Stockfish incorect evaluation?

I was going through my opening lines since I was having a bit of a dry spell with my favorite opening (ponziani), when I came across this EXTREMELY peculiar evaluation.

I would say white is not doing well here and actually managed to "lose" the opening despite a technically favorable line. I literally have no development, yet the engine gives me almost a +5 in this position.



Edit: I was intrigued by @ThunderClap 's comment about centerpawns, so I decided to gift black a d7 pawn to check evaluation. Lo and behold, it is now +1. The 3.5 point pawn lmao.

eek! Dude

I'm not quite able to identify why white is so good in this position as material is equal, but maybe the evaluation is high because white is a pawn up and has the bishop pair? Perhaps the engine evaluation is slightly optimistic in human terms as the engine only thinks in terms of close to perfect play and it feels confident about this position. Although in practice the opening database shows that this position has a win rate for white of about 70%, so maybe it is accurate after all.

Black will have NO CENTER (Center pawns are gone for Black) & White has THE TWO BISHOPS & an EXTRA PAWN to boot !

Yeah I think I agree. Maybe stockfish thinks the endgame is easily traded down to and then won? One thing I noticed is that black's pieces, despite their seemingly good looks, work horribly against white's structure. The dark square bishop gets shut out as white moves all his pawns to the dark squares. The c6 knight has no future after d4 and simply blocks black's move c5. At MAX I (as a human) would give it a +2.5, but maybe these weaknesses are too much, and perhaps there is more I am missing.

All stockfish ratings are accurate. They are never inaccurate. It's like an adding machine, it does exactly what it's told to do.

But some +3 positions are easy wins and some +10 positions are drawn. It's up to you to interpret the results of the adding machine.

It's +4.27 at 13, and +4.58 at 33, and +4.42 at 36. At 36 stockfish looked at 1.6 billion nodes. In other words about 500 times deeper than the end-of-game analysis that LiChess does.

The way I interpret those numbers is to say it's going to be very difficult for white to make any progress over at least the medium-term assuming perfect play from black.

Normally if the advantage stops growing at some point and in fact starts to drop I assume the position is drawn. Typically if a position is winning the advantage grows as more moves are played and never turns down.

Without a tablebase it's impossible to tell for sure, but seeing a +4.4 advantage and believing it's a draw doesn't tell me that the evaluation is wrong.