performance" rating merely refers to strength of play in one particular tournament.

Because of some unscrupulous players who have intentionally lost games in order to qualify for prize money in lower class sections of tournaments (a trick known as "sandbagging"), the USCF established rating floors a number of years ago. Rating floors prevent a player's rating from dropping much more than 200 points below their highest established rating.

One last note: there are those who debate the validity and/or accuracy of the currently accepted rating system, and some would like to change it. For an example of an alternative approach, check out WorldChessRating.com.

Also check out:
ChessMetrics (Jeff Sonas)
'The New & Improved USCF Rating System'
Chess Express

Click here to return to the Chess Odyssey home page.