From: Eric Cheymol ESPACE AEROSPATIALE FR> Date: 1 aug 1996 Subject: No Subject I give here some answers to the questions about the elo system. 1. ------- George I. Fernandez wrote : >What rating formula do you use? The formula used is the FIDE formula explained in the "Official Laws of Chess" (see p 57 in the 1985 edition). On the pratical point of view, I use : Bonus = K * (V - 1 / [1 + 10**((EloOpp - EloOld) / 400)]) where EloOld = Your elo at the beginning of a period of 6 months EloOpp = Elo of your opponent at the beginning ot the same period V = (1 for a win, 0.5 for a draw, -1 for a loss) K is the development coefficient K is the only difference with the original FIDE formula, it depends on your elo at the beginning of the 6 months period. K is set as follows : Elo K >= 2400 16 2350-2399 17 2300-2349 18 ... 2000-2049 24 ... 1700-1749 30 1650-1699 31 <= 1649 32 This correspondance favours low rated players and encourage improvements : a kyu player needs less to be promoted (on a elo point of view) than a Dan player. And the difference between theoritical elos is 40 or 60 points for Kyu grades and 100 points for Dan grades. 2. ------- >How often are handicaps used in European tournaments? >Do you have statistics as to results at each handicap? >Or between players of different ranks? What are your recommended handicaps? Handicap games don't count for the elo list. We chose to use only even games. 3. ------- >Do you give "boosts" to players who are obviously underrated? When I see that a player has made a big improvement, I give him for his new elo his performance : I calculate the average of the elo of his opponents (at least 14 last games) and I had a bonus (table p 58 of the FIDE Laws of Chess) depending on the ratio Number_of_points / Number_of_games. If the new performance is more than 50 - 60 points than his new elo (calculated with the formula explained in paragraph 1), I give this performance as his new elo (Players concerned by this "boost system" are marked ++++ in the list). 4. ------- >What are your requirements for promotion? For kyu grades (below or equal as 2 kyu), there is no requirement : each country grade officer decides for promotion of the players of his country. This is why there is some big differences between the theoritical grades and elos. Promotions should take in account that winning against a 2 kyu player with 1500 is not at all the same as winning a game against a 2 kyu player with 1680 ! It shows the limit of the points system. And it leads to have many players below the theoritical elo. For 1 kyu to 3 Dan promotions, players should follow the points system (explained by Pieter Stouten) or the elo requirement. You can be promoted by the elo requirement if your elo is higher than the theoritical elo for at least 4 consecutive games (Or if your elo is once at least 40 points higher than the theoritical elo). In practice, this never happens as those players were already promoted by the grade officer. Finally, for 4 Dan promotion, both system requirements must be fulfilled. Only two players were promoted 4 Dan this way. 4 Dan promotion must be very hard to get because this is the last grade western Shogi associations can award. 5. ------- Various questions : Albrecht Heeffer wrote : >I have the feeling there is something wrong with these "theoretical" elo ratings >for grades: possibly the fact that they ARE so theoretical. If you compare the >ratings with the actual average rating of each grade from the latest rating list, >you get to see a picture of systematically overrated theoretical grades: > >Grade AVG Theory Diff >3 Dan: 1998 2020 22 >2 Dan: 1846 1920 74 >1 Dan: 1800 1820 20 >1 Kyu: 1689 1740 51 >2 Kyu: 1619 1680 61 >3 Kuy: 1597 1620 23 >4 Kyu: 1453 1560 107 >5 Kyu: 1454 1500 46 >etc. The official elo list doesn't have all the players : - The players who haven't played any tournament since July 1st 1994 don't appear in the list - Nonrated players are not put in the list (ie players with less than 12 games played). On the 1st july 1996, there is 209 players on the complete list, and only 110 on the official one ! I recalculated the average table in the complete list : 4 Dan => 2166 with 9 players 3 Dan => 2007 with 24 players 2 Dan => 1845 with 8 players 1 Dan => 1800 with 20 players 1 Kyu => 1689 with 23 players 2 Kyu => 1631 with 13 players 3 Kyu => 1574 with 13 players 4 Kyu => 1498 with 11 players While looking at this table, you can see that there is a very small number of players for a grade (except for 3 Dan players, due to an inflation with the points system). This prevents me for upgrading the elos to compensate for the overall variation of ratings (If some strong japanese players appear in the european Shogi world and if some beginners stop playing [as it often happen], should I substract some points to all the other players to have a constant average elo ? I don't think so for now.) >I believe these theoretical grades are used for acquiring a higher grade. If they >are overrated this would imply that one has to perform significantly above the >average rating of a grade during 6 months to get one, which seems to be unfair to >me. Theoretical grades are not used for promotion below 4 Dan : they can be used as an indication but they are not necessary. Theoritical elos are used to introduce new players in the list. Best Shogi Wishes, Eric