"Analysis of a difficult position" by Paul Keres

The games don't get adjourned anymore due to the possibility of computer analysis. Nevertheless, the ability to analyze complicated position is an important measure of a chess player's class. So Keres's lecture is still of much interest.

Each chess player ever taking part in a tournament understands the concerns of a player adjourning the game in a complicated, unclear position. Such chess player often gets back to his adjourned position again and again, seeks and tests various continuations.

Adjourned games bring many concerns. The way to analyze them depends on personal tastes more than anything. I don't know if our Grandmasters use different analysis methods because no-one of them ever said anything on that topic.

In the modern competitions, the analysis of adjourned positions became very important, and I think that more attention should be paid to this aspect of chess creativity.

I'm not trying to theoretically prove the methods of analysis of adjourned positions. I'll just try to get the reader into the creative lab of a Grandmaster and show them what happens while thousands of chess enthusiasts impatiently wait for the game to continue.

Making my last move before adjourning, 40... a3, I felt great because the queening of the pawn looks inevitable. For instance: 41. Ra8 a2 42. Ne5 Nb4 and Rb1 -+. White can't prevent this to happen, it was no doubt an easy win, and I was glad that I won't have to work hard on the home analysis.

But it turned out that actually it all wasn't so easy. First of all, I concluded that White has no direct defence from the Black pawn queening, because the White Knight cannot hold it. So all I had to do is understand whether White has any counterplay on the other side of the board. If they don't, then the position can be evaluated as easily won for Black, and stop the further analysis.

At the first glance, it can be seen that by moving the pawn to h5 White can create mating threats to the Black King. This threat, however, is easily met: 41. h4 a2 42. h5 Rh1+ 43. Kxh1 a1=Q+ 44. Kh2 g6.

But can White improve this variant? I found another move order: 41. h4 a2 42. Ra8 Nb4 43. h5, and it's not that easy to repel these threats. Though Black still win after 43... g5!, both after 44. hxg6+ Kxg6 45. Ne5+ Kf5 46. f3 Rf1 and 44. Ra7 Rc1! White is defenceless. They don't have a perpetual check: 45. Nxg5+ Kg8 46. Ra8+ Kg7 47. Ra7+ Kf8 48. Nxe6+ Ke8 or 45. Ne5+ Kg8 46. Ra8+ Kg7 47. Ra7+ Kf8, and the Black King escapes to the queenside.

Thus, the White's threats weren't satisfying. Nevertheless, the analysis indicated that their possibilities weren't exhausted yet. After 41. Rh8+ Kg6 42. Ne5+ White gains an important tempo. But there are no direct mating threats anymore, and Black can queen their pawn. After the Black King escapes the checks, the game is won. All that looked convincing, but all variants should have been checked.

So, where the Black King can go? It has four squares to escape the check, and I considered the moves one by one.

The move 42… Kh5 seems dubious because the King gets caught in a narrow cage on the edge of the board. After 43. Rg8 White creates dangerous threats: 43… Rf1 44. Ra8 Rxf2 45. Rxa3 =. It’s clear that 42… Kh5 wasn’t worth considering.

Does 42… Kg5 lead to a win? At first, I didn’t want to consider this move at all because of 43. Rf8!, creating threats: 43… a2 44. f4+ Nxf4 45. h4+ Kxh4 (or 45… Kh5 46. Rxf4) 46. Rxf4+ Kg5 47. Rf2! Rh1+ 48. Kxh1 a1=Q+ 49. Kh2, and it’s unlikely that Black can win here.

Then I tried 43… Nf6!, and the winning hopes returned: 44. Rf7 Kf5! 45. Rxg7 Ne4, and Black get the Knight to the queenside and make sure the pawn gets queened.

Still, I didn’t like 43… Nf6 because it gave my partner some counterplay possibilities. After further analysis, I found that 44. Ra8! poses new and quite difficult problems for Black. For instance, they threaten 45. Nc4 a2 46. Ne3 with subsequent 47. Nc2 and Rxa2; the threat can’t be met with Kf5 or Kf4. So Black has to play 44… a2, but White has a strong reply: 45. f3!

It seems that Black retains his winning chances. He needs only to get his King on e3 in time or, countering the immediate threats to the King, perform the maneuver Nf6-d5-b4 and Rb1. This plan seems very simple but it’s quite difficult to carry out. The straightforward 45… Kf4 is met with 46. Nd3+ Ke3 47. Nb4, winning an important passed pawn.

White doesn’t pose any immediate threats, so worth considering is 45… h5, intending to block the White kingside pawns with 46… h4. But White can move 46. h4+! Now, if 46… Kxh4, then 47. Ra5! Kg5 (not 47… Nd5? 48. f4! Nxf4 49. Nf3+ and 50. Rg5#) 48. Nd3+ Nd5 9. Nb4, and White capture the a-pawn, with serious chances of drawing the game. And if 46… Kf5, there’s 47. Ra5 Nd5 48. g3!, and the Black pieces’ positions are so awkward that it’s unlikely they can win. For example, after 48… g5 49. g4+ hxg4 50. fxg4+ and 51. h5 or 49… Kf6 50. gxh5 gxh4 51. h6 White have a very dangerous passed h-pawn.

Analyzing the variants and seeing that neither 45… Kf5 46. Ra5 Nd5 47. g3 nor 45… Nd5 46. g3 h5 47. h4+ Kf5 48. Ra7 helps Black to win, I decided not to play 42… Kg5.

During the analysis, I understood more and more that I vastly overestimated my adjourned position. Though I was still hopeful, because it seemed that 42… Kf5 wins an important tempo because White can’t let the King to d4. For instance, Black win easily after 43. Ra8 Ke4 44. Nc4 a2 45. Nd2+ Kd3 46. Nb3 Rb1 47. Nc5+ Kxd4 48. Nxe6+ Kd5 etc. 43. Nd3 or 43. Nd7 isn’t satisfactory too because of 43… Nc7 (not 43… Ke4 44. Nc5+). So 43. f3 is almost forced, and it might seem that 43… Kf4! is decisive.

But how will the game continue if White moves 44. Ra8!? Now after 44… Ke3, there’s 45. Rxa3+ Rxa3 46. Nc4+, winning the passed pawn, and after 44… a2 45. Ra3, a position with no clear way to win appears.

The variants 45… g5 46. g3+ Kf5 47. g4+ Kf6 48. Ra7 and 45… h5 46. h4 g5 47. g3+ Kf5 48. g4+ hxg4 (48… Kf6 49. gxh5 gxh4 50. h6) 9. fxg4+ Ke4 10. h5 don’t gain anything for Black. 45… Nc3 is also not enough due to 46. g3+ Ke3 (not 46… Kf5 47. Ra7! with mating threats) 47. Rxc3+ Kxd4 48. Ra3 Kxe5, and the chances of Black winning the rook ending aren’t too great.

Even weaker is 45… Ne3 because White gets a choice between the simple 46. Nd3+ Kf5 47. Nb4 and the subtle 46. g3+ Kf5 47. Ra7!, and Black can’t play 47… Nf1+ 48. Kg2 Nxg3 because of the sudden 49. Kxg3! Rg1+ 50. Kh4, and White win because of the threat 51. Rf7#.

After this hard analytical work, I decided to check the last variant again and find some improvement. First of all, I considered the immediate 43… a2, forcing 44. Ra8. After that, 44… Kf4 5. Ra3 or 4… h5 5. h4 leads to the positions we have already seen, and 4… Nb4? 5. Rf8+ Kg5 6. f4+ leads to mate.

Maybe then, it was worth to try and get the King to f6, intending to reach the queenside? If White tries to prevent that with 45. Ra7, then 45… Kg5, trying to escape on f4 or h6. After the analysis, I had to abandon this idea altogether, because White has enough defence resources. For example: 46. g3 h5 47. Rxg7+ Kh6 48. Ra7 Nb4 (or 48… h4 49. gxh4 Nb4 50. Ng4+ Kh5 51. Ne5!) 49. h4 Rc1 50. g4 hxg4 51. fxg4, and Black can’t queen their pawn due to mating threats.

So 42… Kf5 was also refuted. What’s the sense of making the King go into the mating net, if after 42… Kf6 the King can easily reach the queenside? White can give some annoying checks, but perhaps there’s nothing terrible in that?

All that should have been checked, and the exhausting analysis turned in that direction.

White play 43. Ra8, and now if Black are to avoid the variants after 44. Ra7, they are forced to play 43… Ke7. Then, after 44. Ra7+ Kd8 45. Nc6+ Ke8, or 45. Nf7+ Kc8 46. Nd6+ Kb8 47. Rb7+ Ka8, or 45. Ra8+ Kc7 the Black King escapes the checks. Finally, it’s all right!

But just in case, it’s worth to check the variant again. Unfortunately, it also led to no good. After 42… Kf6 43. Ra8 Ke7 44. Ra7+ Kd8 White can play 45. Rxg7, and Black again faces the familiar challenge. How to queen the a-pawn now?

45… Nb4 doesn’t meet the objective because of 46. Rb7 Rb1 47. Rxb4 a2 48. Ra4 a1=Q 49. Rxa1 Rxa1 50. Nf7+ and 51. Nxh6 or 46… a2 47. Rxb4 Rh1+ 48. Kxh1 a1=Q+ 49. Kh2, and Black has no winning chances. Slightly better is 45… Kc8 to meet 46. Nc6 with 46… Rc1, but 46. Ra7 Nb4 47. Ra4 a2 48. Rxb4 also leads to a drawn ending.

After 45… a2 46. Nc6+ Ke8 47. Ra7 it’s also hard to win for Black. 47… Rc1 is met with 48. Ra8+, 49. Ne5+ and 50. Rxa2, and 47… Nc3 – with 48. Nb4 Rb1 49. Nxa2 Ra1 50. Ra3 Nxa2 51. Ra7. There’s also 47… Kf8, threatening 48… Rc1, but White still can defend after 48. h4!, for example: 48… Rc1 49. Ra8+ Kg7 (49… Kf7 50. Ne5+) 50. Ra7, and Black can’t play 50… Kf6 due to 51. Ne5!, threatening Rf7#.

Black have other continuations, but neither of them gives clear chances given White’s perfect play due to Black King’s bad position.

Trying all the possible King moves after 41. Rh8+ Kg6 42. Ne5+, I suddenly understood that none of them give real chances. I abandoned all the dreams of easy win, I couldn’t even see how to get any chance to win. Was the adjourned position unwinnable?

Again and again I considered the possible moves, failing to find anything new. Very soon I rejected 42… Kh5 and 42… Kg5, then also 42… Kf6. Finally, I was sure I had to find improvements after 42… Kf5 43. f3 Kf4 44. Ra8.

After a long study, I decided that after 44… a2 45. Ra3 Kg5! Black has some winning chances. Earlier, I rejected this variant because of 46. h4+ Kxh4 47. f4 Nxf4 48. g3+, capturing the Knight. But more thorough analysis showed that instead of 47… Nxf4? Black can play 47… Nc3! Now if 48. Rxc3, then 48… Rh1+ 49. Kh1 a1=Q+, and after any other moves the simple 48… Rb1 and 49… a1=Q wins. If White play 46. g3, then Black replies with 46… h5, getting a safe haven on h6 for the King.

But those variants, while good for Black, had their downsides too. For example, instead of 46. h4+ or 46. g3 White could play 46. Ra8 to get a drawn position after 46… h5 47. h4+ Kf4 (47… Kxh4 48. f4 or 47… Kf6 48. Ra7) 48. Ra3. Thus, Black has to play 46… Nb4, but then follows 47. h4+ Kxh4 (47… Kf6 48. Ra7 or 47… Kf4 48. Ng4 creates mating threats) 48. Ra5! Nd5 (or 48… Kg5 49. Nd3+ Nd5 50. Nb4 etc.) 49. Ng4 with annoying threat Ng4-e3-c2. If 46… Nc3 47. h4+ Kxh4 48. Ra5 Kg5, White has 49. Nd3+ Kf6 50. Nb4 Rb1 51. Nxa2 Ra1 52. Ra3 Nxa2 53. Ra7, and it’s not easy to win for Black despite an extra Knight.

Finally I managed to find a variant where Black at least retain some practical winning chances. It’s hard for the defending side to find a series of the only good moves, if this variant hasn’t been thoroughly analyzed at home. The result, however, didn’t satisfy me, so I reserved it in case there’ll be no clear win.

Finally I found a shockingmove: 44… Ke3!! Black willingly give away their strongest asset – the passed a-pawn. Do they hope to win a Knight endgame, being a pawn behind with all the pawns concentrated on kingside?

It comes out that this endgame gives Black best chances. 45. Nc4+ Kf2 46. Nxa3 Ne3 loses immediately, so White has to capture with the Rook. After 45. Rxa3 Rxa3 46. Nc4+ Kf2 47. Nxa3 Ne3 Black retain the sacrificed pawn. But this is not the most important thing: much more important is that the White King’s position is bad, and his Knight is too far away from the kingside. As a result, Black captures at least one more pawn, which gives him decisive advantage.

Now, the variants. White can’t protect the g2 pawn: 48. g4 Nf1+ 49. Kh1 Ng3+ 50. Kh2 Kxf3 51. Nb5 Ne2 leads to loss of the d4 pawn and hopeless endgame for White. Also bad is 48. Nb5 Nxg2 49. Nc7 Nf4 and 50… Kxf3. So White is confined to the variants where he gets a passed pawn which can give some problems to Black. 48. f4 Nxg2 49. f5 exf5 50. d5. Now it’s easy to see that Black can quickly use their advantage: 50… Nf4 51. d6 g5!, and if 52. d7, then 52… g4 53. hxg4 fxg4 54. d8=Q g3+ 55. Kh1 g2+ 56. Kh2 g1=Q#.

So the adjourned game problem was solved. My partner also thoroughly analyzed the position, but my move 44… Ke3 was totally unexpected. Staring at the board in disbelief, Szapiel quickly captured on a3. Only after 47… Ne3 he had a long think, but couldn’t defend.

White resigned after 52. Nc2 g4 53. hxg4 fxg4 54. Ne1 g3+ 55. Kh1 Kxe1.