More Ponzing About

ROBIN Moss (2269/201) has given me his analysis of what he reckons is a critical line in the Ponziani - a line in which Black sacrifices two, and sometimes even three, pieces.
1.e4 e5 2.Nf3 Nc6 3.c3 Nf6
Black's most popular move. 3...d5 is generally considered sharper but, as we shall see, 3...Nf6 can easily become extremely sharp if White follows the main line.
4.d4 Nxe4 5.d5 Bc5!?
The knight-retreats 5...Ne7 and 5...Nb8 are much more popular in ChessBase's 2019 Mega database, but the text, which seems to have first been played, unsuccessfully, by Ernst Falkbeer in 1855, has been tried by GMs and scores a better percentage for Black.
6.dxc6 Bxf2+
Falkbeer, in a match against Robert Brien, played 6...Nxf2. More than 20 years were to pass before Mikhail Chigorin used the text to beat Eugen von Schmidt, also in a match, in 1879.
7.Ke2
White's king is short of squares and his development is disrupted, but he has a knight for two pawns, and Black's two developed minor pieces are precariously placed.

How should Black proceed?

*****
*****
*****
*****
*****
7...Bb6!?
More popular is 7...bxc6, which gives Black a third pawn for his missing knight. However, White can reply 8.Qa4, which prevents 8...Ba6+ and at the same time attacks e4. After the further moves 8...f5 9.Nbd2 0-0 10.Nxe4 fxe4 11.Qxe4 d5, White is obliged to offer his queen with 12.Qxe5. However, 12...Re8 13.Qxe8+ Qxe8+ 14.Kxf2 leaves White with rook, bishop and knight for queen and pawn, which has proved very good for White in games such as Jonny Hector (2535) - Peter Heine Nielsen (2420), Tastrup 1992 (1-0, 23 moves).
However, RM recommends the text, which is preferred by the analysis engines Stockfish10 and Komodo10. One point is that not only is Black's dark-square bishop returned to safety, but Black again threatens a knight fork at f2.
8.Qd5 Nf2
Saving the knight while gaining a tempo on the king's rook.
9.Rg1
White has a major alternative in 9.cxb7!? Bxb7 10.Qxb7, when 10...Nxh1 gives Black rook and two pawns for bishop and knight. The position is unclear, but the engines slightly prefer Black.
9...0-0
The only continuation given by RM in his analysis, but the engines also quite like a move that is not in Mega19, 9...dxc6!? The further moves 10.Qxd8+ Kxd8 11.Be3 Ng4 12.Bxb6 give another unclear position in which Black has three pawns for a knight.
10.cxb7 Bxb7 11.Qxb7 Qf6!?
This may be a new idea (the only game to reach this position in Mega19 saw a 2389 play 11...e4).

RM treats this as the critical position for the whole variation, giving four possible continuations for White

*****
*****
*****
*****
*****
A) 12.Qd5 c6 13.Qd2 Rae8
Black has two pawns for bishop and knight, but Black also has a strong attack that the engines reckon gives full compensation.
B) 12.Na3 e4
The engines continue 13.Nc4 Rab8 14.Qd5 exf3+, slightly favouring Black.
C) 12.Nbd2 e4
This transposes to B) after 13.Nc4.
D) 12.Qa6 Rfe8
The engines at first strongly favour White, but their evaluations sharply change with almost every move in the sequence 13.Na3 Qf5 14.Qc4 Ng4 15.hxg4 Qxg4 16.Kd1 d5!?, when Black is, temporarily, three pieces down for three pawns, but the position is unclear.
There is much to think about here for anyone who plays, or plays against, the Ponziani.