Hot novelty in Marshall attack

Hello,

I want to bring to your attention a game I played recently at chess.com. It features the well-known Marshall attack and a very new and interesting move made by my opponent. I'm speaking about this line:

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O 8. c3 d5 9. exd5 Nxd5 10. Nxe5 Nxe5 11. Rxe5 c6 12. d4 Bd6 13. Re1 Qh4 14. g3 Qh3 15. Re4 g5 (see the diagram below)

Black is a pawn down, but White is under a powerful attack. In this position, the moves encountered until now where 16. Qf3, 16.Qf1 or 16. Qe2. Basically White is trying to defend.

My opponent found the calm, excellent move 16.a4. The idea is that White will force the exchange of one pair of rooks and thus Black's attacking potential is significantly reduced. Black will not have control over the e file and White's defense seems to be easy. White can force further exchanges (moves 22-24) and Black will soon find himself defending a worse position, with a central pawn down.

You can see the whole game below, I had no idea how to deal with this novelty and I was swept away after 12 moves. Any suggestions are greatly appreciated.

As a coincidence, Toga (powerful chess engine) also says that 16. a4 is the optimal move in this position. It also agrees with all the moves my opponent made starting from move 16.

True, f5 is also a key move in Marshall. The point is that 16. a4 solves some of White's development problems (connecting his rooks). White will either exchange his innactive rook from a1 or will get counterplay by controlling the a file.

What Rybka says is not so important to me, Rybka does think that Marshall gambit is not sound at all. For it, the position is easy to play and evaluated as winning for White (since Black is a pawn down). For humans it looks complicated and not clear at all.

Thanks for your comments!

tonydal: I thought about this: 16...Bf5 17. Nd2 Bxe4 18.Nxe4 and there is no defense for Black's g pawn. It looks to me that taking the rook leaves white with easy play, no problem and maybe advantage (or this is what I felt in the game). In the normal Marshall (without the novelty 16.a4), White usually does not defend the rook with Nd2, since this blocks white's dark-square bishop.

Thanks! No gonnosuke, nobody had any idea about how to deal with 16.a4 Indeed, going all in with f5 and not allowing White to exchange a pair of rooks (as I did in the game) seems to be the only chance for Black.

You say that 16.a4 is not a new move? I'd be very surprised, it has never been played in a recorded game (for instance, chessgames.com have no reference for such a move). I am surprised how such a move - you can't say it is not a reasonable try - has never been played before.

Hehhe, what gonnosuke siad... or read those messages I sent you costelus :P

 Agreed.  You played very conservatively, and sometimes it's difficult when playing against very good players to know if there's a trap waiting around the corner.  In this case though, taking the rook was the right thing to do.

First, I explained above why after Nd2, taking the rook is not such a big deal.

Gonnosuke: thanks! Indeed, this continuation was very hard to see. What database of 7 million games do you have? As far as I know Chessabase has about 3 millions.

   I've played the marshall as black several times but never got the 15 Re4 response, so this is very instructive for me. Thanks.

Sounds like you might have been playing Toga?

Chessbase contains not only grandmaster games, there are many games even from players rated 1900-2000. As for this move, 16. a4, there is no reference also on chessgames.com. To me it looked like a reasonable move, Black must play very energically to secure an advantage. In the actual game, I could not find the winning line. I thought of f5, as this is also a key move in this line of Marshall, but I thought of playing f5 after Bf5 and capturing the rook.

I tried to use chessbase lite, but I find it very painful that, each time I want to look for a position, I have to set up the board myself. I cannot just simply paste a FEN string.

Of course chessbase reader allows you to paste the fen, but it does not allow you to paste the fen and then search the position in a database. Unless you set up the position manually (piece by piece).