November signals the start of the 4NCL season and I was in action on Board 1 for Guildford on the 19th & 20th November. I emerged with 2 wins from tough and fraught fights. The Saturday game featured an extremely sharp rook and knight ending which I analysed at some length with my friendly young opponent afterwards. I’ve since spent some train time analysing the ending further and it seems that although I’d seen plenty of good stuff during the game, my opponent’s judgement of the position was closer to the mark than my customary optimism!
If you’d rather play through the game online than with a set, try my Chessbase Cloud site: http://cloudserver.chessbase.com/MTIyMTYx/replay.html. I’ve split up this article into 2 games for readability: the first deals with the actual game continuation; the second focusses on the improvement 39.f5.
Sadler,Matthew D – Sanders,Isaac B
After a balanced opening, the position – somewhat to my surprise – turned my way due to the vulnerability of Black’s queenside (in particular the b6 pawn). I was attracted to the additional possibility of targeting the soft spot in Black’s kingside structure (f5) with e3–e4 which seemed very strong.
Supporting the e4 push
I was aware that this was possible, but I dismissed it with a vague thought of “exchange the queens I’ll just get the f3 pawn in the end”. However, once my opponent started thinking, I began to get uneasy. I noticed the knight hop into c2 and the new entry point on g4 and I suddenly wondered how I was ever going to round up f3! My opponent played this move after 6 minutes’ thought (leaving himself with just 4 minutes plus increments to move 40) I took 17 precious minutes on my next move leaving myself with 9 minutes.
31.Qd2 This felt crazy but I was attracted for a while to the threat of Bxf6 followed by Nfe4 trapping the queen 31…Rd7 32.Bxf6 Bxf6 33.Nfe4 fxe4 34.Rxf3 exf3 was what put me off as White can’t hold the knight on d6, though Komodo finds some equalising counterplay with 35.h3 gxh3 36.Qh2
I got a number of dubious remarks about this move after the game which was a shame as I was really pleased with it. I’d seen clearly that after the following exchanges, the remaining White rook and knight would combine powerfully to keep the Black rook away from the d-file while eating up pawns on the 6th rank. I’d seen a lot, but as you will see, the position was even more complicated than I’d realised.
32…Rxd1 33.Rxd1 Ng4
Much the better way of doing this.
33…Ne4 34.Nxe4 fxe4 35.Bxg7 Kxg7 36.Nd6 picks up the e-pawn as 36…Rd8 37.Nf5+ wins
34.Bxg7 Kxg7 35.Rd6 Nxe3 felt a little wrong to me: the White knight on f2 is passive while the Black knight on e3 is active. It felt more reasonable to exchange off the knights even though it gives Black a protected passed pawn on f3. Stockfish gives the following lines: 36.Rxb6 Rb8 37.Rxe6 Rxb7 38.Rxe3 Nc2 39.Rxf3 Nd4 40.Re3 Rb4 41.Nd3 Rxb3 42.Kf2 Rc3 43.Nxc5 Rc2+ 44.Kf1 Rxc4 with an edge for White although I would expect Black to be sufficiently active to reach a draw.
34…fxg4 35.Bxg7 Kxg7 36.Rd6
I evaluated this position fairly well from afar. I was particularly happy when I grasped how well-placed the knight was on the unusual square b7. First of all, it takes away the d8 square from the Black rook, preventing deadly counterplay along the d-file. Secondly, that means that White can allow his own rook to leave the d-file and start grabbing loose pawns on e6 and b6. Thirdly, once White does grab the pawn on b6, the knight on b7 attacks 2 further loose pawns on a5 and c5. As an added bonus, from c5 the White knight still restricts …Rd8 which can be met by the fork Ne6+. Finally, from b7, the knight eyes 2 excellent squares: d8 and d6. I had a lovely feeling that lots of tactics were working for me – which they were – but I maybe got a little carried away by it all. After all, tactics can be easily broken…
An obvious and good idea, liquidating one loose pawn and (half)-opening the e-file thus creating a further potential entry channel for the Black rook.
I was a little nervous about this as …Rb8, pinning the knight to the rook, was bound to happen in reply. However, I knew I had at least Rb5 unpinning and threatening to win either the a5 or c5 pawns so I felt the danger of the pin was minimal. I also wasn’t sure what Black could do while I was wasting time grabbing pawns as my knight on b7 was still covering d8! Note also that …Re8 is well met by Nd6 as …Rxe3 loses to Nf5+! More tactics!
37…exf4 38.gxf4 Rb8
The critical position in this ending. I spent 7 of my remaining 10 minutes on this move. In the post-mortem after the game, my opponent was upbeat about his chances here which surprised me greatly. I had the feeling that I was pushing Black around. It was only when I accidentally looked at the position from the Black side on my pocket chess set that I understood what he meant! The position looks scary for White! Obviously, White wants to unpin his rook as quickly as possible and I saw 2 ways of doing this:
- 39.Rb5: my “banker” and the move I had foreseen several moves earlier
- The sharper 39.f5, bringing the f-pawn in to the fray and preparing to unpin with Rg6+.During the game, I was very tempted by 39.f5. I considered the following lines:
a) 39…Re8 40.Rg6+ Kf8 41.Rf6+ Kg7 42.Re6
While considering this line, I got the first mental glimpse of Black’s counterplay in this ending:
42…Rxe6 43.fxe6 Nd3 44.Nxa5 h5 45.Nc6 h4 46.a5 g3 47.hxg3 h3
A lovely idea! White is not in time to deal with …f2+ followed by …h2.
The main reason I rejected this line. I wasn’t sure how to continue; 40.Rb5 Re8 looked fine for Black as due to the inclusion of f5 and …Kh7, White no longer has his Nd6–f5 trick. I decided therefore to retain all my tactical possibilities and leave the king on g7.
In the game therefore, I went for 39.Rb5
This was my main line too, but in the post-mortem, we (well my opponent, actually) were much cleverer:
This is what I meant by tactics can be broken! White’s rook and knight are performing wonders in combining pawn grabbing with restricting the Black rook’s potential activity on the d and e-files, but it’s all based on one tactical detail: the position of the king on g7. 39…Re8 fails to 40.Nd6 Rxe3 41.Nf5+. Quite wonderful that a single knight on b7 can prevent a rook from activating itself on 2 central open files! However, take the king on g7 out of the equation and the knight is suddenly powerless. Of course it feels very weird to put the king on to the back rank, but it’s all about tactics. If Black can get his rook onto White’s back ranks, the position of his own king won’t matter in the slightest.
In fact, after this move I think that White is in trouble! My engines take a very long time to realise this which gives you a now rare feeling of seeing more than the silicon monster!
My first reaction at the post-mortem. I’m sure I would have played this in the game. By advancing his e-pawn, White ensures he can play his e-pawn to a protected square after …Re8, while after e5, he can also block the e-file with Nd6.
Lots of other alternatives:
a) 40.Nxa5 Rd8
b) 40.Nxc5 Rd8 41.Ne4
is a better way of grabbing a pawn but White is the one under pressure.
41…Rd1+ 42.Kf2 Nd3+ 43.Kg3 Rg1+ 44.Kh4 f2 45.Nxf2 Nxf2 46.Kh5 (46.Rxa5 g3 47.hxg3 Rh1#
is a gorgeous mate!) 46…Kg7 47.Rxa5 Rg2 48.Ra7+ Kf6
c) 40.Nd6 Rd8 41.Rb6 h5
will lead to other lines.
d) 40.f5 Re8
The inclusion of …Kg8 makes this move possible
is very scary. This line illustrates what I said earlier: don’t trust your engine! The turning points in this endgame are clearly too far away even for Komodo 10. His assessment of the ending starts at +1.18, drops the further you play on to 0.00 (here at move 41). It’s only when you carry on with some natural rook checks that the evaluation drops right down to -2.73! That’s a big swing! In other words, you can only use the engine assessment when you reach the end of a variation. Until that moment, if it feels scary, trust yourself!
42.Rxa5 Re1+ 43.Kf2 Re2+ 44.Kg3 (44.Kg1 Nc2 intending …Ne3 or …Rg2+) 44…Rg2+ 45.Kh4 f2 46.Ra8+ Kf7 47.Ra7+ Ke8 48.Ra8+ Ke7 49.Ra7+ Kd8 wins
Back to my intuitive choice 40.e4
An annoying move, challenging the loosened kingside pawns. Komodo initially very optimistic and then …losing!
41.f5 h5 42.Nd6 Rd8 43.Rb6 h4 Suddenly the full horror dawns on Komodo and he jumps to –1.44 44.f6 g3 45.hxg3 hxg3 46.Nf5 Ne5 47.Nxg3 Rd2 48.Nf5 Ng4 wins for Black
41…Nxf4 42.Nd6 Rd8 43.Rxa5 h5 44.Kf2 Nh3+ 45.Ke1 h4
The Black pawns are too fast
In the game, Black’s 39…Nd3 was much weaker:
40.Nd6 Rd8 41.Rb7+
A key difference compared to 39…Kg8. This check with gain of tempo re-establishes the fantastic coordination of White’s rook and knight. We also see a nice result of Black’s 39…Nd3: the d-file is blocked which means that White can move his knight away from the d-file without allowing immediate counterplay.
Allows a neat finish
a) 41…Kf6 42.Nf7 g3 An amazing Komodo resource I hadn’t seen 43.hxg3 (43.Nxd8 f2+ wins) 43…Rg8 44.Kf1 Rxg3 45.Rd7 Nc1 46.Kf2 wins for White
b) 41…Kg8 42.Nf7
c) 41…Kg6 Black’s toughest defence. I had planned a good move thank goodness: there are enough ways to go wrong!
42.Rb6 Kh7 43.Ne4
43.Nf5 h5 44.Rh6+ Kg8 45.Rxh5 Rb8 is irritating for White. The e- and d-files have been covered, but the b-file is one too many! 46.Nh6+ Kf8 47.Nxg4 Rxb3 48.h4 Rb1+ 49.Kh2 f2 50.Nxf2 Nxf2 51.Rxc5 Rb4 will lead to a draw;
43. Nf7 was an equivalent option I thought during the game but…. 43…Re8 (43…Rd7 44.Nxh6) 44.Rxh6+ Kg7 45.Nd6 So clever I thought but… 45…Rb8 (45…Rxe3 46.Nf5+) 46.Nf5+ Kf7 is better for Black! I can’t hold b3!;
43…Kg7 44.Nf6 Kf7 45.Nd5
45.Nxg4 Rg8 46.h3 h5
threatening Rb7+ looked pretty good for White though Black is not finished yet
42.Nf7 Re8 43.Nxh6 Rxe3 44.Rb8+
The key idea, forcing the exchange of rooks.
44…Kg7 45.Nf5+; 44…Ke7 45.Nf5+
45.Rxe8+ Kxe8 46.Nxg4 Nc1 47.Kf2 Nxb3 48.Nf6+ Ke7 49.Ne4
As we have seen, 39.Rb5 was a fairly serious mistake after 39…Kg8! 39.f5 has the advantage of bringing an additional unit into play to coordinate with White’s rook and knight.
Now White has 3 interesting possibilities. It would have been very tough to make a choice on the 40th move!
Note also that after 39…Re8, we need to correct my analysis during the game: the simple 40.Rg6+ Kf8 41.Nxc5 Rxe3 42.Rxg4 wins according to Komodo
A move that Komodo always wants to play. Usually his evaluation drops sharply after I push Black’s h-pawn a few times! The advantage of this move is that after …Re8, Nd6 defends the e-pawn and thus prevents Black from breaking through
i) 41.e5 h4
very scary for White!
is even the move both engines want to play 42.e6 g3 (42…Nd3 is even stronger 43.e7 f2+ 44.Kf1 g3 45.hxg3 h3 wins!) 43.hxg3 hxg3 44.e7 Nc2 45.Re6 f2+ (45…Nd4 46.Nd8 Ne2+ (46…Rxb3 47.e8Q Rb1+ 48.Re1 f2+ 49.Kg2) 47.Rxe2 fxe2 48.e8Q) 46.Kg2 (46.Kf1 Rg8) 46…Ne1+ 47.Kf1 Nf3
is Komodo’s suggestion. He likes Black though after
43…Kg7 44.Kf2 Rg8 45.f6+ Kg6
Stockfish also thinks that Black is clearly better though a human could easily get confused!
ii) 41.f6 Nd3
Good timing, preventing White from supporting his f-pawn with e4–e5
41…h4 42.f7 Rf8 (42…Kg7 43.e5 g3 44.hxg3 hxg3 45.Rf6) 43.Nd6 Kg6 (43…g3 44.Nf5) 44.e5 g3 45.hxg3 hxg3 46.Ne4+ Kxf7 47.Nxg3 Nd3 is almost equal
42.Rd6 Rxb7 43.Rxd3 Kg6 44.e5 Kf5 45.Re3 Rb8
Both Stockfish and Komodo prefer Black here!
46.Kf1 Rd8 47.Re1 h4 48.e6 Kxf6 49.e7 Re8 50.Re4 Kf5 51.Re3 h3 52.Kf2 Kf6 53.Re4 Rxe7 54.Rxg4 Re2+ 55.Kxf3 Rxh2
should lead to a draw;
Logical as Black is kept under wraps after 41…Re8 42.Nd6
feels unpleasant! 42…g3 is a fascinating alternative! 43.hxg3 hxg3
44.Rxb8 f2+ 45.Kg2 Nc2 46.Rb7+ Kh6 47.Rb6+ Kh7 (47…Kh5 48.Nd3 Ne1+ 49.Kxg3 f1Q 50.Nf4+ Kg5 51.Rg6#) 48.Rb7+ is a draw!)
42…g3 43.hxg3 hxg3 44.Rxb8 Nc2 45.Rb7+ Kg8 (45…Kh6 46.Nf7+ Kh5 47.Rb6 f2+ 48.Kg2 Ne3+ 49.Kxg3 f1Q 50.Rh6#) 46.f6 (46.Rb8+ Kh7) 46…f2+ 47.Kg2 Ne3+ 48.Kxg3 f1Q 49.Rg7+ Kf8 50.Rf7+ Kg8 is a draw
This is a winning attempt for Black, stopping Rb7+
43…Nd3 44.Rb7+ Kg8 45.Ne4 Nxe5 46.Nf6+ Kf8 47.Nh7+ is a draw
44.Rxb4 axb4 45.Kf2 is a better version of the previous ending. Komodo claims an edge for Black but it’s very murky!
44.Rb7 Rxb7 45.Nxb7 g3 46.hxg3 hxg3 47.Nxc5 f2+ 48.Kg2 Nc2 wins
44…g3 45.hxg3 hxg3 46.e6 Rg7 47.f6 Nc2 48.fxg7 Nd4
wins according to Komodo. Black will follow up with …Ne2+ and …g2+
B. 40.f6 Again trying to use the f-pawn to free the White rook from the pin on the knight on b7
Very natural, but allows a very neat rejoinder
A very strong idea, breaking the pin on the knight on b7 while blocking the e-file with the rook
41…Rxb7 42.Re7+; 41…Kg6 42.Nd6 The key structure which forces Black into passivity 42…Rf8 43.f7+ Kg7 44.Re5 and Black’s pawns start to fall
My line! 43.Nd6 is also strong 43…h4 44.Rf6 Threatening Nf5 and Rh6# 44…Kg7 45.Rf5 Winning according to Komodo (45.Ne8+ Kh7 46.Nd6 Kg7)
44.Re8 Kg7 45.Ne6+ Kxf7 46.Rxf8+ Kxe6 47.Rh8 h3 48.Rh5
b) 40…Re8 41.f7 Rf8
Forcing the pawn to f7 before White can set up the ideal structure with Re6 & Nd6 / Nd8
42…h5 43.Ne4 is very strong: the knight is superb on g5
43.Nf5 Rxf7 44.Nxh6 Rg7 45.Nf5 was my line and only a little better for White;
Stops White from establishing his ideal structure with Re6
The pawn on f7 is hard to capture and takes e8 away from the Black rook
41…Kg7 42.Rb5 Nd3 43.Nxa5 Rd8
44.Nb7 Rd7 45.Nxc5
45.a5 Nb4 46.Nxc5 Rd1+ 47.Kf2 Rd2+ 48.Kg3 Rg2+ 49.Kh4 f2 50.Rb7 Rxh2+ 51.Kg3 f1Q 52.Ne6+ Kg6 53.Nf4+ is an unexpected Komodo draw
45…Nxc5 46.Rxc5 Rd1+ 47.Kf2 Rd2+ 48.Kg3 Rg2+ 49.Kh4
This looks to be a pretty good version: Black really misses …h5 protecting the pawn on g4
i) 41.Rd6 Rxb7 42.Rxd3 Kg6
is fine for Black
ii) 41.f7 Kg7 42.Rb5
Felt like the best way to play this idea: f7 takes away e8 from the Black rook. 42.Nxc5 Rxb6 43.Nd7 Kxf7 44.Nxb6 h5 It’s a lovely tactic, but I was very worried about the result. Komodo thinks it’s lost for White!
44.Nc6 Rd6 45.Ne7 Kxf7 46.Nd5 is a Komodo alternative but after 46…h4 I don’t think that Black is going to have any problems here
44…Rd7 45.Nxc5 Nxc5 46.Rxc5 Rd1+ 47.Kf2 Rd2+ 48.Kf1
ii1) 48.Ke1 Rxh2 49.Rf5 Kf8 50.a5 g3 51.Rxf3 Rh1+ wins
ii2) 48.Kg3 Rg2+ 49.Kh4 Rxh2+ 50.Kg5 Kxf7 wins.
50…g3 was what I had thought but… 51.Kf4 is the amazing Komodo draw! (51.Rf5 Kf8 52.Rxf3 g2 53.Rg3 h4 54.Rg4 Kxf7 was my win) 51…h4 (51…g2 52.Rg5+ Kxf7 53.Kxf3; 51…f2 52.f8Q+ Kxf8 53.Rf5+ Ke7 54.Kxg3 Curses!) 52.Kxf3 Kxf7 53.Rg5 holds for White)
48…Rxh2 49.Rf5 Kf8 50.Rf4
Komodo thinks this is just about ok for White. All feels very fraught! (50.a5 h4 is very dangerous
iii) 41.Rb5 Re8 42.f7 Rxe3;
The same trick as before, but there is a difference…
42.Nd6 Rxb3 The difference with 40…h5: the b-file is open so this wins for Black
is Komodo’s line: White seems to be keeping the knight under control for now, but it still feels messy. Stockfish is very keen on Black!
43…Ne5 44.Rd5 Kxf6 45.Rxc5
45.Rxe5 Kxe5 46.Nc6+ Ke4 47.Nxb8 Kxe3 48.Kf1 h4 wins for Black.
45…Rg8 46.Nc6 g3 47.hxg3 Rxg3+ 48.Kf1 Ng4 is another Black win according to Stockfish
Not the engines’ first choice, but the best attempt in my opinion
and Black’s rook has escaped the clutches of White’s rook and knight. However, his king is now badly-placed
41.Nd6 Rxe3 42.Rb7+
42…Kh8 43.Nf7+ Kh7 44.Ne5+ Kg8 (44…Kh8 45.Ng6+ Kg8 46.f6 f2+ 47.Kf1 Re1+ 48.Kxf2 Nd3+ 49.Kg3 Re3+ 50.Kh4 Rh3+ 51.Kxg4 Nf2+ 52.Kf4) 45.Nxg4 is a definite White advantage
43.f6 Re1+ 44.Kf2 Nd3+ 45.Kg3 Rg1+ 46.Kh4
I thought even this might be winning for White, but that’s not correct.
46…f2 47.f7+ Kg7 48.Ne8+ Kf8 49.Nf6
49…f1Q 50.Nh7+ Kg7 51.f8Q+
is the gorgeous mate that Komodo finds
47.f7+ Kg7 48.Nf5+ Kf8 49.Nxh6 Kg7 50.Nf5+ Kf8 51.Nd6 Kg7 52.Rb8 Ng6+ 53.Kg5 Nf8 54.Nf5+ Kxf7 55.Nh6+ Ke7 56.Nf5+ is a draw too
47…Kh7 48.Kh5 f2 49.Rb7+ Kg8 50.Rb8+
is a draw
What a fascinating ending! It seems therefore that the position after 39.f5 was a draw with best play though the main line is pretty fraught for Black. My 39.Rb5 was a poor move that should have landed me into trouble after 39…Kg8!