Martin Burrows v Andrew Greet
Leicestershire v Cornwall, 2016/17 (Bd.1)
The last time I played Martin, he took a draw off me with the black pieces, so I knew this would be a tough game.
There was no real chance to prepare before the game, so I opted for a fairly solid line without too much sharp theory.
This is one of several respectable lines at Black's disposal. The idea is obvious: Black commences queenside counterplay as quickly as possible.
Funnily enough, I have played the White side of this variation a few times, facing 9...Qb6 ; and even 9...g5!? . Clearly it's a rich position with many possibilities for both sides; but I happen to like the move I played in the game. The idea is to prepare ...b4 without allowing the reply axb4.
Other moves are possible, but this is considered one of White's principal tries for an advantage. With the rook now defended, White is ready to meet ...b4 with axb4.
Is Black losing a tempo? Not really. Rather, I'm angling for a slightly improved version of the 9...Qb6 line, as White's rook is slightly worse on a2 than a1, as it can't reach the centre so easily.
A good move, reinforcing the crucial d4-square.
I didn't have a chance to prepare before the game and I was nearing the limits of my exact theoretical knowledge. I did, however, remember that this quirky queen retreat had been covered in "Playing the French" by Aagaard and Ntirlis - a book which I edited! It looks weird to move the queen for the third time in a row, but Black is claiming that the free moves Ra2 and Ne2 don't really help White.
11...b4 12.axb4 Nxb4 13.Ra1 Nc6 is another line covered in the aforementioned book. I considered this as well but, since I couldn't remember any exact theory, I thought it would be safer to keep the position more closed.
This new move is not especially threatening, although White's position remains okay. After the game I consulted "Playing the French" and found an impressive illustrative line: 12.g3 cxd4 13.Nexd4 Nc5 A similar position has been tested after 9...Qb6 10.Ne2 Qc7 etc., where White has good chances for an advantage. The difference here is that the rook is on a2 instead of a1, and this small detail is enough to give Black equal chances! The key line goes as follows: 14.Nxc6 Ne4 15.Qe2 Qxc6 16.Nd4 Qc7 17.Bg2 Bb7 18.0-0 Bc5 19.c3 0-0 20.Nb3 Bxe3+ 21.Qxe3 Qc4! Finally the difference becomes clear, as White must move his knight to a passive position and then spend time rerouting his rook via a1. 22.Nc1 Rac8 23.Ra1 (23.Kh1? Nxc3! 24.bxc3 Qxf1+ 25.Bxf1 d4+-+ is a nice variation) 23...Qc5 24.Re1 Qxe3+ 25.Rxe3 Nc5 With equal chances. Obviously I had no recollection of this during the game, other than the general point of realising that the rook was worse on a2 than a1.; 12.c3 is another natural move which has been tested in a few games. I was intending to meet this with 12...Na5 , aiming at the light squares on the queenside.
This is the right recapture, sending the knight towards e4. It's also the move that reaps some benefit out of retreating the queen to c7. If the queen was still on b6, White would win a piece with b2-b4.(12...Bxc5?! 13.Bxc5 (13.Ned4!?) 13...Nxc5 14.Ned4 gives White easy play, as 14...Ne4 can be conveniently met by 15.Qe3 followed by Bd3.
We've reached a typical French structure where the active knight on e4 and the odd placement of White's queen and rook mean that Black shouldn't be worse. It's still an interesting battle though, full of complexity and challenges for both sides.
14...Bb7 is a playable alternative, which might well come to the same thing after an exchange on c6. On b7 the bishop provides additional support to the knight on e4, but I decided to put it on d7 in order to restrain f4-f5 and support the b5-pawn, in case ...a5 becomes a good option.; The computer points out 14...g5!? as an interesting possibility; I didn't consider this at all during the game.
I thought about 18...Rac8 but was put off by 19.c3 , when I'd prefer my rook on b8 to prepare ...a5 and ...b4. My chosen move leaves the kingside slightly short of defenders, but I decided I could afford to take this small risk as his queen and a2-rook are not well placed to support an attack.; 18...Bd7 followed by ...f6 is another possible plan.
White takes the opportunity to put some pressure on my kingside.
This seems too passive. It's important to realize that 25.Qg4 doesn't really threaten Nxe6, because the reply ...Qxe5 will defend against mate while leaving White's knight in a deadly pin. 25...bxc3 26.Bxe4 dxe4 (26...Qxe5!? is also good) 27.bxc3 Qxc3 28.Qxe4 The position is close to equal but I'd take Black with the two bishops.; 25.Qh5!? Rf8 26.c4!? is a better alternative, when the position remains balanced.
I decided to keep some tension in the position. My move challenges for the f-file while making ...Qxe5 into a real threat, as White will not have the reply Bf4 available. 25...bxc3 was simple and good: 26.Bxe4 dxe4 27.Rxc3 Qb7=/+ Black is slightly better with the two bishops and targets on b2 and e5.
Exploiting White's poor coordination.
I spent some time looking at 29...Qxc3? 30.Rb3 Qa5 and thought I was doing well, but eventually decided I liked the text move slightly more.; For some reason I didn't spend much time looking at 29...Nxc3! because 30.Qc2 looked annoying. However, I should really have seen that 30.Bxh7+ Kh8! should also win for Black (but not 30...Kxh7 31.Qc2+ Kg8 32.Rb3 ) 30...Ba4! 31.Bxh7+ Kh8 would be winning for Black, since 32.Qg6 can be met by 32...Bc5! 33.Bxc5 Qxc5+ 34.Kh1 Nxb1 and White has nothing on the kingside, as the queen can join the defence via c1 and h6.
Martin defends well and prepares to take on e4.
A good practical decision. I was running short of time, and this move leads to an endgame which should be drawn, but where I can keep playing for a win without taking the slightest risk. 33...Nc3 is the computer's preference, but it loses some coordination and leaves the kingside slightly vulnerable.
With the time control reached, I could relax and enjoy testing my opponent's defensive skills. Clearly the endgame should be a draw, but Black has a few things in his favour: a) The bishop is more powerful than the knight. Moreover, it has a perfect, stable outpost on d5, whereas it is hard to see a decent square for the knight other than f3, where it serves a purely defensive function. b) White's pawns are more vulnerable. The e5-pawn is a clear target, and even the g2-pawn could become weak,especially if Black can get a rook to the second rank. c) Black has the practical advantage of being able to simplify by taking on f3 at a moment of his choosing, whereas White can only sit and wait. And as we will soon see, king-and-pawn endgames are likely to be winning for Black.
37...Bxf3?! would be a mistake, as it would force White to choose the correct defensive structure: 38.gxf3 (it's easy to see that the pawn ending after 38.Rxf3?? Rxf3 39.gxf3 Kf7 would be winning for Black) 38...Kh7=/+ Black can march the king to f5 with good chances to win the e5-pawn, but the ensuing rook ending with three pawns versus two should be a fairly easy draw.
This is a sensible defensive move; White ensures that any future attempt by me to advance my kingside pawns will involve a pawn exchange, which will bring White closer to a draw.
Introducing the possibility of ...Bxf3 followed by gobbling up the e5-pawn.
43.gxf3 Rxe5 should be a fairly easy draw, as Black will have a hard time creating a passed pawn without allowing further pawn exchanges. I suspect that Martin recaptured with his king in order to avoid creating an additional pawn island, but it's not too difficult to see that Black will not be able to gang up on the h4- or f3-pawns without losing a pawn or two of his own.
My guess is that the endgame is still drawn with perfect defence. However, Black has many more winning chances with a passed e-pawn.
It's important to prevent his pawn from getting to h5, as this would fix my g7-pawn as a target. A continuation such as 44...Ra5 45.h5+ Kf6 46.Rb7 g6 47.hxg6 Kxg6 would be much worse for me than the game, as it's easier for White to defend a pawn on g4 than on h4, while virtually any endgame involving a lone h-pawn will be a draw.
Martin opts to eliminate another pair of pawns. After 45.g5 Ra5 followed by ...Ra4 White risks having his king tied to the defence of the h4-pawn. Obviously Black's king has to mind the g7-pawn, so maybe White could hold this, but it's not pleasant.
Even though pawn exchanges bring the defender closer to a draw, I was happy to trade off my h-pawn. The point is that certain rook endings and king/pawn endings would be easily drawn with a lone h-pawn, but easily winning with a lone g- or e-pawn. Moreover, this structure opens up the possibility of attacking the h-pawn and cutting off the white king, as occurs at move 60 of the game.
Black has to advance his king to make progress but the g7-pawn is vulnerable, so I prepare to advance it to g6.
When analysing this game, I did a quick Google search for '7 piece tablebases' and was amazed to discover that the Lomonosov tablebases are available as a free Android 'app'. So, barely five minutes after my online search, I've downloaded the app to my phone and entered the position, and can confirm that the position is still drawn.
By now White had a little over six minutes to complete the game, while I had a little over five.
Obviously Martin avoids allowing his king to become cut off without a good reason.
Trying to make a run for the weak h-pawn, but he attacks the e-pawn to force my king back.
I stopped notating here but have added some moves; they may not be correct in every detail, but they are more or less what happened in the game. If any errors do exist, they will be cases where a king or rook could choose between different squares without changing the evaluation of the position.
White is still drawing after this, but I was happy to be able to advance my king without allowing a check on f7. [White should have 'passed' with a move like 58.Rc7
The Lomonosov tablebase confirms my suspicion that this was the decisive error. Martin wants to keep my king tied to the defence of g6, but he has overlooked a trick. White should have preferred 59.Ra8 ; or 59.Ra5 , or something similar.
That's the point: with White's king cut off on the h-file, I can afford to swap down to a lone extra e-pawn.
After the game, someone asked me if 61.Rf6+ Kxf6 62.Kxg4 was possible; the answer is that Black wins with 62...gxh4 63.Kxh4 Kf5! 64.Kg3 Ke4 65.Kf2 Kd3 and the active king decides the issue.; The path of most resistance would have been 61.h5 Rh4+ 62.Kg3 Rxh5 when Black is winning, but still has to take some care.61...Rxg5
My memory of the remaining moves is too hazy, as White has umpteen possible rook moves at every turn. I won easily by advancing my pawn to e2 and king to e1 while keeping his king cut off.
For any readers who aren't completely clear about how to force the pawn through to the promotion square while escaping rook checks, I highly recommend looking up Lucena's position in an endgame book.
Every player can benefit from learning certain basic theoretical endgames. Not only does this help you get the desired result when you reach those endgames; it's also tremendously useful to know in advance whether certain positions will be winning or drawing, as this knowledge will enable you to decide much more quickly and accurately whether to exchange pieces as I did at the end of this game.
Notes by Andrew Greet