PDA

View Full Version : CM Ratings



andrewaclt
11-04-2007, 11:05 AM
Is there any correlation between the ratings of your CM opponents and FIDE or USCF ratings? i.e. If I am consistently beating 1700-1800 opponents in CM will I fair similar over the board?

diskamyl
11-04-2007, 03:08 PM
Originally posted by andrewaclt:
Is there any correlation between the ratings of your CM opponents and FIDE or USCF ratings? i.e. If I am consistently beating 1700-1800 opponents in CM will I fair similar over the board?

although I've never had an official rating, I believe they should correlate somehow after 1800, because after 1800, CM personalities stop giving up pieces for nothing, and play a reasonable game. below that level however, again it should correlate somehow but the variation range should be grater.

but still, bear in mind that the mistakes of CM personalities don't really look like their human correspondents'.

likesforests
11-04-2007, 07:12 PM
I think the ratings in CM 9000 were calibrated based on tests against human USCF players and so they should be accurate to within 50 points. But I don't know if there's been any re-calibration since then.

bear in mind that the mistakes of CM personalities don't really look like their human correspondents'.
That's true of my experiences with previous Chessmaster versions, but I think it's unrelated to their accuracy. If Kasparov always dropped a knight and then played full-strength his rating would still be consistent but much less than his normal rating. In CM 11 Josh, Age 8 (1600) didn't leave any pieces or pawns en prise when I played him... although he played many weak moves.

Biker1985
11-07-2007, 12:38 PM
Originally posted by likesforests:
I think the ratings in CM 9000 were calibrated based on tests against human USCF players and so they should be accurate to within 50 points. But I don't know if there's been any re-calibration since then.
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">bear in mind that the mistakes of CM personalities don't really look like their human correspondents'.
That's true of my experiences with previous Chessmaster versions, but I think it's unrelated to their accuracy. If Kasparov always dropped a knight and then played full-strength his rating would still be consistent but much less than his normal rating. In CM 11 Josh, Age 8 (1600) didn't leave any pieces or pawns en prise when I played him... although he played many weak moves. </div></BLOCKQUOTE>


The AI-personalities in CMX are very weird:

In a Fisher 5/3 game, Rybka v.2.3.1.w32 won against CMX without any problem
but Odile (1801 CMX rating) won from the same Rybka v.2.3.1.w32 in the same Fisher 5/3.
No need to tell you that CMX 2800? had no problem with Odile 1801.


Thus CMX Odile 1801 has a special tactic that can beat Rybka? Or was it lucky randomness?