Accuracy Score in the Analyzer

Sort:
dcb1970

First let me say how terrific the analyzer is. The "retry" mistakes feature and ease of use of all the functions is really first rate. I do look forward to an API offering where I can pass a pgn with N games in it, and get back a nice .pdf or .xls file with stats and reports for all N games.  Anyhow, the question:

What is the range of reasonable accuracy scores to expect in a slow chess game?  I ask because I ran my 3rd graders quads through it today, and 2 of the  games he had 95 and 97.5 accuracy scores. His opponent in one had 55, another 92.  All the ELO ratings are 1400 - 1700.  Curious what one expects for this metric.

blueemu

The CAPS accuracy metric is statistical, so the score for a single game doesn't mean much. It might not mean anything at all.

The average nuclear family in the USA has two-and-a-half kids, but if you went into a neighbor's house and saw two-and-a-half kids on the couch, you would probably find it disturbing, not normal.

There are far too many factors that affect the CAPS accuracy... whether or not the two players follow a known book line, how far out they follow it, whether the play is relatively forced (with few options for either side), whether one players makes several blunders (it doesn't take Bobby Fischer to see that capturing a free piece is a good move), etc.

The CAPS metric is only meaningful if it is applied to a large sample of games by the same player.

dcb1970
Agree it is statistical. So over a large enough sample, what would equally matched players get as their CAPs score, ELO 1500 / 1800 / 2000 etc?  Any sense?
 

 

blueemu

Prometheus_Fuschs
blueemu escribió:

 

So I have played like a "Typical 2400 OTB" before.

moo6o6

These are average xCAPS not from one terrific game

blueemu
Prometheus_Fuschs wrote:

So I have played like a "Typical 2400 OTB" before.

In one game, the CAPS result is meaningless.

If you do a least-squares regression curve-fit on the data, and then plug in my CAPS score from this game:

https://www.chess.com/daily/game/200865884

... you'll get the equivalent of a 2808 performance rating. 

Means nothing.

blueemu
ghost_of_pushwood wrote:

Which is why he never even bothers to bring it up.

Not more than once or twice a week, anyway.

Prometheus_Fuschs
blueemu escribió:
Prometheus_Fuschs wrote:

So I have played like a "Typical 2400 OTB" before.

In one game, the CAPS result is meaningless.

If you do a least-squares regression curve-fit on the data, and then plug in my CAPS score from this game:

https://www.chess.com/daily/game/200865884

... you'll get the equivalent of a 2808 performance rating. 

Means nothing.

Which begs the question, why isn't there an automatic average CAPS score in the statistics section? It'd be interesting to see its reduction in different time formats.

blueemu
blueemu wrote:
ghost_of_pushwood wrote:

Which is why he never even bothers to bring it up.

Not more than once or twice a week, anyway.

OK, maybe three or four times a week, tops.

This is the essence of modesty!

iamnotanumber7

How many games make a statistically significant sample for measuring CAPS? 

Should we expect a player's mean CAPS score for one time control to be similar to their CAPS score for other time controls? eg Blitz vs. Daily

I'd be interested in members' opinions and experience of CAPS. I'd be even more interested in an explanation from Chess.com staff.

 

 

Prometheus_Fuschs

@blueemu

I've got the same CAPS score in a recent game of mine!

 

wollyhood

Yay am not sure if I should post this here or start a new thread, but I got 92% accuracy in this game and yeh the analyzer wants me to move my knights all the time, and castle only when they want me too, but very pleased to get that win so early in the game. https://www.chess.com/daily/game/230615272

I wish I wish I wish I could get a brilliant move one day, but is it really going to be brilliant, or just luck that someone like me does a move and it gets called brilliant?

Prometheus_Fuschs
wollyhood escribió:

Yay am not sure if I should post this here or start a new thread, but I got 92% accuracy in this game and yeh the analyzer wants me to move my knights all the time, and castle only when they want me too, but very pleased to get that win so early in the game. https://www.chess.com/daily/game/230615272

I wish I wish I wish I could get a brilliant move one day, but is it really going to be brilliant, or just luck that someone like me does a move and it gets called brilliant?

Getting a brilliant move on the analysis here is only a matter of time.

wollyhood

Does time have a bank account? Give-a-Little Page?

Prometheus_Fuschs
wollyhood escribió:

Does time have a bank account? Give-a-Little Page?

Yes, you have to deposit games in it to (eventually) get the brilliant move, regardless, I wouldn't mind this too much.

wollyhood

maow I really want it but yeh just a pointless desire

wollyhood

YAY!!! first brilliant move that I have found in my games:

https://www.chess.com/analysis/game/daily/231112722

no donations required!

Prometheus_Fuschs

One thing off the bucket list.

blueemu
BEASTmode9898 wrote:
blueemu wrote:

 

What time control is this?

Postal. Three days per move.

Oh, wait... maybe you meant that CAPS vs Rating chart, not the game I posted... standard (slow) tournament time controls.