I was analyzing a recent game of mine, and I noticed something interesting. After I blundered back a bishop on move 16, the evaluation was -3.4, which is still strongly winning for Black.
[White "White"]
[Black "Black"]
[Result "0-1"]
[WhiteElo "1700"]
[BlackElo "1500"]
[startply "31"]
[fen ""]
1. e4 c5 2. f4 { B21 Sicilian Defense: McDonnell Attack } b6 3. Nf3 Bb7 4. Nc3 a6 5. Bc4 e6 6. O-O d5 7. exd5 exd5 8. Be2 d4 9. Ne4?? { (-1.03 → -4.31) Blunder. Nb1 was best. } Bxe4 10. d3 Bb7 11. Ng5 Be7 12. Bf3 Nc6 13. Re1 Nf6 14. Qe2 O-O? { (-5.42 → -3.17) Mistake. Qd7 was best. } 15. Bxc6 Bxc6 16. Qxe7 Qd5 17. Nf3?! { (-3.25 → -5.04) Inaccuracy. Qe2 was best. } Rae8 18. Qc7 Rxe1+ 19. Kf2? { (-4.63 → -19.29) Mistake. Nxe1 was best. } Ree8 20. Qxb6 Ng4+ 21. Kg3 Ne3 22. Qb3 Nf5+ 23. Kf2 Qd7 24. Bd2 Bxf3 25. gxf3 Nh4 26. c4 Qh3 27. Qd1 Qxh2+ { White resigns. } 0-1
How is this so, given that the material had evened out? Obviously, if it's not due to material, it must be due to something positional, but even then I'm not very convinced. After move 16, White owns the e-file and maybe even the 7th rank. They also has a knight on g5 which, to me, is a bit worrisome (should it not be?), and it's supported by a pawn and a bishop. What merits such a strong evaluation for Black?