Analysis by chess.com

Here is the important ingredient: upon release of Stockfish 15.1 going forwards, evaluation interpretation has changed. An evaluation of +1.0 no longer means White effectively has an extra pawn. Rather, an evaluation of +1.0 means there is a 50% chance to win (for the engine to play against an equal opponent).

By the time there is an eval of +2.0, this means almost 100% certainty that White is winning. Extending that to your game's eval of +5.0, Stockfish is saying it is practically impossible for the position to be anything other than a win for White.

Here is the official documentation: https://github.com/official-stockfish/Stockfish/wiki/Stockfish-FAQ#interpretation-of-the-stockfish-evaluation

Thanks for your feedback- I'm not sure we're talking about same analysis score though? I'm talking about analysis score that can get as high as 35 on chess.com, before becoming M3 or M1 etc

I guess it depends which engine you are running in your analysis. If you are running Stockfish on Chess.com, then we are talking about the same analysis scoring method. If you are running something else (like Komodo), then it is likely scored differently.

For something as extreme as +35 (regardless of engine), then it simply means the engine know that is about to win, but it has not found a forced mate yet.

I'm not sure the site is using the winning expectancy values , even with Stockfish 16 NNUE

When Chess.com made their official announcement, they pointed out the expectancy value. If they are not actually using this, then clarification would be greatly appreciated.

Article: https://www.chess.com/news/view/stockfish-16-on-chesscom

I asked and it was said that whatever Stockfish provides is what's used. That said, it appears that Stockfish may actually be using some conversions to output something more similar to a regular evaluation scores.

https://thechessmind.substack.com/p/news-you-can-use-stockfish-normalizes

So, if that's correct, it's not outputting winning expectancy scores.

I am sad. I wish I knew which evaluation Chess.com is outputting - especially since it pointed out the "new" evaluation method upon integration with the site.

I haven't read this but staff mentioned this as well, which was said aligns with the above article.

https://github.com/official-stockfish/Stockfish/discussions/4150

Can you show an example position from one of your games? I am interested.

Check out this #chess game: sibkk vs patimaritimissimitosan - https://www.chess.com/live/game/87830584309

I can't recall example where I.

I can't recall specific example where one player is ahead massively in late endgame, but only given advantage of 5 or 6, but it seems to happen very often