Halloween Gambit

I've never been a lover of gambits, sound or unsound, but the other day I was intrigued by an article I saw on the Halloween Gambit: 1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5

In this gambit (also known as the Müller–Schulze Gambit or Leipzig Gambit) White sacrifices a knight for a single pawn. White's objective is to seize the center with pawns and drive back Black's knights.

The theoretician Oskar Cordel reported in 1888 that Leipzig club players used the opening to dangerous effect, but he did not believe it was sound. Their name for it, Gambit Müller und Schulze, the German equivalent of Smith and Jones. It was renamed the Halloween Gambit in the magazine Randspringer in 1993. The name is founded on the fact that players who are for the first time confronted with it they become shocked as if they were suddenly confronted with the horror of some scary Halloween mask.

Clearly the gambit is unsound, but I let Houdini 1.5a examine the position after 5.d4 for about a half hour and was kind of surprised at the results.

1.e4 e5 2.Nf3 Nc6 3.Nc3 Nf6 4.Nxe5 Nxe5
NowWhite usually plays
5.d4
5.f4 does nothing for his development. Black can retreat the attacked knight to either g6 or c6. For example, 5...Nc6 6.e5 Ng8 7.d4 d5 which favors Black by 2P's according to Houdini.  Here is Houdini's analysis of the variations after about half an hour.
5...Ng6
5...Nc6 6.d5 Ne5 7.f4 Ng6 8.e5 Ng8 9.d6 cxd6 10.exd6 Qf6 11.Nb5 Nxf4 12.Nc7+ Kd8 13.Qf3 Bxd6 14.Nxa8 b6 15.c3 Qe6+ 16.Kd1 Ng6 with Black being given the advantage of only slightly over one pawn.
5...Bd6 only results in equality after 6.Nb5 Neg4 7.e5 Be7 8.exf6 Nxf6 9.Bf4 d6 10.Be2 0–0 11.Nc3 c6 12.0–0 Be6 13.Re1
6.e5 Ng8 7.Bc4 d5 8.Bxd5 N8e7 9.Be4 c5 10.d5 Nxe5 11.0–0 f5 12.Bd3 N7g6 13.Re1 Kf7 14.h3 Bd6 15.Nb5 Bb8 16.Bf1 a6 17.Nc3 Re8 with a nearly 2P advantage for Black.
Handing the opponent a two pawn advantage right out of the opening isn't something I'd be comfortable doing and GM Larry Kaufmann wrote in 2004 that the gambit is refuted by 4...Nxe5 5.d4 Nc6 6.d5 Bb4! 7.dxc6 Nxe4 8.Qd4 Qe7 which he attributes to the Polish IM Jan Pinski. Of course most of us aren't playing Hourdini or a GM and if the large number of short wins for White means anything, this gambit looks like something blitz players or club players might want to give a try...there's nothing to lose but the game.

For more analysis check out Chessville's coverage of it: PART 1 PART 2