Actually the line you (and engines) give after 27.Qxc4 Nf3, culminating in 36.-b3 "there's no good answer to ...Ba3, and the b-pawn promotes" continues. The b-pawn doesn't promote but black is still winning: 37.Nf6 Ba3 38.Nd7 Bxb2 39.Nc5 - white is just in time to stop the b-pawn but 39.-Bxd4 40.Nxb3 Bxf2. And this - two connected passers for black - is of course resignable.
Not to "correct" you, I just wonder whether there is any "logic" to black winning in such a way, made possible by three white pawns (b2, d4, f2) on dark squares. Who would predict the game to end like this?
Actually the line you (and engines) give after 27.Qxc4 Nf3, culminating in 36.-b3 "there's no good answer to ...Ba3, and the b-pawn promotes" continues. The b-pawn doesn't promote but black is still winning: 37.Nf6 Ba3 38.Nd7 Bxb2 39.Nc5 - white is just in time to stop the b-pawn but 39.-Bxd4 40.Nxb3 Bxf2. And this - two connected passers for black - is of course resignable.
Not to "correct" you, I just wonder whether there is any "logic" to black winning in such a way, made possible by three white pawns (b2, d4, f2) on dark squares. Who would predict the game to end like this?