Endgame defense can be difficult. Some endgames can be extremely tough to defend. In many, you may be able to spot a light at the end of the tunnel which saves a draw. However, there are times when you’re lost and conventional means won’t save you. Now, I’m not talking about insane swindles drawing from -5 positions. I’m talking about positions where playing the objectively best moves won’t necessarily help you.
What can you do in such positions? You’ve got to spice things up. You’ve got to create confusion. You have to make your opponent think and use his time. Your job is to make your opponent’s life as hard as possible. Is it unreasonable to assume that your opponent will let something slip after long hours of play and his time ticking away? Of course not. Besides, if you’re busted, what do you have to lose?
First, here’s an example of me on the receiving end of this from a couple years ago. After a wild fight (let’s not go there) against a strong IM, I emerged with a winning position. Here’s what happened:
White is a pawn up, and more of black’s pawns are quite loose to say the least. Black’s only chance here is active play. 48.Ra6! is winning here. The point is that after 48… Kd5 49.Nxc6 Rf6, which looks like it wins the knight, white has 50.Ra8! attacking the bishop in return, and he’s just winning after that. I missed this idea, but I chose what seemed to be a reasonable path: 48.Rg7? going after the g6-pawn. Then I got hit with 48… Bf6 49.Rxg6 Ke7
Black is giving up pawns left and right, but Rxc2 will save him. The b2-pawn will fall after that, ruining white’s winning chances. I ended up sacrificing an exchange with 50.Nxc6+ Kf7 51.Rxf6+ Kxf6. I did win black’s remaining queenside pawns at the cost of my c- and h-pawns. Though it wasn’t technically winning, I still had winning chances there, but I butchered those too and the game ended in a draw.
Now it’s not like this was a particularly complicated endgame, but I got bamboozled. All my opponent’s activity forced me to figure things out instead of just converting easily. Anyway, I had been lost earlier in the game, and went on to get my first IM Norm in that tournament, so everything ended happily…
A tricky pawn endgame
I was black against a GM, had been much worse for most of the game, and after a difficult rook endgame reached this position. Though material is equal, black is in awful shape. His king is cut off along the 7th rank, while his counterpart is beautifully active on a5. Black’s pawns have barely gotten moving, while white has a passed pawn which is already on c5. Oh man.
How does white get through? One key idea is to go Ra7 with the idea of winning the a6-pawn. If black cleverly tries to stop that with …Kb8, white will go Rd7-d6, which (could) result in a winning pawn endgame. But ok, white can go 55.Ra7 right here, right now. The a6-pawn is going down, but there’s a catch: black is still kicking in the pawn endgame after 55… g4 56.Rxa6 Rxa6+ 57.Kxa6 h5. While white has connected passed pawns on the queenside, black is getting his own passer on the queenside, and only accurate calculation can determine who will win the race. Long story short, it boils down to a queen endgame after 58.b5 h4 59.b6 g3 60.hxg3 hxg3 61.Ka7 g2 62.b7+ Kd7 63.b8Q g1Q
I saw all this during the game, and I thought it was a draw. I was, however, surprised after the game when I found out that this is mate in 44 according to tablebases. For me, queen + pawn vs. queen make up a mysterious class off endgames, where tablebases are more valuable than Dvorestky’s Endgame Manual. In this endgame, however, white’s queen will do an excellent job shielding the white king from checks (by “counterchecking” the black king), and white’s king will assist the queen in pushing the black king out and escorting the pawn to victory. Still, from the starting position, even getting here isn’t clear—not to mention the evaluation of the endgame.
Instead, my opponent played 55.Rh7. After 55… g4, the pawn endgames are now a completely different story. Still, 56.Ra7! ironically still wins here. The idea is that black is in zugzwang. If 56… Kb8, white goes 57.Rd7!, and the pawn endgame after 57… Kc8 (57… h5 runs into 58.Rh7!, after which black’s position collapses) 58.Rd6 Rxd6 59.cxd6 h5 60.Kxa6 h4 61.b5 g3 62.hxg3 hxg3 63.b6 g2 64.b7+ Kd7 65.b8Q g1Q, and even to my human eyes it’s pretty clear that white is winning after 66.Qc7+ Ke6 67.d7. Still, this is a long line which could easily get “fuzzy” in human calculation, and it’s not unlikely your opponent will miscalculate/hallucinate somewhere along the way. My opponent instead tried a different path 56.Ka4, which doesn’t blow the win but isn’t on the right track. He could attempt to sneak back around with Kb3, but it will run into …g3!, since after hxg3 Rxg3+, white obviously doesn’t have time to take the h6-pawn because he’s in check. He probably didn’t believe the critical lines were winning for him. The game continued 56… Kb8 57.Rf7 Kc8 58.Re7 Kd8 59.Re3? (this is what blows the win), and after 59… h5 I was actually completely all right, since I’ve finally managed to mobilize my pawns on the kingside. After 60.Rg3 we agreed to a draw.
While there were multiple paths to Rome for my opponent, they were all long and complicated. At the end of the day, that’s what saved me.
Sometimes, things just get so complicated that neither you nor your opponent have any idea what’s going on. It can be a curse or a blessing (depending on whose calculations are more accurate), but if there are no reasonable alternatives, mutual confusion is not a bad idea.
Here’s a final example from one of my own games. I had been a bit worse for a while, and after both my opponent and I built up our positions a bit, things exploded.
(I was black) This goes to the old land of queen + knight vs. queen + bishop, filled with passionate debates about which combination is better… My personal rule of thumb is that, no matter the situation, whichever duo I have is worse. Just kidding, but it really depends on the situation. Here, white is for choice, not only because his queen + knight duo is better than queen + bishop, but because he has a passed b-pawn that is very dangerous. Black’s king is also fairly exposed, due to the kingside expansions, while white’s king is still fairly safe for the moment.
White is actually winning here with the move 50.Qb1!, hitting the f5-pawn and aiding the passer at the same time. After 50… Be6 51.hxg4 fxg4 52.b6 g3, however, it gets messy. The most natural line goes 53.b7 gxf2+ 54.Kxf2 Qf4+
The b-pawn is beyond black’s control, and white’s only job is to get his king to safety. The most natural move is to run back with 55.Kg1?, but that actually blows the win. After 55… Qxe3+ 56.Kh1, the desperado 56… h3! actually secures a draw! Yes, white can go 57.Qf1+ Ke7 58.b8Q, but after 58… hxg2+ 59.Qxg2 Qe1+ 60.Kh2 Qh4+, white can’t escape the checks.
Instead, white has to play 55.Ke2!, ironically going into the middle. The white knight, however, covers everything, and black has nothing. After 55… Bg4+ 56.Nxg4 Qxg4+ 57.Kf1, white’s king will run to the corner, and …h3 desperados no longer work, since black no longer has a bishop to provide backup.
Instead, my opponent played 50.hxg4? fxg4 51.b6!? (if 51.Qb1 black has Bg6! attacking the queen). Here’s where I returned the favor and slipped up as well.
51… Qxb6 52.Nxg4+ is obviously white’s idea, but I had 52… Kf5. What I missed was rather embarrassing: I thought that white had 53.Qxe5+ Kxg4 54.f3#, but the f-pawn is pinned. Oops!! After 52… Kf5, black is also all right after 53.Nxe5 Qe6 54.f4 Kxf4, where the situation looks scary but is actually harmless.
Instead of that, I played 51… Be6?, giving my opponent another chance to return to the winning variation by playing 52.Qb1!. I naturally didn’t know that that variation was winning during the game, and neither did my opponent, because he instead played 52.b7? Qb6 53.Qd1, giving up the b-pawn with the aim of collecting my kingside pawns instead. But after 53… g3! 54.Qf3+ Ke7 55.fxg3 hxg3 56.Qxg3 Qb1+ 57.Nf1 Qxb7, black is holding. The game was soon drawn.
These variations are naturally difficult to see during a game, especially after a long fight and with little time, and that’s what you’ve got to use to both ends. However, the moral of the story is clear: the more complicated the win is, the less likely your opponent will play it. Complications aren’t necessarily your enemy when defending. They can be a lifeline if conventional defensive means fail. And, as I hope these examples illustrate, complicating matters can save you half points here and there that just giving up wouldn’t do.