Explanation
by Peter Ritmeester

The ratings are calculated in almost the same way as the World Chess Federation FIDE does with its ratings. The procedure is as follows:

1. First a player is given a rating to begin with. This rating is not very important, it is only a means to start right away. If the rating is too high, you will lose points quickly, and vice versa.

2. After a match, the difference between the two ratings is calculated, with a maximum of 350.

3. Next, the expected win-percentages are looked up in the following table, made by the FIDE. The first column indicates the rating difference, the next two the expected result. We use these numbers only for a 3-0 victory. The numbers for a 3-1 or 3-2 victory are different. 4. Transpose the result to a victory percentage. Chess only has 100%, 50% (draw) and 0 %, we use 100% (3-0), 75% (3-1), 60% (3-2), 40% (2-3), 25% (1-3) and 0% (0-3). In case of a best-of-three that is 100% (2-0), 67% (2-1), 33% (1-2) and 0% (0-2).

5. Deduct the expected percentage (Pe) from the real percentage (Pr), and multiply the result by the so-called K-factor (=50*), divided by 100**. The result is added to the old rating and voilà, you have your new rating. The formula is therefore (Pr - Pe) * K / 100 = correction. In case of a best-of-three, K=30.

6. Example: player A has a rating of 1800 and player B 1650. A wins 3-2. Pe of A is 70, Pr is 60. The correction is therefore (60 - 70) * 50 / 100 = -5. Therefore, A loses 5 points: he should have won by a larger margin!

* FIDE uses K = 25 for new players who did not yet play 30 matches, K = 15 after that, and K = 10 for players above 2400 who played at least 30 matches. By using an educated begin rating we do not need to take into consideration how many matches were played. By using a higher K = 50, players rise and drop much faster, which is more exciting. At any time K can be lowered if that is preferred. A higher K does not mean that the ratings will become higher: it is only an indication of the speed of change..

** The FIDE does not use percentages but probabilities: a percentage of 75% is called by them a probability of .75. Therefore we need to divide the result by 100.