Game Review Game Rating

Hi All,

After going over a game review, I see there is a estimated rating given by the computer based on the moves played in the game. My questions are how accurate does the community feel this rating is for a single game? If you are noticing a trend during these reviews, would you agree this is close to what your office rating would be for USCF or FIDE? Any thoughts on this topic would be great, Thanks!!

A great question that i have often wondered myself. I have been playing on and off for years, but never really dove in until now. I've aeemed to have leveled out my skillset and striving fot improvement.

Not accurate at all.

I had posted the below comment in another forum, but I move it here because I also am curious about these game ratings. Maybe someone here has some insight.

. . . There's another issue in the new "game review features" involving the "performance rating", which claims to be "an estimate of your level of play based on that single game". I am a beginner / intermediate player, and I recently played 17 games against the Komodo9 computer, rated supposedly at 1300. Across my games, my individual performance ratings ranged from 1100 - 1650, with an average of 1332. Fair enough. The computer, however, had individual performance ratings ranging from 800 - 1350, with an average of 1053. It only matched or exceeded 1300 in 2 of the 17 games. How is it that a computer "rated" for 1300 only plays at an average rating of 1053? Is the computer rating too high or are the individual game performance ratings too low?

Both are too high (compared with official USCF or FIDE rating methods), but the computer's rating is inflated hundreds of points relative to the player's rating.

Not accurate at all. I ran through one game I had played as a guest but saved as a pgn the other day. I was surprised and a little delighted when it gave my performance in that game 2700! I should confess that in the real chess world I have never been higher than 1850 FIDE more or less. So, although like all of us at one time or another, I have had some games when I've felt I'm on fire and played above myself, I don't think I've ever produced anything in the master, let alone grandmaster, class, as that game review rating would suggest.

1850? I don't mean to be a downer, but 1600 on chess.com is not even close to 1850 FIDE (which should be closer to 2000 USCF or approximately 2300 chess.com).

Personally, I've noticed that game review ratings tend to be closer to chess.com ratings than OTB ratings, because it seems that the ratings are definitely inflated by quite a bit.

I'm pretty sure that Chess.com ratings, as @EndZoneX has stated, are definitely inflated by a lot more than OTB ratings. If you think about this; many chess players have exceeded the 3000 rating boundary on Chess.com, but the highest rated player is only 2800+ OTB.

However, if you are considering the game review rating in terms of Chess.com ratings, then yes, I am pretty sure it is somewhat accurate to what you played like in terms of rating.

I do find it, however, unbelievable that I could ever play a game like someone rated 1900

I've only played, I think, six games on here while logged in, none of which I was taking very seriously while trying to cope with my misbehaving four-year old at the same time, so I don't think my chess.com rating (and those ratings, I give you, are inflated) should be taken as statistically significant. I lost three of those six games as I recall. I should add that my FIDE rating was a good while ago when chess seemed to be more important than it later became to me, and I wouldn't claim to be anything like that now. Happy to play you, if you're willing to risk some of those oh so important chess.com rating points to a 1600.

Regardless of whether chess.com ratings are high compared to ratings elsewhere, one would think that a chess.com computer rated at, say 1500, would actually play games at a chess.com individual game rating of around 1500. Instead, we see individual game ratings for the computer ranging by 500 or more points, with few games rated as highly as what chess.com rates the computer. Maybe the computer engine adjusts itself somewhat to the lousy play of the opponent rather than playing at a fixed level. Still, there is a huge disconnect here.

This is most likely due to how many players on Chess.com can play games in such a way so as to preserve their rating, making it only go up instead of down.

Or, at least, as much in that way as possible.

For the most part, I understand the new game rating feature. When I play unusually we'll, the rating is higher than my current rating (circa 1300), and vice versa when I play poorly. But sometimes I see a very low number (100-200). This seems to happen when I score a lopsided win, so it cannot be my game elo. Has anyone else seen this? Any idea what it is supposed to mean?

I wonder if the Game Review ratings are meant more as an incentive or reward than an accurate play assessment. On this day, in this game, your skills were comparable to those rated 1600, for example. That's either good news or letting you know you're having a bad day.

It does not seem to be accurate at all. At least not in the 15|10 chess.com player pool.

Most games I get a performance rating well above my actual rating and if I play at a level the computer thinks is accurate (performance rating close to my actual rating or below it), I will most likely lose the game.

So I'd say it gives performance ratings that are a couple hundred points higher than they should be.

For example the last game I played it gave me a performance rating of 1850 and my opponent 1600. I won the game, but in reality I am around 1400 and so was my opponent. Of course if this only happened once or rarely, it would not say much, but it seems to be this way most games I play (both players with performance ratings above their actual ratings).