PDA

View Full Version : Fritz/Rybka ratings VS reality



Paul Cavezza
14-02-2010, 01:46 AM
Hey there,

Would anybody hazard a guess at the difference in rating, if you put Fritz on 1800 strength and compare it to a human of 1800?

Basil
14-02-2010, 08:22 AM
zero?

Kevin Bonham
14-02-2010, 09:50 PM
Hey there,

Would anybody hazard a guess at the difference in rating, if you put Fritz on 1800 strength and compare it to a human of 1800?

May depend which Fritz. I actually find when playing blitz against Fritz 11 that its low-rating strengths are surprisingly strong, to the point that it seems quite underrated. That is as opposed to its higher levels, on which it clearly overrates itself.

Paul Cavezza
15-02-2010, 08:22 AM
hehe overrates itself:D damn cocky fritz!

I agree with Kevin I find they make far less human style tactical lapses and positional mistakes when you drop the rating. Most software struggles with it- if you play an engine online it will often just give you 3 or 4 tempi in the opening then play normally..

Kevin Bonham
15-02-2010, 10:24 AM
Fritz 11 seems quite different to Fritz 6 in this regard; Fritz 6 on low rating settings often plays unsound lines that are easily and repeatedly beatable.

Denis_Jessop
15-02-2010, 02:23 PM
I have Junior 8 though I've never got to play it. I suppose it is very under-rated :)

DJ

Paul Cavezza
15-02-2010, 11:17 PM
Ok 2nd question, which program do you find best at giving natural language analysis? I have Deep Rybka 3 at the moment and apart from giving the occasional "this move gains space" (duh!) I don't get much out of it in that sense and I have to deduce for myself why it's rating of a position has changed.

Davidflude
19-02-2010, 11:56 PM
Deep Rybka is very strong but it has not got a clue in some positions. I distrust programs that give evaluations to two decimal places.

Paul Cavezza
24-02-2010, 09:31 AM
Hi David,

Where do you find it weak? Personally I think it might over-value space advantages... just an inkling from some Italian games I annotated using it.

BearDrinkingBeer
17-04-2010, 06:05 PM
Personally I think it might over-value space advantages... just an inkling from some Italian games I annotated using it.
I believe it is the most modest of the strongest chess engines. According to John Nunn, it is also the most balanced.

Kevin Bonham
17-04-2010, 09:36 PM
I find Fritz 11 is very prone to overvalue space very early in the game.

Paul Cavezza
28-04-2010, 03:11 AM
I would definitely agree with that Kevin. Pretty much any variation where you get in e5/d5 or something like that (in a c3 Italian game for example) and it always favours the variations where you push the pawn.