Thursday, February 3, 2011

On-line Chess - Part 1.6 - Rating and levels


Why playing a game like chess is addictive? Because it is a beautiful game! Yes, of course. Because you meet people! Yes. Because you like the theory behind! Yah. But unconsciously a better answer is because you want to prove you’re stronger than the opponent. Another variant: because you want to prove you’re stronger than a lot of people. A more modest variant: because you want to prove you’re the strongest player, that’s all!

Rating helps to find your strength among other player. ELO rating is the most used. It is a simple formula returning a number: bigger the number, stronger you are.


Figure 6 – Rating computation formula. W=1(won) or 0.5(tied) or 0(lost); R1 is the current rating of the player, R1’ is the new rating, R2 is the rating of the opponent, K is the maximum possible adjustment per game and this constant is set to 32 unless the player has a provisional rating, in that case K depends on the number of games played by the opponent.

Talking about ELO rating is not completely correct. It is more correct to talk about a rating used to have a rank in a top list.

Every provider uses its own formula, especially because ELO rating formula doesn’t work well when the players have played too few games. This is the problem of how to start computing the rating.

In order to start the rating computation you should have an initial rating. There are different ways:
  • get a fixed initial rating (1200, 1400 or 1500 for beginners) from the system; 
  • play a bunch of games without having assigned a rating so that after this period the system can determine the best estimation of your rating. Sometimes the system accept hints about your experience in chess in order to find the most appropriate opponents for you during the estimation; 
  • play with a provisional rating for 20 games and after that your rating is established. In this case the rating is computed using different formulas for established and provisional players. The initial rating is chosen by the player during the registration.
A derived problem is to take into account the fact that a player could have been away from tournaments for years before start playing again. This is a special case of the previous problem in fact in both cases the rating of the player is not sure. For this reason some providers use a different and more complicated formula called Glicko System [1].

Ladder is another possibility. No rating, only a rank in a ladder. People who win will raise the ladder, those who lose does not rise and soon will be taken over by others. The goal is to be at the top of the ladder (rank 1) instead of having the highest rating.

Famous chess programs have the concept of levels: beginner, intermediate, master. Splitting the number of games into rooms is useful to spread the load over more game servers. So why not create rooms based of player levels? Yes it is possible but less attractive than rating, isn’t it?

Ratings, ladders and levels are usually different for different categories of game: correspondence game, classical game, blitz game (3-5min), bullet (less than 3 min) and chess variants have their own rating.

Today I don’t feel concentrated enough but I would like to play a game anyway, just to spend time, but if a play and I lose I will compromise my rating. Some web sites give you the possibility to play a not rated game, so you can still play for training for example without compromise your rank or your rating.
Almost all providers ask you to sign up before starting use the service. It is reasonable because at least your nickname should be visible by your adversary. Someone could think: “why should I give my personal information like e-mail and nationality if I’m just trying to play a single game?” Very few web sites let you play chess as anonymous: no ratings, no registered games, no customizations, just play! In case they like the service they will sign up whenever they want.

[1] The Glicko System, Mark E. Glickman, http://math.bu.edu/people/mg/glicko/glicko.doc/glicko.html

No comments:

Post a Comment