Author: KarinsDad
Date: 15:29:41 02/02/99
Go up one level in this thread
On February 01, 1999 at 20:25:36, John Coffey wrote: >It makes sense to me that Elo could predict that at a 200 point difference the >stronger player would win 75% of the time. (Think about it: Elo can predict a >200 point differnce because that is how the players got to be 200 points higher >... they win 75% of the time against players who are 200 points lower. But why >on earth does that mean that a player rated 400 points higher would lose 5% of >the time? One conclusion does not necessarily back up the other.) I agree. I think that the problem is that the formula comes close to matching a high percentage of the cases (90% or higher), however, that same percentage of cases mostly falls within a +-400 rating differential. For example, in a 6 round open (non-sectioned) tournament, you may play someone 800 points higher than you in the first round, but the other 5 games would usually be within 400 points max. In fact, I just went to my state web site and looked at a one section open tournament (the only non-sectioned open each year in my state to my knowledge). There was one game about 1000 points apart, 1 900 points, 2 games 800 points (one of these was my game, figures I would be paired against the tournament favorite), and about 20 600 apart. All of these were in the first round except for 2 games. There were 135 games total, so less than 20% were against people 600 or more points. Out of those games, one person won at the 600 point differential (his rating was 150 points higher after the next tournament after this event, he is definitely an up and coming player) and another person drew at the 800 point differential (the upset of the year in my state). But in sectioned open tournaments or in quads, the chances of playing someone more than 400 points above you drops dramatically (unless you play in a higher section than required). > >The classic way to raise your rating on ICC is called "inching." You play >lower rated players and get 1 point here, 2 points there, another point from >someone else. Yeah, I try to avoid this on fics. I prefer to play people higher (or just slightly lower) rated than myself to give me a more accurate rating there. KarinsDad > >John Coffey
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.