How does CMX ratings compare vs. human (across-the-board-o..

Len

Distinguished
Apr 6, 2004
149
0
18,680
Archived from groups: rec.games.chess.computer (More info?)

Is there an accepted formula for translating ChessMaster 10 rating into
human rating? At this point in time it is not convenient for me to seek out
face-to-face human appointments. But I am interested in how well I would
fair against rated players based on how well I am doing against CMX
 

JVMerlino

Distinguished
May 20, 2004
41
0
18,530
Archived from groups: rec.games.chess.computer (More info?)

It really depends on what rating range of personalities you are playing
against. In general, the ratings for the mid-range personalities
(1000-1800) are within 100-150 rating points of their USCF equivalent
(and they are typically too high -- meaning that a CMX personality that
is rated 1450 is likely to be equivalent to about 1300-1350 USCF).

The low-end personalities (less than 1000) play so randomly, for the
most part, that their ratings are simply estimates, as there are very
few people who play that way "in real life".

The high-end personalities (greater than 1800) also tend to be a bit
higher ranked than their USCF equivalents, but typically by less than
100 points, and usually no more than by 50 points.

Of course, this is based on my knowledge of how the personalities were
rated for CM7000 through CM9000. I don't think the new programming team
changed the personalities, nor the way that the ratings for them were
calculated. So it is a pretty good guess that everything in CMX is very
similar to what it was in CM9000.

jm
 

Ruud

Distinguished
Apr 9, 2004
39
0
18,530
Archived from groups: rec.games.chess.computer (More info?)

On the internet you see players with ratings of over 3000, because they play
against slightly lesser players all the time, so they gather points.
Engine fritz for instance mixxes in and gets wins all the time, getting
rating of 3200, sometimes.
That doesn't mean they are actually that strong. On the bottom half of the
lists are people with lower ratings than factually true.
The only way to find out is to try.

<JVMerlino@aol.com> schreef in bericht
news:1117907528.667565.274500@g14g2000cwa.googlegroups.com...
> It really depends on what rating range of personalities you are playing
> against. In general, the ratings for the mid-range personalities
> (1000-1800) are within 100-150 rating points of their USCF equivalent
> (and they are typically too high -- meaning that a CMX personality that
> is rated 1450 is likely to be equivalent to about 1300-1350 USCF).
>
> The low-end personalities (less than 1000) play so randomly, for the
> most part, that their ratings are simply estimates, as there are very
> few people who play that way "in real life".
>
> The high-end personalities (greater than 1800) also tend to be a bit
> higher ranked than their USCF equivalents, but typically by less than
> 100 points, and usually no more than by 50 points.
>
> Of course, this is based on my knowledge of how the personalities were
> rated for CM7000 through CM9000. I don't think the new programming team
> changed the personalities, nor the way that the ratings for them were
> calculated. So it is a pretty good guess that everything in CMX is very
> similar to what it was in CM9000.
>
> jm
>
 
G

Guest

Guest
Archived from groups: rec.games.chess.computer (More info?)

Ruud <rhilll@home.nl> wrote:
> On the internet you see players with ratings of over 3000, because they
> play against slightly lesser players all the time, so they gather
> points. Engine fritz for instance mixxes in and gets wins all the time,
> getting rating of 3200, sometimes. That doesn't mean they are actually
> that strong.

What do you mena by ``actually are that strong''?

Firstly, ratings only allow you to compare the strength of people within
the same pool of players. Ratings given by, e.g., the USCF and FIDE
cannot be directly compared because there is relatively little overlap
between the two fields of players. The best that can be done is to work
out the correlation between the two ratings by looking at the players that
are rated in both systems. This is only an approximation.

These 3000-rated players really are that strong but only relative to the
other players on the same system. As noted above, you can't just say that
Kasparov is rated 2800 [by FIDE] and NetDude is rated 3000 [by some
internet site] and therefore NetDude would beat Kasparov wih ease. What
you can say, though, is that NetDude will score about 95% against people
rated 2500 on the same server.

Notice that you don't gain any rating points for beating somebody very
much weaker than you so it isn't possible to gain a 3000 rating by beating
any number of 1000 players.


Dave.

--
David Richerby Edible Tool (TM): it's like a hammer
www.chiark.greenend.org.uk/~davidr/ but you can eat it!