Tuesday, 12 December 2017

029. Avoiding the Four Knights

White: tripoduk - Chess.com, 2017

In my previous post I mentioned that I've only ever had six OTB games with the Four Knights, and “it hardly seems worthwhile studying an opening that you only face once every five years”. The point is that there's actually quite a lot of theory on the Four Knights, especially if you meet 4 Bb5 with Rubinstein's 4...Nd4. If you can't be bothered to try and remember it all, the question is what to do other than 3...Nf6.

Obviously, 3...f5 is the move I'd most like to play. Unfortunately, 4 d4! fxe4 5 Nxe5 Nf6 6 Bc4! d5 7 Nxd5! Nxd5 8 Qh5+, as in G.Breyer-Z.Von Balla, Bad Pistyan 1912, pretty much wins for White. And there aren't any significant improvements for Black which make 3...f5 acceptable.

Alternatively, 3...g6 is a popular avoidance strategy; but as IM Andrei Obodchuk says in his book on the Four Knights (New in Chess 2011), this “is employed most often by stronger players against weaker opponents”, with the aim of outplaying them in less familiar positions. Whereas my 3 Nc3 opponents tend to be of comparable strength to me – or stronger. Outplaying them from a slightly inferior position is therefore not the most likely outcome.

So, I've recently been looking at 3...Bc5!? instead. The book refutation of this move runs 4 Nxe5! Nxe5 5 d4 Bd6 6 dxe5 Bxe5 and so on.

Back in 1993, Nunn wrote that “Examination of published theory gives the impression that this line is very bad for Black, which is probably true”. Or not. GM Oleg Romanishin has been playing 3...Bc5 for years with decent results (P24, W8, D10, L6 in the database); and as Obodchuk shows in his more serious investigation of the variation, it really isn't that bad at all. Even his various suggestions for White don't seem to trouble Black much since, if nothing else, they can often be circumvented.

I tested one such in the game below. This followed D.Pedzich-O.Romanishin, Koszalin 1997, where after 12...Nf6 13 0-0 0-0 Obodchuk suggests 14 Re1 as offering White a slight advantage. That may or may not be the case, but Black doesn't have to go in for it anyway. The plan of 12...Bd7, ...Bc6, ...0-0-0, and ...Nf6, as in B.Bogoevski-R.Navarro Segura, Merida 2001, seemed preferable to me.

As it happens, I've since had the chance to test 3...Bc5!? out over the board as well. J.Willow-J.Tait, Notts County Championship 2017, saw 7 Bc4 Nf6 8 0-0 0-0 9 Re1 Re8 10 Bg5 c6 11 f4?? Bxc3 12 bxc3 Qb6+ and Black won a pawn for nothing.

Okay, I didn't in the end manage to win either game, but at least I enjoyed them a whole lot more.


  1. I've been having quite a bit of fun with Rubinstein's 4.Bb5 Nd4 recently, often getting into those gambit lines where White takes on e5, but perhaps it's a different story when you get to a higher level than I (circa 1700-1800).

    As for 3...Bc5, I note that the colours-reversed "fork trick" lines like 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Nc3 Nxe4 5.Nxe4 d5 are often overestimated for Black, as if Black can actually get the advantage. 6.Bd3 dxe4 7.Bxe4 is probably equal with best play. Thus while I imagine White should be able to get an edge after 3...Bc5 because of the extra tempo, it shouldn't be too dramatic.

  2. Hi again Ian :)

    Yes, I've generally played 4...Nd4 before (against 4 Bb5). But since I have no particular ideas in that line, it means playing on my opponent's turf, which I never much like to do, even if that turf is theoretically equal turf.