CEE CAPS2

What is this CEE CAPS2 that I see when I mouseover the accuracy score:

I have the same question.

Better Than Ratings? Chess.com's New 'CAPS' System - Chess.com

Yeah - that's all I could find too, but it dates back to 2017. Then they abolished it I believe. So I guess this is a new incarnation of it, strange to see no post or description though.

I believe it is only in testing. It actually was gone the last time I looked and doesn't show on mobile web.

?? Chess Estimation Engine Computer Aggregated Precision Score version 2 ??

either close to that maybe or 

Cheetos entertainment experience commercial advertisement project simulator

version 2 obviously

This video dated today (March 31, 2021) addresses the launch of CAPS 2. 

Thanks! A bit of a letdown though - he managed to speak for three minutes and say nothing

It currently only applies to Guest accounts, perhaps they forgot to remove it.

It applies to everyone that has signed up for Beta testing w/ Chess.com. As Danny indicates, it will be a revised Computer Aggregated Precision Score. He mentions that this system will "be based less on the extremes of good and bad moves."

Recently, I had an opponent play 12.3% per CAPS1 and 58.25% per CAPS2. In the game, while they made 3 blunders, they were heavily penalized for 2 inaccuracies and 3 mistakes. Hope this helps. As you can infer, it is far less critical of mistakes. 

So, what is then more useful to take a look at ?

I sometimes watch lessons on youtube and even GM Aman was amazed that he played a game, controlling the game complete and then having an accuracy  of 33 procent.
Also fun to see when you use the system he explains to just use basic rules, the computer is telling me that they are mistakes.
Kinda odd sometimes how the computer and a GM are in conflict

Yes. There are two main issues with CAPS 1 that I should highlight. I think aggregating relative to the computer best moves does not take into account a lot of practical issues that result in weirdly inflated or deflated scores. Hopefully problems such as the ones I listed below will be resolved with CAPS 2.

1) CAPS tends to be MUCH more kind to games with a higher number of moves. Some of my god awful games got 70+ CAPS scores just because they lasted for 50 moves or more, while some 20 moves where I dominated my opponent from start to finish barely broke 50.

2) CAPS does not change the way it weights move based on the engine evaluation in the current position. I should not be penalized for making a mistake in a +7 or -7 position than I should  in a +0.5 or -0.5 position.

3) Your CAPS score is often severely penalized for not making a K + Q vs K checkmate or a K + R vs K checkmate in the lowest number of possible. You should not be getting missed wins for a move that has you +61 when could have had mate in 7. However, this a regular thing that happens. Most players on the site cannot not find a mate in 7 in one of these checkmates and will therefore gladly settle on a 15 or 20 move checkmate as long as they know the checkmate will work. I don't think these mistakes should be penalized at all as they do not matter towards the outcome of a game.

Caps isn't that useful unless you review your game.  Caps is just an encouragement to analyze your game and lead you to videos and  learning aspects of the website...and perhaps a way to slow down the number of live games on the server as people look at their games.   Remember when Chess.com went down because of their y2k type bug tied to the total number of games played?  Ahhh good times.

i always review my games, but it is confusing to follow lessons online, for example, some GM`s tell it is better to wait to capture a pawn, when you not developed all of your pieces, but the computer tells you that it is a mistake.

But i admit, i learn from it, but it is sometimes real dumb to see that you have an accuracy of only say, ( it is just an example ) 33 %, but the Caps is saying 79 % and for my feeling and also as a result, a victory.

i will be the first person to say that my games need real improvement, but it is real hard when you got in between good advice from youtube for example, but the computer says, nah, that is a bad move.
then it feels good to look at your Caps and it says that the game wasn't that bad at all.

There are many problems here:

1. Strong players often give general advice but call it "rules". It's not rules, and it's not as generally applicable as they say, even if they aren't saying it as poorly as most do.
2. What strong players say is often taken out of context. For example, GM Igor Smirnov is all over the place saying "To take is a mistake" but what he really means, when you go to the original source where he first said that, is that if you cannot determine whether making an exchange will definitely improve your position, err on the side of not making it. In most cases (at master level - maybe not at yours), that's the right decision, and when it's not, you get a lesson to help you improve a similar decision later on.
3. Players often (I could probably say always) overgeneralize what strong players say. A strong player may be explaining a position and make a comment about how this or that is often the case in positions like this, and then weaker players go off and apply the same conclusions to other, completely different positions thinking they are similar.
4. Strong players give advice usually assuming that the player receiving it has mastered elementary material chess. They assume that you ALMOST NEVER miss forced checkmates or lose material by leaving it en prise or to simple tactics, or fail to take advantage of the same errors by your opponents. The reality is that until you reach that basic level, nothing else matters at all, so at first ignore everything but eliminating material blunders.
5. General advice should be used not as ready made conclusions to be repeated to avoid having to think for yourself, but as an aid to your thinking process. What really matters is your own concrete calculation and evaluation - there is no substitute for that.
6. The engines could be wrong. They often are, depending on the particular circumstances.
7. The engine may see something concrete many moves in advance that the GM cannot see, and thus that neither you nor your opponents could possibly see, so ... you need to decide whether you'd like to play the game like a thinking, imaginative human, or like an unthinking, unimaginative engine. We have overcome our deficiency in calculation (compared to computing engines that do nothing else) by inventing general guidelines, etc. That comes at a cost. Live with the tradeoff.

In 1972 I played black in this game.
https://www.chess.com/a/35ZDr7Fwctyz6

A few months ago I analyzed this game, and CAPS only gave me 36.4 for it.

Today I would see what CAPS2 (with mouse-over) gives me.
Then I noticed that the old CAPS had changed to 45.9.
And CAPS2 gave me 80.1!

Something is happening...!!

I have compared the CAPS1 and CAPS2 scores for about 30 or so different games now.

In every case CAPS2 has given scores that are more reflective of how well any reasonable human would say that the players played the game.

The original CAPS is badly flawed, the new version is not perfect but it is significantly better. It is time to retire the old broken scoring system and replace it with CAPS2. Or at least give people the option of which scoring system they would like to see.

Was the CAPS 2 just removed for anyone else as of today?  I was really liking the CAPS 2 valuation method as I found it to more closely align to the "common sense" evaluation of the play.

Yes, CAPS2 is gone now.
Let us hope it will return soon!