HERE is a way to spice up the Caro-Kann by effectively turning it into a Blackmar-Diemer Gambit:
1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.f3!?
Spanton (2021)- Alan Ormsby (1995)
Isle Of Man 2003
4...e3?!
Accepting the gambit, which I will come to next, is overwhelmingly Black's most-popular choice in ChessBase's 2020 Mega database.
The text is Black's second-most common move, but if declining the gambit is what Black wants to do, it is almost certainly better to play 4...e5. This is sometimes cited as a refutation of White's fourth move, and is one of the reasons many whites prefer 4.Bc4!?, planning f3 on the next move.
After 4...e5, the main line runs 5.dxe5 (Milner-Barry preferred 5.Be3!?) Qxd1+ 6.Nxd1 exf3 7.Nxf3, with a position that may be better than is commonly supposed. Black's most-popular continuation in Mega20 is 7...Bc5, whereupon the analysis engines Stockfish10 and Komodo10 continue 8.Bc4 Ne7 9.Be3 Bb6 10.Ng5 0-0 11.0-0 with what they reckon is a slight edge for White, presumably based on White's lead in development.
5.Bxe3 Bf5
Marginally more popular in Mega20 is 5...Nf6, but then 6.Bd3 leaves Black struggling to find a good square for his light-square bishop. That is by no means the end of the world, but the engines give White a slight edge.
6.Nge2
This was the choice of German international master René Stern - now a GM - when he had this position in 2001, but the engines reckon even stronger is 6.g4!? Bg6 7.Nge2.
6...h6 7.Ng3 Bh7 8.Bd3 Bxd3 9.Qxd3
White is better thanks to having four pieces developed to Black's none, and Black is yet to open a diagonal for his king's bishop
(1-0, 93 moves).
**********
Accepting the gambit is much more popular.
Nicolas Bertrand (1923) - Laurent Picart (2170)
Fouesnant (France) 2009
4...exf3 5.Nxf3 Nf6
This is Black's most-flexible response, leaving open the question of where to develop the light-square bishop. It is also a direct transposition to the Ziegler Defence in the Blackmar-Diemer Gambit.
Most popular in Mega20 is 5...Bg4, when the engines want White to sac a second pawn with 6.h3 Bxf3 7.Qxf3!? Qxd4 8.Be3.
Then the normal continuation of 8...Qf6 9.Qg3 gives White at least full compensation (bishop-pair, lead in development, exposed black queen) for the pawns, according to the engines. So they suggest 8...Qh4+ 9.g3 Qf6, when White does not have the sidestep Qg3. The only game to reach this position in Mega20, Helmut Küspert (1948) - Hubertus Schulze (2096), Landesliga North 2006, saw 10.Qg4 Nd7 11.0-0-0 e6, when Black is surely better, and did go on to win, but the position is tricky to play for Black.
Although Komodo10 likes 6.h3, Stockfish10 comes to reckon calmer play with 6.Be3 or 6.Bd3 is better. After the latter, Black can still win a second pawn by 6...Bxf3?! 7.Qxf3 Qxd4, but Bd3 is a much-more useful move for White than h3, and the engines reckon White gets an advantage with 8.Bf4.
6.Bc4
Gary Lane's recommendation in his 1995 Batsford book Blackmar-Diemer Gambit.
The engines prefer 6.Bd3 (the game reached the position with 6.Bc4 in a slightly different move-order, so 6.Bd3 was not an option), but since Black has not moved his e pawn, 6...g6 seems to be a good answer, as does the more-popular 6...Bg4, although in both cases White has some compensation for the pawn in the shape of a lead in development and useful half-open files.
6...e6
Black's most-popular reply in Mega20, but there is a lot to be said for first developing Black's light-square bishop. This is best done by 6...Bf5, which warrants consideration in a separate game.
Note, however, that 6...Bg4?, as played in 44 games in Mega20, is embarrassing if White finds 7.Bxf7+! etc.
7.0-0 Be7 8.Qe1
White is not being subtle about which part of the board he will attack on, but Black is more-or-less committed to castling short as his queenside is completely undeveloped.
8...0-0 9.Bg5
The engines suggest an interesting apparent-novelty, 9.Ne4!?, eg 9...Nbd7 10.Bd3 Nxe4?! 11.Qxe4 Nf6 12.Qh4 with advantage. Probably better in this line is Stockfish10's 10...c5 11.Nxf6+ Nxf6 12.Qh4, but with strong kingside pressure for White.
9...Nbd7 10.Qh4!?
Stockfish10 prefers 10.Bd3!?, although Lane calls that "a Teichmann Defence [1.d4 d5. 2.e4 dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 Bg4] where White has lost time by moving the king's bishop twice."
10...Re8
This comes to be Stockfish10's choice, but Komodo10 prefers 10...h6, although then Stockfish10 gives 11.Bd3 Re8 12.Bxh6! gxh6 13.Ne5 Bf8 14.Rf3 Bg7 15.Rg3, when Black seems forced to give back his extra piece by 15...Ne4. After 16.Qxe4, Black keeps in the game with 16...f5, but White is better.
 |
| How should White proceed? |
*****
*****
*****
*****
*****
11.Bd3?!
Probably better is the typical BDG move, 11.Ne5. White may have feared the exchanges that seem forced after 11...Nxe5 12.dxe5 Nd7, but Stockfish10 gives 13.Bd3 Nf8 14.Bf6!, which it reckons is winning, eg 14...gxf6 15.exf6 Ng6 16.Bxg6 hxg6 17.Ne4!. Komodo10 suggests 14...Bd7, and after 15.Ne4 it gives 15...b5, but 16.Qh5! wins, eg 16...Bxf6 17.Nxf6+! gxf6 18.exf6 etc. A fantastic line, but very typical of the Blackmar-Diemer Gambit.
11...Nf8
The engines suggest 11...h6, but then Stockfish10 finds 12.Bxh6! gxh6 13.Ne5 with similar play to the note at Black's 10th move.
12.Ne5 Ng6
This is best, according to the engines, but after …
13.Bxg6 hxg6?
… Black is lost. He had to play 13...fxg6, although White has a big initiative for his pawn after both 14.Rf3 and 14.Rad1.
14.Rf3 Nh7
There is no satisfactory defence.
15.Rh3 Bxg5 16.Qxh7+ Kf8 17.Qh8+ Ke7 18.Qxg7 1-0
After 18...Qxd4+ 19.Kh1 Rf8 comes 20.Nxg6+.
**********
Although 6...e6 is the most-popular move in Mega20, Lane treats
6...Bf5 as the main line, and that move is preferred by the engines.
Spanton (2004) - Richard Almond (2158)
Isle Of Man 2004
7.0-0 e6 8.Ne5
This threatens 9.Nxf7!
 |
| How should Black proceed? |
*****
*****
*****
*****
*****
8...Bg6
The tempting 8...Bxc2?! is probably less good, as was shown by future IM Gerard Welling in a 1981 game, where he found 9.Nxf7! After 9...Kxf7 Welling played 10.Qg4?!, which was thought strong, but that verdict seems to have been reversed by the game Ian Webster (2116) - Michael Compston (1835), 4NCL (Telford) 2005, where Black's 10...Qd6! was instantly winning, according to the engines. The game continued 11.Bg5 Nbd7 12.Rae1 Re8, when it is hard to see what, if anything, White has for his sacrificed knight and pawn (0-1, 48 moves). However, Welling's concept still works, but with the follow-up 10.Qxc2!, which was played in Steven G Burke (2055) - Ian G Gallagher (2005), 4NCL (Sunningdale) 2007. The game continued 10...Qxd4+, upon which White unleashed 11.Be3!! The game saw 11...Qxc4 12.Rf4!, trapping the queen (if 12...Qa6, then 13.Ra4). Materially, Black is not badly off, but his lack of development and weak e pawn give White a large advantage, according to the engines (but 0-1, 57 moves).
Going back to the position after 9.Nxf7!, Black can try 9...Bxd1 10.Nxd8 Kxd8, but after 11.Rxd1, White's bishop-pair and Black's weak e pawn give White good compensation for his pawn-minus.
9.Be3
If you give the engines enough time, Stockfish10 settles on a novelty, or at least a move that is not in Mega20, viz 9.Nxg6!? The move also comes to be Komodo10's second choice - just 0.02 of a pawn behind 9.Bf4.
The idea, at least as I understand it, is that White will not go for a quick tactical knockout, as is often White's aim in the BDG. Instead White will rely on the bishop-pair and better pawn-structure to create long-term compensation. I am sympathetic to this, as will become clear, but there is no need to commit to it this early.
Instead I have gone for the developing 9.Be3, which for a long time is Komodo10's top choice, and is second in popularity in Mega20 to 9.Bg5.
9...Nbd7
As Black can capture on e5, now is the time to implement Stockfish10's idea:
10.Nxg6 hxg6
After the further moves ...
11.Qf3 Be7
… we have reached a position I wish to elaborate on. The game Spanton - Almond got here via a slightly different move-order, but I have used the move-order that arises from the Caro-Kann.
Lane says of the Ziegler Defence: "This line is a popular and important method of defence. It involves a pawn-formation similar to the Caro-Kann and has a sound reputation."
I would add that it is particularly effective when Black avoids the popular 6...e6 and instead develops his light-square bishop outside the pawn-chain. This is reflected in White scoring a healthy 54% after 6...e6, but just 33% after 6...Bf5.
The stats and the engines' assessments add up to the fact that Black has a good position that cannot realistically be blown away.
That is why I believe, if Black gets as far as the position after 9...Nbd7, and it requires some precise choices along the way, White should take the bishop-pair and settle in for the long-term.
I think this game is a reasonable example of how that can pan out in actual play.
12.Ne4
I see from my original notes that I realised, pretty soon after the game, that completing development with 12.Rae1 or 12.Rad1 is more logical, and the engines agree, although I also see Crafty19.01 preferred the text.
12...Nb6
It was too much to hope for 12...Nxe4?? 13.Qxf7# (but I did hope for it!).
13.Bb3
White loses the bishop-pair after 13.Bd3?! Nbd5 14.Bf2 Nb4.
13...Nbd5 14.Bf2
Komodo10, which I believe has a more positional feel than Stockfish10, which is usually better at finding deep combinations, likes the text. Stockfish10 prefers 14.Ng5?! Nxe3 15.Qxe3, when maybe the opposite-coloured bishops would give hope of drawing chances, but it would be a bleak position to play as White.
14...Nxe4!?
This is the engines' choice, but exchanging a pair of knights makes it less likely White will forfeit the bishop-pair.
15.Qxe4 Bd6 16.g3!?
I felt 16.h3 left White very weak on kingside dark squares, as well as setting up a possible sac on h3. The text also denies Black's pieces use of the f4 square.
16...Qg5 17.h4!?
I gave this an exclamation mark when I first wrote up the game, but today's engines are not overly impressed, preferring 17.Qf3 or 17.Rae1.
17...Qe7
Black could seek an exchange of queens with 17...Qf5, but such an exchange normally enhances the power of the bishops because they become the only pieces with long-range diagonal movement. There is also the point that in this specific position, White's king looks the more vulnerable, which is a reason to keep queens on the board.
18.c4 Nf6 19.Qf3 0-0
This is Komodo10's choice, but Stockfish10 prefers 19...0-0-0. From a practical view, long castling offers White more chances of counterplay.
20.g4?
I apparently played this to prevent ...Nh5, but getting on with development with 20.Rae1 is better.
20...Bc7
Preparing a bishop-queen battery on the b8-h2 diagonal.
21.Kg2 Qd6 22.Rh1 b5
The engines prefer a developing move such as 22...Rad8.
23.Rac1 Qf4?!
The engines reckon Black keeps a bigger edge with 23...bxc4 or 23...a5.
24.cxb5 cxb5 25.Rc5 a6 26.Qxf4 Bxf4 27.Kf3 Bd2
Or 27...Bd6 28.Rc6, when again Black's bishop is short of good squares.
28.Rd1
The engines reckon 28.g5 equalises, eg 28...Nd5 29.Bxd5 exd5 30.Rxd5 f6 31.gxf6 Rxf6+ 32.Ke2 Raf8 33.Bg3 Bh6, when White's better pawn-structure presumably cancels out Black's piece-activity and better coordination.
28...Bb4 29.Rc6 Rfd8
If 29...Rfc8, then 30.Rdc1, when 30...Rxc6 31.Rxc6 leaves Black with a horribly passive queen's rook. Therefore 30...Rd8 is better, in which case it is better to play it at move 29, as in the game.
30.Rdc1 Nd5 31.a3 Bf8 32.Ke4
White can equalise by giving up the bishop-pair, according to the engines, ie 32.Bxd5!? Rxd5 33.Rc8.
32...Ne7 33.R6c2 Rac8?!
Black contests the open file and is happy if the rooks come off as that gets him into an ending a pawn up, but it would also enhance the power of the bishops as, with rooks off, White will be 2-1 up on long-range pieces.
The engines reckon Black keeps an edge with 33...a5.
34.Bg3
Also good is 34.Rxc8.
34...Rxc2
This is fine, but after …
35.Rxc2
… Black should probably play 35...Nd5, when 36.Bxd5 Rxd5 37.Rc6 gives White full compensation for his missing pawn, according to the engines.
Instead the game saw …
35...Rc8?! 36.Rxc8 Nxc8 37.d5
… when more lines open for the bishops.
37...f5+?!
Better, according to the engines, is 37...exd5+ 38.Kxd5 Ne7!, when the white king cannot safely advance. My old Crafty19.01 thought the game would be equal, while whatever version of Fritz I had at the time gave Black a slight edge. But my modern engines prefer White, one line running 39.Ke4 Nc6 40.g5 Ne7 41.Bf4 Nf5 42.h5 Ne7 43.hxg6 Nxg6 44.Bd2, when Black's queenside looks vulnerable, but the position is unclear.
38.Kf3
After 38.gxf5? exf5+ 39.Kd4, Black has a protected passed pawn and activates his king with 39...Kf7, which the engines reckon makes Black slightly better.
38...Kf7?!
Black faced a tricky decision, but this is probably not the right answer.
The engines suggest 38...Be7 39.dxe6 Kf8, but then White has restored material equality while keeping the bishop-pair.
If 38...fxg4+ 39.Kxg4 exd5, the black king is sidelined by 40.Bxd5+ Kh7. But if instead 39...Kf7, then 40.dxe6+ again restores material equality while keeping the bishops.
39.dxe6+ Ke7
I thought 39...Kf6!? might be better as, although 40.g5+ drives the black king to e7 anyway, Black gets a protected passed pawn. But the engines prefer 40.Bc1, eg 40...g5 41.Bc3+ Kg6 42.h5+ with a large advantage. If Black meets 40.Bc1 with 40...Nd6, then the engines like 41.g5+ Ke7 42.Bd5 with massive coordination problems for Black.
40.Be5 Nb6
The engines suggest 40...Nd6 41.gxf5 Nxf5, but then comes 42.Kg4 with Kg5 to follow.
41.Bd4 Nc8
If 41...Nc4, then 42.Bxc4 bxc4 43.Bc5+ brings a winning pawn-ending.
42.gxf5 gxf5 43.Kf4 Kd6
Or 43...g6 44.Kg5.
44.Kxf5 Ne7+ 45.Ke4 Ng6
Enticing the h pawn forward does not help, but Black is lost anyway.
46.h5 Ne7 47.Bc3 Kc6
Or 47...Nc6 48.Bd5 Ne7 and 49.Bb4+ or 49.Be5+.
48.Ke5 Kc7 49.Bd5 Nc8 50.Ba5+ Kb8 51.Bb4 Ne7
Perhaps 51...Bxb4!? 52.axb4 Kc7 was worth a try, but then the quickest win, according to the engines, comes from 52.e7! Nxe7 53.Ke6 as 53...Nxd5 54.Kxd5 Kd7 loses to 55.Kc5 (but not 55.b3? Ke7) Kc7 56.b3 etc.
52.Bxe7!
Bringing about opposite-coloured bishops, but this is the clearest way to win.
52...Bxe7 53.Kf5 Kc7
Or 53...Bf6 54.h6 Bxb2 55.e7.
54.Kg6 Kd6 55.Ba2 a5 56.Kxg7 Bg5 57.Kf7 b4 58.axb4 axb4 59.h6 Bxh6 60.e7 1-0
This game illustrates how tricky it can be to play against a bishop-pair, even with a small material advantage (in this case an extra-but-doubled pawn).
**********
CONCLUSIONS
Ending up in a Blackmar-Diemer Gambit will come as an unpleasant surprise for many Caro-Kann players. Even more so if they in their games as Black against 1.d4, they usually reply with something other than 1...d5 (so ruling out a pure BDG).
And if they do happen to meet 1.d4 with 1...d5, and so by necessity have a line ready for the BDG, it may not be the Ziegler Defence with an early ...c6.
All in all, White has a good chance with 1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.f3!? of getting the opponent on unfamiliar ground.
