Talk:Elo rating system

The math behind the 400 rule

It seems like it makes winning against much stronger players (>400 of difference) exponentially less and less significant. Wouldn't that discourage players from competing against much stronger players? In the sense that a reasonable player would never compete against a player with more than 400 of difference, if he has a choice of competing against another player with a smaller difference. I'm just wondering if there is a logic behind that. —Preceding unsigned comment added by 41.239.69.171 (talk) 21:23, 4 November 2010 (UTC)

I read that the idea behind it is if the ratings are close to accurate, beating someone more than 400 higher is a fluke, and shouldn't be weighted too heavily. Also the lower-rated player has practically nothing to lose and a lot to gain, rating wise. And you learn by playing a higher-rated player. Bubba73 ^{You talkin' to me?} 22:34, 4 November 2010 (UTC)

To call it "just a fluke" is to hugely misunderstand the statistical ideas of Elo and of chess. Carlsen should beat anyone that's 1200 or lower 9,999/10,000 times, provided he was taking each game seriously - I am not coming up with this 10,000 number as a random guess, every 400 points in difference leads to a factor of 10 in expectancy of scoring against someone.

1/10,000 times he would lose due to an absurd blunder on his part (known to happen to the strongest), and the other player playing the game of his life. Alternatively, he could draw twice in 10,000 games, which might in fact be more likely. This should be reflected in the rating system, ie. he gets a large amount of points if he wins, and Carlsen would get 1/10,000 of that for every win he gets. - And remember that 10,000 games is one HELL of a lot of games, people don't play that amount of FIDE games in their entire career, even 100 games is more than you might think, so yes it could happen. Anonywiki (talk) 21:20, 18 November 2011 (UTC)

ELO inflation

(Sorry cause my poor english)

Hello,

I listened about ELO inflation as some inaccurate of the system but after read about it is not what I spected. The firsts sites appears in Google: http://members.shaw.ca/redwards1/ http://www.chessbase.com/newsdetail.asp?newsid=5608 Talk about the "inflation" of just the ELOs of the best players! I understood as inflation a grow of the mean, not a grow of the best ELOs. The grow of the best ELOs is, in my point of view, a too logical consecuence of two things: 1-There are more players now than before so it is more possible that talent guys that not player chess with years ago situation, now play. 2-Actual "real" level of the best players is higher than the level of old best players cause the progress of the way to be prepared to the competition.

In anyway, exists a simple method to determine if this inflation exists as an inaccurate of the system or just cause really the actual level of the best is best than before: By construction, the ELO system just says that a gap of 200 points in ELO means that the better ELO guy takes the 75% of the points against the lower ELO guy (see this same site). So if that inflation is an inaccurate of the system and the borders was wrongly expanded, now a 200 points gap means less "level" gap than before. So, to refuse or proof that inaccurate expanded of the borders, you just need to compare older and newer results between guys with 200 points ELO gap and then can stablish: 1-If actual results between 200 ELO points gap guys are more tie than in the pass, then that inflation exists. 2-If actual results between 200 ELO points gap guys are the same than in the pass, then that inflation is a lie.

Regards,

Luis from Paraná, Argentina. — Preceding unsigned comment added by Luis Babboni (talk • contribs) 12:23, 28 July 2011 (UTC)

Deflation explained

The article talks a lot about deflation, but not about the causes. The cause(s) of deflation are fundamental to the discussion and should be included, even if to say nothing else than the causes are not known (if that is the case). - 98.225.38.209 (talk) 00:50, 12 September 2011 (UTC)ATBS

Starcraft 2's Use of Elo

From this [[1]] I can't tell if SC2 uses Elo, but it very clearly doesn't matter whether Elo or TrueSkill is used to calculate ratings, because there are many other things going on. It seems to me that SC2 uses a modified TrueSkill system that only accounts for limited game histories, but then has other mechanics going on as well such as a second "hidden" rating (with a much larger K value), and with bonus pool points (described in the blog) that would help to obfuscate the actual workings. Given all of this, it seems like it may be preferable to simply not list Starcraft 2 on this page, because until Blizzard makes transparent how their system works, claiming that it is backed in Elo vs TrueSkill is clearly speculation. 128.101.35.157 (talk) 18:48, 12 December 2011 (UTC)

League of Legends Elo usage.

The online competitive game League of Legends also uses the Elo system for their ranked games, although modified to fit their needs: http://na.leagueoflegends.com/learn/gameplay/matchmaking I would just add this but I've never done formatting/editing on wikipedia before. I'll have to look up the wiki page for that soon... 72.226.12.214 (talk) 04:56, 19 December 2011 (UTC)Mnenomenon

Lead-in

The lead-in of this article is not particularly useful. I came looking for a brief overview of what the Elo system is and how it works, instead I get some biographical information, talk about how different organisations use different types of calculations and something about the maths behind it - no actual definition. Find it problematic, but can't correct it myself since I have no knowledge of how Elo works - which is why I came here in the first place! --109.148.43.57 (talk) 21:11, 28 December 2011 (UTC)

I'll try to add a paragraph to the lead giving the general idea. The details are in the body of the article. Bubba73 ^{You talkin' to me?} 21:46, 28 December 2011 (UTC)

Y Done Bubba73 ^{You talkin' to me?} 00:13, 29 December 2011 (UTC)