Stockfish 14+ NNUE or Stockfish 16?

You can trust SF 16. No doubt about it. It's also using an NNUE file, although you have to make sure that that exact file is in the same directory as the executable. (tests.stockfishchess.org/nns)

also stockfish evaluation changed between sf14 and sf16 and is no longer tied to the value of one pawn, see sf 15.1 release note: stockfishchess.org/blog/2022/stockfish-15-1/

and a quick addition of one more eval to the bowl, sf 20230706, 16 gb hash, depth 50:
rnbq3r/ppp2kpp/5n2/2b5/2P5/8/PP1P1PPP/RNBQKBNR w KQ - 0 6 acd 50; acs 0; bm Nf3; ce +142; Ae "Stockfish dev-20230706-e699fee5";
that means +1.42 for white
not that it matters ;)

Thanks to everyone who took their time and replied to my post. I have concluded that I must have downloaded some corrupted version of SF 16 as right now I'm looking at a position where all other engines says it's +15 and my SF 16 is like, nah.. it's just +2 :D

"So which engine should I trust?"
It's like a Lamborghini or a Ferrari? And in reality you would be fine with a Dacia.

it uses a different evaluation method, for stockfish 14, +1 means 1 pawn up in material, but for 16 score has nothing to do with material, it represents the winning probability

@hchh25 said in #15:
> it uses a different evaluation method, for stockfish 14, +1 means 1 pawn up in material, but for 16 score has nothing to do with material, it represents the winning probability

And what exactly is the reason for that? The score was always calculated in terms of pawns so why the change? I don't think any chess player thinks in terms of winning probability.

@Hagredion said in #16:
> And what exactly is the reason for that? The score was always calculated in terms of pawns so why the change? I don't think any chess player thinks in terms of winning probability.
Because the goal is to win right? Being one pawn up sometimes is a draw and sometimes is a win, have to consider other things

@Hagredion said in #16:
> And what exactly is the reason for that? The score was always calculated in terms of pawns so why the change? I don't think any chess player thinks in terms of winning probability.

"Normalize evaluation

Normalizes the internal value as reported by evaluate or search
to the UCI centipawn result used in output. This value is derived from
the win_rate_model() such that Stockfish outputs an advantage of
"100 centipawns" for a position if the engine has a 50% probability to win
from this position in selfplay at fishtest LTC time control.

The reason to introduce this normalization is that our evaluation is, since NNUE,
no longer related to the classical parameter PawnValueEg (=208). This leads to
the current evaluation changing quite a bit from release to release, for example,
the eval needed to have 50% win probability at fishtest LTC (in cp and internal Value):

June 2020 : 113cp (237)
June 2021 : 115cp (240)
April 2022 : 134cp (279)
July 2022 : 167cp (348)

With this patch, a 100cp advantage will have a fixed interpretation,
i.e. a 50% win chance. To keep this value steady, it will be needed to update the win_rate_model()
from time to time, based on fishtest data. This analysis can be performed with
a set of scripts currently available at github.com/vondele/WLD_model"

well I think stockfish 16 is the best

@pointlesswindows said in #14:
> "So which engine should I trust?"
> It's like a Lamborghini or a Ferrari? And in reality you would be fine with a Dacia.
I am fine with UAZ too