in-games rating

I don't know what the proper name is to these ratings, but when analysing my games, I see like ratings defining which part is taking advantage (positive = white, neg = black).

Would be curious to know what the calculation is, and what the variants are. I know -0.5 isslight advantage to black and ashite I still have chances, and -10, I only can rely on big blunder from black to equalize. but what does -1 (for instance ) means? 10% chances ? can we translate it and quantize it in probability ?

edit : sorry for the text format, my laptop is locked on insert function and I don't know how to deactivate it

Press insert again.

Those numbers represent Pawn-equivalents. +1.2 means that White has an advantage roughly equal to one and one-fifth Pawns.

-9.0 means that Black is the equivalent of a Queen ahead.

Thanks, no ins touch on my laptop, would be too easy otherwise.

Actually, I am speaking about this 

Yes, as I said that indicates what fraction of a Pawn the advantage is worth. In your pic, the engine feels that White has an advantage roughly equivalent to 1/6th of a Pawn.

@blueemo, sorry, after your reply I see that i misformulated the topic. Itùs not "in game rating" but more about "analysis rating". But I still don' t know what is real name of this...

oh ok. any clue on how it is calculated, or what is considered to get this very precise number which is, in this particular case, based on position and not any material advantage?

I know here is onmy first move, so the computer can easily compute statistics based on a database of already played games, but we still get this calculations on advanced parts of every games, wich by definition, are unique, and for balanced materials, it implies to compute positional advantages, which is my question.

It's just the result of the engine's evaluation function, normalized into Pawn-equivalents.

Naturally, the engine needs an evaluation function, or it couldn't decide that one move was better than another.

It considers all the usual things... Pawn structure, King safety, piece activity...

So... no possible way for us, mortal humans, to translate a -1, or -1.5, -10, etc, in mortal human words? 

I mean, I understand your explaination, and thanks for your time, but if I am translating it into a very personal way, it would be, like saying money doesn t make people happy :

I am receiving a survey, filling it in, and knowing afterwards that even if I am quite poor, my happiness is equivalent to 1.000.000 dollars. Cool, I am happy then. But I would like to know what factors the survey took in consideration... positional and material is really different, so how to put them in a serial calculation?

Stockfish, the best traditional chess engine out there, is open source. you could go and have a look at the source code if you are interested in the details.

No.

This is one of the key concepts that "engine monkeys" don't seem to grasp. Since top-ranking chess engines can defeat any Human, including the World Champion, some people think that we should be learning from engines instead of learning from other Humans.

The problem with that idea is that engines don't teach. They don't explain. There is only a limited amount of information that can be extracted from a laconic remark like "+0.72". A Human, on the other hand, can explain exactly WHY White is better in that position, and can offer both general and concrete suggestions on how to improve (or exploit) Black's play.