CAPS Conversions

How to get the conversions from CAPS to projected elo?

Read this article. It talks about how CAPS replaces rating ( and then gives a suggested conversion chart).

https://www.chess.com/article/view/better-than-ratings-chess-com-s-new-caps-system

You may have seen it but you evidently did not understand it. That chart is the best information that has been released to date.

Ok.

Using the table you can do a linear regression. You get Estimated rating= CAPS*60.2-3375

Thanks

For what they say, the important is the average and not the CAPS of one specific game.

Try reading the chart again. It suggests 96 corresponds to 2400.

But then again what lparreira said goes too, it is not meant to rate an individual game.

Is your CAPS average 96? I doubt it.

LOL, I'm off the chart.

Not really. You only play daily and I assume you make use of databases.   

I believe that CAPS should exclude the first 25 or so moves you get from a database. Your 2 inaccuracies might get you a different CAPS rating for the game.

Hmmm... I was wondering since in this game I just played I had CAPS 98.9 yet I was losing for over half the game until my opponent blundered.

https://www.chess.com/analysis/game/live/3614610328

The other question to consider here is if the conversion is really linear.

Btw, using linear regression is pretty problematic for this.

Notice that the formula would give you a rating of -3375 at 0% accuracy which makes very little sense. In fact, any accuracy below 56% gives a negative rating. Basically, this model gives a horrible downward skew to values lower than the dataset chesscom provided.

It is better to model with a polynomial recognizing that no elo values less than 0 should be given. This gives you something more like 

Re = 2.05+12.9*Acc-0.256*acc^2+0.00401*acc^3.

Where Re is the estimated rating and Acc is the CAPS accuracy score. This seems to be a much better conversion. 

This model is still pretty imperfect and pretty far from Chesscom's data right at 1200 elo, however, it seems like a much better overall model. 

But that gives my game (above) an estimated Elo of 7752.

My model only gives you 2693 Elo. And of course, that would only be your elo performance in that game, not a reflection of an actual rating. 

And the whole thing is changing soon anyway since the new version of CAPs will be released soon (it's already available in beta and site-wide for premium)

I wonder if this could be expected to change the conversion? 

I would think so since the Accuracy numbers are going to be different. I don't know what the magnitude is overall and how that will translate, but I would expect some differences.

https://support.chess.com/article/364-how-does-the-game-report-analysis-work

https://www.chess.com/news/view/chesscom-releases-new-game-review

Interesting. Sadly, I doubt we'll get much data on the conversion (even the original conversion was pretty vague) so it will probably be difficult to find a way to independently calculate the elo corresponding to the accuracy score. And if chess.com doesn't provide something there may be no way to perform this conversion at all. 

Hmm