Originally posted by welsharnieyes but I note that you haven't played many games here yet, it seems to me that your rating has yet to reach it's peak
I was recently playing someone who happened to have a profile stating their age and location. From chatting over the board, what he told me about his position in OTB junior circles, his OTB rating would have had to be around 500 points below his rating here!!! (My rating here, for example, is 250 points below my FIDE grade).
Originally posted by XanthosNZTrue, but the point being his play was good enough to suggest there was no way he could be as weak as would have had to be for what he was claiming his OTB situation was (in addition the only games he had lost were ones he had inexplicably resigned in good positions, some of them winning positions). He was finding moves in our game a 1200 rated OTB player would not find, no matter how long he had to think...
I have a provisional NZCF rating. It's horrible. It's also based off one 5 round OTB tournament quite a while back. It was my first OTB tournament and as such I played horribly mostly due to nerves. Rating itself does not tell the entire story.
Also, correspondence and OTB chess are very different games. I take my time on this site, I use databases and ...[text shortened]... lightning (based on only a few games though). The shorter the time controls the worse I play.
Originally posted by stevetoddI expect to reach around 1900 here. To me that makes it even more surprising that this player, who must be around 1200 OTB, was playing some very good chess against me...
yes but I note that you haven't played many games here yet, it seems to me that your rating has yet to reach it's peak
Originally posted by SkorjRunning it through the computer is not the be all of detection. I have run several club games that match up perfectly with the computer. If your opponent is very studied in a particular opening or a couple they will have high match ups. I only get suspicious if said player can play every opening flawlessly or if they are matching up for a large amount of time at about 50-85 % and then all of the sudden start matching up at a consistant 90+ matchup for the rest of the duration of about 10 games in a row or more. I will also grow suspicious if a player is rated very high and makes 1700 perfect match moves in about 14 days. It is not uncommon for a low rated player to have match ups depending on the positions on the board. The moves may just be odvious simple to spot moves or ones they have studdied.
I just finished a pair of games on another server with an opponent rated 100 pts below me, but who won both games. No big deal, it happens. I subjected both games to computer analysis and found their play was flawless both games. Since it is extremely for this to happen even once for someone who plays at my level I thought there was a sure sign that they had u ...[text shortened]... a strong human player or somebody just entering the moves an engine was spitting out at them.
RTh
When you play a book opening it is not hard too play a perfect opening with a good database. A popular opening at the GM level may be analyzed as deep as twenty moves or more. Twenty moves will take the game past the start of the middle game. Too analyze a game for program use you might want to have your program analyze every move and not the game. Also you need too start analyzing the game after the game has left the book.
To clarify, the 100% agreement of which I originally spoke is the "agreement percentage" in CM9000. This is the percentage of moves, not counting what it considers book moves, which does not damage the computer's assesment of your position more than a certain amount. This is different than a 100% correlation with the moves the computer would have played itself. This statistic usually serves as a rough measure of how well I or an opponent have played. For someone of my calibre, 100% agreement is rare.
I have heard of more sophisticated cheat detection methods by which the moves played are matched against the moves suggested by a variety of engines to see if any of them provided the alogrithm that dictated the moves. I was wondering if anyone knew of similar methods that wouldn't require me to have access to every popular engine out there and avoids even the appearance that I am accusing my opponent when I believe theire's an exellent chance that they're really legit.
- Joined
- 03 Sep 03
- Moves
- 87628
Originally posted by SkorjYou are correct about that!
is different than a 100% correlation with the moves the computer would have played itself. This statistic usually serves as a rough measure of how well I or an opponent have played. For someone of my calibre, 100% agreement is rare.
I have heard of more sophisticated cheat detection methods by which the moves played are matched against the moves suggested ...[text shortened]... m accusing my opponent when I believe theire's an exellent chance that they're really legit.
It is very common for players rated 1400+ to have > 90% agreement percentage with CM9. But that same 90% will drop to 40-50% if you compare it to the top 2 or even 3 moves suggested by CM9.
Like you said it is rare, but I my own games have had a 100% agreement. But when compared to the actual moves suggested by CM9 it is usually closer to 40%.
Originally posted by welsharnieJunior ratings for OTB are weird. Some games get rated, some don't, the time control are not used sometimes, etc... Look at ark13.
I expect to reach around 1900 here. To me that makes it even more surprising that this player, who must be around 1200 OTB, was playing some very good chess against me...
The other thing is that they don't play enough OTB. You gotta play regularly otherwise your rating will not increase while you rhp rating will.