About Our New Statewide Rating System
Ever lose 50 points to a sandbagger? That wont happen in this system.
Starting soon, a new rating system can add more fun to our table tennis statewide. Not only is this system more stable and accurate, it can also be used for individual challenge matches at your club, or even in your basement. Of course, any matches played at club round robins or at official USATT tournaments can be entered into the system as well. The results will be processed quickly, and theyre easily viewable on the web.
The neatest thing about this new system (and probably the trickiest to understand) is the addition of another number beside your actual rating. For example, my current rating in the system is 1751(59), and an unrated newcomer, NewGuy, who visited our club for the first time recently is rated 1450(300). The 59 or 300 is the Accuracy Factor, which, in statistical terms, is the standard deviation. This number helps in two main ways.
First, it tells us how accurate the ratings are. The lower the number, the better. Obviously we know more about my level of play after 15 years of my being at that level (hold the wisecracks, please) than we do about someone who just walked into the club.
Second, the Accuracy Factor protects my rating from those dreaded players who can steal lots of points so easily. Im talking about sandbaggers, unrated players, those who dont play tournaments for a long time, or those darn kids who gain 800 points in a year.
For example, lets look at the very unlikely scenario that NewGuy beats me in a match. The current USATT system would take 50 from me and give 50 to NewGuy. However, this new system would only take about 16 points from me, while giving almost 400 to NewGuy. The new system recognizes that we dont know as much about NewGuy, so he needs more adjustment to raise his rating to where it belongs. The system also knows that I shouldnt be punished just because we didnt know that NewGuys rating was really closer to 1850. In his next matches, all of his opponents will also be protected until his Accuracy Factor comes down to below 100 (which will happen as he plays more).
Send any questions or comments about system implementation for Colorado to me at ttcivilwar@hotmail.com. You can also find statistical information at the links below.
Pongingly,
David Huggins
For rationale and to see the system in action elsewhere, visit davidmarcus.com.
Another link of explanation can be found at:
http://tabletennis.about.com/library/ratings/bl-virginia-rating-scale.htm. Here is an excerpt.
The Virginia Table Tennis Association is using the Marcus Rating System which was developed with the help of a grant from the US Olympic Committee and was recently featured in the Journal of the Royal Statistical Society: Series D (The Statistician), 2001, vol. 50, part 2, pp. 191-208. Like other new systems, it sometimes take a little time to become comfortable with it, but understanding the basic logic behind the system should speed up this process.
The system is not a magical black box, just a statistically sound algorithm based on probability that treats every player fairly regardless of level of play and level of activity. Here are three charts (see link above) that show how ratings under the system will change, along with some commentary to help you better understand what the system is doing.
The goal of any rating system should be to accurately and fairly rate all players based on their tournament results. If you are not that familiar with the inner workings of the present USATT rating system, then you may not realize that while the system does a good job overall, it is far from perfect. The Marcus system is a very significant improvement over the current USATT rating system that produces more accurate ratings while treating all players fairly and impartially.