Chess Mentor Ratings need revamping...

Hi Wildman,

I've stated many times that I understand that CM is a teaching tool...and it is an excellent one.  I'm only suggesting that there is much room for improvement in one aspect of it.

If your only goal is to expand your understanding of Chess, then ratings are never NEEDED, per se.  Some people have probably never played a rated game in their life, yet study and still find time to improve their game. 

But for the vast majority of players who play tournament Chess, ratings are an important means (and often only way) by which to measure progress and see how they stack up against other players. 

If the rating system used for CM will not allow you to see how you are performing relative to other chess.com members, then the ratings should at least loosely mirror one's regular chess rating (give or take 300 points).  I'm fairly certain the CM staff's tried to apply (subjective) difficulty ratings that they felt were as accurate as possible in comparison to OTB ratings.  Given the widely varying opinions here, it's evident that there is a lot of room for improvement. 

Hi Careyfan,

Yes, but in the case of ratings against other players we have a fair idea of what is being rated. 

In the case of something like Tactics Trainer, both the  problems and the players involved are being rated as if they were playing a series of games, so the pool of players plus the clock keep the ratings somewhat meaningful.

In the case of Chess Mentor you're looking at a series of lessons where the lesson has a fixed subjective rating, percentage points get knocked off in subjective amounts for wrong answers, you can get hints if you get stumped, etc.  It isn't clear what, if anything, is being measured by a rating.  So why bother?  It's like reading a chess book and then asking what your new rating is.  If you're really learning anything then it will translate into an improvement in your ratings against other opponents;  if you're not it won't.

 I agree. I want an indicator of my playing sdtrength which is reliable. I actually feel that my level is somewhat higher than it should be, therefore, Chess Mentor should have been there as good approximator for my level, but fails to do so. I would rather it was hard than easy!!

Scanning the posts on this topic so far, I didn't see this suggestion: Increase the penalty for wrong answers. In a real game, the wrong move can cause a loss in a completely won position. Most training programs don't seem to realistically penalize mistakes, and subsequently give players these inflated "ratings." That number has no correlation with OTB play, but it would make puzzles more challenging (increased psychological pressure) if the penalty for a mistake were more severe.

I like the way Mentor ticks down the success percentage as time goes by, and I've enjoyed the lessons I've tried. Nice program.

i agree that the ratings are way too high on chess mentor, but does it really matter? ive started from the very first lesson and all i have done so far is make legal moves, castle, some very simple tactics.....my rating? 2420 LOL :-)

Wow, That is impressive and probably well deserved!

Gumpty, your rating is disappointing :)) When I went through the CM courses not on the CD, doing only the lessons rated >1600 and never using "correct square" or other such help, I got to over 2700 :)) I imagine that if you start from mate in one or opposition, you'll get over 3000.

I guess that Erik simply doesn't know how to balance these ratings. And I don't think anybody has an idea. Note that a begginer must be encouraged a little bit, you cannot simply give him a negative score consistently.

Heres a suggestion.  Eliminate the chess mentor ratings per user and just put an accuracy avg.

2555 now :)

I too largely ignore ratings from Chess Mentor, etc. I just want to sharpen up my tactical skills and expose myself to new possibilities. However, its nice to get a high rating - I am hoping to beat Topolov's 2812. 

i already have :-)))

My Chess Mentor rating is about 600 points higher than my blitz rating.

(see full progress report)

Yes, the Chess Mentor ratings should not really be compared to your official Chess ratings. It should be considered separately from other rating systems.

Or, careyfan, the Chess Mentor shouldn't give a quantitive but a qualitive rating-instead of a number, say something like GM-strength.

Oddly enough, I take the lessons to learn chess.  The "ratings" are just a form of positive feedback which I appreciate, just as when my hybrid car flashes "excellent" when I have good mpg on a trip.   It is not exactly meaningless but I would never expect that it would correlate with over the board strength.

OK, so I was looking for info on how my Chess Mentor rating should compare to my turn-based Chess.com rating, and I found this (oldish) thread. Before I go on, I should emphasize that I really do appreciate CM - many of the tactical and strategic concepts I have learnt have strengthened my game and certainly (in my mind) 'justify' the annual subscription I pay.

Back to the point: CM (using the 'adaptive' mode) rates me fairly consistently at around 2150 (up or down by 10 or 15 points over time) - and I must admit I find this quite flattering. However, I have played games on-line against 2150 and 2200 rated guys and, quite frankly, I was completely outclassed (words like "hammered," "cuffed" and "don't call us, we'll call you" come to mind) ... with the exception of one fine game where my opponent fell for the Elephant Trap :-).

My on-line rating, which is now reaching a sort of stability, bounces around between 1800 and 1900 or so (and I definitely think my game has improved over the past 6 months due to my use of Chess Mentor).

I think one of the problems is - as has been said already in this thread - that the mentor is too 'forgiving' of wrong moves: in a real game, there is no "takeback-and-try-again" option.

I don't have any sort of official OTB rating, but I've heard that for strongish players, Chess.com scores are around 250 or so points higher than OTB; maybe there's a similar 'fiddle factor' we can apply to on-line vs mentor ratings. Any ideas?

Look at my chess mentor rating 0-0