Chess Online

Play online chess!

From	Message
goldentweety 16-Jun-21, 19:05	Matching Parameters I got into a discussion with a fellow captain as to our team policy in using the 90 day high and the player's graph instead of the 90 day average. We also take into account player history, time per move when required, and highest won against. My question is what system do you use? I am always willing to adapt to a better matching system but we have been using this since I was captain in 2016 and before when I was cocaptain and it has been successful for us. I am always willing to learn more. What do you think?
neurokarma 17-Jun-21, 01:48	I tend to first look at the 90 day avg...Also 90 day high and low and avg opp rating. Time per move only when requested. Check under rivals to see if they've played before and results. I don't take into consideration results or ratings more than 6 months old.
goldentweety 17-Jun-21, 02:46	Deleted by goldentweety on 17-Jun-21, 02:46.
goldentweety 17-Jun-21, 02:48	Hovering the mouse Mouse over the link and it gives you the 90 days. Is this what you are talking about? This is their top rating and indicates player potential. I have always used this and we have never had any problems. I only check further if they are close to being 50 out of range
neurokarma 17-Jun-21, 08:32	Mouse over name only gives 90 day. I look at the player profile for the other stats
joveyboy1 18-Jun-21, 21:30	A team my size, I would like to look at extra stats, but it takes me too much time. For me, I primarily look at current and top 90 day ratings, but I do feel the average 90 day ratings are significant, so I am willing to spend a little more time looking at that. I also check to make sure the players I am planning on matching aren't already playing or have played recently, I try to give some diversity. I'll look at average time per move too, but it isn't priority.
goldentweety 19-Jun-21, 12:17	Rivals and the chart I too have a big team so usually if elo and 90 days are close I accept but I am learning to appreciate their chart. It really shows a player's strength and whether they are overated and underated. We are trying to implement rivals. Our players dont like duplicate matches and if they have a losing record against an opponent there is not much sense matching them but this is a judgement call. With the addition of amacivn, we are going to try using rivals as there are two of us using it now...half the work Hopefully my other cocaptains will agree and we can start using this system as of July 1
pmcmurphy 19-Jun-21, 14:05	Rivals Goldentweety: How much weight are you thinking of giving to past records between rivals? At what point does a losing record become bad enough that it rules out a match? I agree that the chart can be very useful. My impression is that it is underused as a matching tool.
goldentweety 19-Jun-21, 14:28	It depends If a player has just finished a match and lost 2-0 a couple of days ago I dont think it is wise to pair them again so soon. Some players have extensive rivalries and I dont see a problem if they are +/- 4 as long as they havent played each other recently. I have had complaints from team members regarding duplicate matches and "playing a superior opponent" I recently turned down a match because the chart showed that the opponent was a 1600 player with a 1400 rating currently. While the rest of the stats were ok. my player was overated at 1400 so this is not a fai rmatch. A couple of things you have to watch out for but they are hard to detect. Some players are in the league or the ladders and they lose these games to keep their rating down. So when they play team games they have an unfair advantage.Also I am wary of people that have completed 100 or so games and are crushing their opponents at low ratings while they achieve their true rating. This is not an exact science and you have to rely on intuition and judgement. Every matchup needs to be evaluated on its own merits
pmcmurphy 19-Jun-21, 15:01	Very true I think your third paragraph identifies two sources of the biggest mismatches: (1) GameKnot newcomers whose ratings do not yet reflect their actual playing strength but are matched as if their current ratings are accurate and (2) players who for whatever reason play much better in team games than in non-team games but again are matched as if their overall ratings are accurate for team games. I click on "Rivals" when I remember to. Seeing the past history can add to the drama of a match. I don't actually know one way or the other how reliable past heads-up records are as a predictive tool, but there are times when the record is lopsided enough that I decide against sending a challenge that I was considering. I do know that if two equal players play enough matches against each other, one or both of them will at some point obtain the "lead" in their rivalry. Even if they're completely equal, not every match will end 1-1. But of course if one player has an advantage over the other, that player will also tend to take the lead in the series. Sometimes it's not easy to figure out if you're looking at variance or a significant trend! Anyhow, the "+/- 4" flexibility that you mention sounds similar to my intuitive reaction to past heads-up histories when I remember to look at them.
amacivn 20-Jun-21, 11:54	Rivals I've always used the feature when accepting and making challenges ....it's probably the 2nd thing i look at ... some players play each other a lot .. and can lose heavily.... I wouldn't match our player if they lose to frequently against the same person ...especially if the losses were very recent ... I wouldn't like it why should they ... if they're old then they're not to significant then I check other stuff ... it's a feature that's part of my system... not the only feature
pmcmurphy 20-Jun-21, 16:00	The easiest player to evaluate The easiest player to evaluate is one who has a long GameKnot history, has never timed out a game, has played to the best of his ability in every game, has a rating floor that is well below his playing strength, and has never gotten any better or any worse at chess since he started playing on GameKnot. For that player, I am pretty sure the best estimate of his abilities is his average rating for either one year or for all rated games. An intelligent reading of his graph, also, will yield a similarly accurate estimate of his playing strength.
goldentweety 21-Jun-21, 17:33	We are making a small change to our matching strategy. Goals: 1) prevent duplicate matches from the same opp0nent 2) prevent loss of matches when a player has played the same opponent several times with a losing score 3) prevent opponents who manipulate their ratings by losing personal games Resolution: 1) evaluate according to elo and 90 day requirement 2) evaluate players graphs to determine if opponents/players are underrated/overrated 3) evaluate current matches to prevent duplication 4) evaluate rivals to see overall record against opponent. I realize some have an extensive rivalry but we wont match unless the overall score is +/- 4 games This additional checking may cause delay in matching but we are seeking quality over quantity enabling our players to have a fair chance in their matches
joveyboy1 11-Sep-21, 08:50	I do try myself to send you matches with these preferences you desire. I know I won’t always do the best, but I’m willing to take the extra time to try to start a good match. I’ll try my best for sure
saguaro 11-Sep-21, 12:14	GoldenTweety Thanks for sharing that! -Dave

GameKnot: play chess online, chess clubs, monthly chess tournaments, Internet chess league, chess teams, online chess puzzles, free online chess games database and more.