Welcome to my decription of the ratings system. Basically, the system works like this.
The whole system is based on percentiles. I begin with the assumption that the average player who has never played in a tournament before is at the 40th percentile, in terms of his expected performance versus that of his peers. He therefore has a starting rating of 40. This is a fairly arbitrary determination, but you have to start somewhere. After each event that the player plays in, his rating is adjusted, up or down, a small part of the difference between his rating going into the tournament, and the percentile score that he scored in the tournament.
For example. Suppose Bob goes into his first tournament. The tournament has 30 players, and Bob finishes in 7th place. This means that Bob scored at the 78.333th percentile in this tournament. [(30+.5-7)/30]*100 = 78.333
Bobs rating is now going to change. It is going to move a certain distance from it's old point, which was 40, toward this new target of 78.3. But how much is it going to change? Well, it's going to change by a pre-determined percentage of the difference between his initial rating, and his percentile score.
The difference between his initial rating, which is theoretically his expected performance, and his percentile score, which represents his actual performance, is 38.333 points. 78.333 - 40 = 38.333 The percentage of this difference that his score is going to change by, varries depending on the size of the tournament that he was in. The amount that a particular tournament will impact its player's ratings is its "Tournament Value".
My goal was for this percentage to be high enough for individual tournaments to be important, but not so high that player could rise to the top of the heap very quickly or easially. The value of a normal tournament is equal to the number of players participating in the event divided by three point five, plus two; with a maximum 15. Since the fewest players that you would play Diplomacy with should be 7, the minimum tournament value is normally 4. This range of 4 to 15 represent percentages of the difference between expected performance and actual performance that a rating can change. This formula has also changed since the system was initially introduced. The old formula was [#/5 + 5].
So, referring to our previous example, Bob performed as we stated in an event with 30 players, then his rating will change by (30/3.5)+2 percent of 28.333. 10.572% of 38.333 is 4.053, so Bob's rating will increase from 40 to 44.053 based on his initial tournament.
There are three important exceptions to this range of 4% - 15%. The World Championship always adjusts ratings at 20%, regardless of the number of players attending. This seems appropriate for the premier tournament of the year. A tournament which has only one round of play, does not get the bonus +2 to its value. A one round tournament simply has a value of #/3.5. Finally, the Diplomatic Pouch results database contains results for a number of events which are not really tournaments at all, but really just club games. These events will show up on a player's "Player History", but they do not effect ratings. Technically, the have a tournament value of zero.
If you take a quick look at one of the player histories, you'll be able to see how a players rating can change over their career based on their performance in each event they play.
[ The Zine | Online Resources | Showcase | Postal | Email | Face to Face ] |