Saturday, July 1, 2017

The Art of the Miracle Draw

I have been a tournament chess player in the city of Charlotte for over twenty years now. Over the course of the years, I have had a lot of games that have resulted in what many would refer to as a "miracle draw". Even as recent as last year, I've heard comments from various members of the club wondering how on earth I've pulled off as many draws as I have from either vastly inferior or even completely lost positions. In the vast majority of cases, it's merely a blunder by the opponent, whether it be due to a tactical oversight or severe time trouble. I have also been on the wrong side of this numerous times, where I go from a completely won position to a draw or even a loss.

However, not all draws from an inferior or lost position are total luck. There is actually an art to surviving lost positions. There are a number of ways to do this, three of which you will see here in this article. These include perpetual check, stalemate tricks, or creating chaos on the board to try to confuse matters, and in many cases, combining more than one may be necessary to pull it off.

However, above all other factors, one thing is critical above and beyond anything else if you plan to succeed in snatching those half points away from your opponents, and that is being in the right frame of mind to create these draws. In order to succeed, the player must be able to do each and every one of the following:
  • First and foremost, recognize that your position is clearly inferior to that of your opponent's, and acknowledge that any playing for a win is a complete pipe dream and that it just isn't happening. It is also critical that this is recognized as early as possible. More on that in a moment.
  • Once it is recognized that you aren't winning and are now in the frame of mind of playing for the draw, you need to familiarize yourself with many of the well-known drawing patterns. It's easier to figure out where your remaining pieces need to go in order to possibly achieve the goal.
  • Make sure that the position is still complex enough such that there is room for your opponent to error.
  • The best move according to a computer, resulting in the smallest "numerical" disadvantage, is not always the best move for the situation at hand given the human aspects of over the board play.
  • Understand that, given the fact that you already recognized that you are either inferior or outright lost, you will not always succeed in drawing. The best you can do is disguise and complicate the matter.

So now let's look at some of the various techniques to snatch the half point.

Perpetual Check

This is probably the most basic way to achieve a draw in a lost position. That said, because of its simplicity, it is also the easiest threat for your opponent to recognize. Therefore, in addition to creating threats to repeatedly check the opponent's king, there needs to be a certain level of disguise in the threat. A combination that is complicated and deep enough for the opponent to overlook. If the threat is a one move threat, you might as well resign and not waste your time. Take the following hypothetical position:

A hypothetical position

Black can safely resign this position. He is down an exchange and two pawns, and he has nothing here to play for. Sure, Black could say "Uh, I could play 1...Qd6 or 1...Qg5 and maybe White won't see my threat to draw with 2...Qxg3+!". Remember in the first bullet that it was said that it is critical that you recognize the fact that your position is inferior as quickly as possible. Here, it is too late. The threat is a one move threat, and there are many ways for White to avoid the draw, whether that be 2.Kh2, 2.Kg2, 2.Qd3, 2.Qf3, 2.Qg4, etc. This is what is meant by the fact that it needs to be complicated enough to disguise your intention.

Now let's look at a more realistic opportunity.

Klaus Pohl - Patrick McCartney, Columbia Open, 2013

White is down a pawn, but the position is actually completely winning for White. That said, Black also has a nasty trap, and it was complicated enough for White to walk right into it. Do you see it?

White here played 29.Rxb7! Rxb7 30.Rxb7??. It looks like Black is dead. Mate is threatened via 31.Qg7#. It appears as though Black's pieces are overworked. The rook and queen are guarding the knight. So a move like 30...Re7 would drop the knight and lose instantly. That said, what White overlooked is that lifting the rook off the back rank has allowed a rook sacrifice followed by perpetual check. 30...Re1+ 31.Kh2 (31.Kf2?? Qe2+ 32.Kg3 Qg4+ 33.Kf2 Re2+ followed by 34...Qxg2 is mate) 31...Rh1+!! and a draw was agreed because after 32.Kxh1, Black has a perpetual with 32...Qe1+ 33.Kh2 Qxh4+ 34.Kg1 Qe1+ etc. Because the threat was well disguised, White missed it.

Instead of 30.Rxb7, White can win if he inserts the move 30.Qg5+ before capturing on b7 for two reasons. The first is that by moving the Queen to g5, the h4-square is guarded and Black has no perpetual check. The second is that capturing on b7 the following move will now come with check and so White gains an extra tempo, and with the b-pawn being a critical pawn that was holding the position intact, Black's position rapidly falls apart and Black could safely resign at that point.

Stalemating Tricks

The next example will illustrate a second method of pulling off the half point. Once again, we are going to look at a position that is, for all intents and purposes, lost. Take a look at the following position:

Patrick McCartney - Samuel Xin, SC Championships 2008

Black's last move was 32...g5, where the pawn was originally on g7. White is down a pawn, and even then, White's pawns are not healthy. For starters, he has crippled pawns on the Kingside via doubled g-pawns. While doubled pawns may be strong in many middle games, they tend to be a liability in the endgame. That said, Black's last move wasn't very good. He still has a significant advantage, and according to the bots, taking en passant is White's best move, giving Black roughly a point advantage, implying that White has little to no compensation for the pawn. This brings up another point when having a vastly inferior or lost position. The best move is not always the "best move"! What is meant by that is often times the move that is theoretically best according to computers may lead to the smallest advantage for the opponent from a theoretical standpoint, but in human play, there is the practical factor, and rather than sit there, suffer, and just wait for Black to defeat White, White sees a potential drawing pattern after Black's last move, this time in the form of stalemate. If White can entice the Black rooks off of the second rank, he can advance the g-pawn to g3, park the king on h3, advance or exchange the a-pawn and f-pawn such that they are locked, eliminate one pair of rooks, and sacrifice the other rook on the second rank. White manages to pull this off, pretty much exactly as planned. Let's see how it was done.


Not accepting the en passant offer and instead working on setting up a stalemate cage.

33...b5 34.axb5 axb5 35.Rd7 Re3+ 36.Kh2

White is not ready yet to advance the g-pawn as 36.g3 Ree2 forces the White Rook into a passive position on h1 to avoid mate.

36...Re4 37.Kh3 Rb3+

Of course, if 37...Re3+, White will again just retreat the king to h2. Our goal is to draw, and so White has no objection if Black just wants to agree to it now via a repetition of moves.


Now that both rooks are off the second rank, White has time and can now advance the g-pawn.

38...Rb2 39.f6!

Locking his last pawn.


The point behind waiting for both rooks to exit the second rank and gaining the necessary time to advance the f-pawn is seen if Black plays 39...Ree2 here, which White will respond with 40.Rg2!, forcing one set of rooks off the board, and if Black takes the rook with 40...Rxg2??, White will draw with the "Eternal Rook" with 41.Rd8+ and then continuing to check the Black King to eternity. If the King ever takes the Rook, White is stalemated!


White now threatens to use the eternal rook trick with both his rooks! Throw one away and then throw the other one away, and because of the rook on b2, White would once again be stalemated!


So Black retreats to avoid the draw, at least for now.

41.Re7 Rf8

Taking the rook on e7 would actually lose for Black after 41...Rxe7?? 42.fxe7 Re2 43.Rc8+ followed by 44.e8=Q and White wins the rook.


Black has got to have a splitting headache by now. White is attempting to sacrifice both Rooks for yet another stalemate. The c8-Rook is poisoned as 42...Rxc8?? 43.Re8+ Kh7 44.Rh8+ Kg6 45.Rxh6+ Kxh6 is again stalemate. Therefore, Black does a tricky rook trade, eliminating his own rook on the second rank to rid himself of the stalemate threats that White is executing.

42...Rh2+ 43.Kxh2 Rxc8 44.Rb7

Even when attempting drawing tricks, many basic principles still apply, and one of them is that rooks belong behind passed pawns.

44...Rc4 45.Kh3 Rd4 46.Rc7 Rd1 47.Rb7 Rb1 48.Ra7 b3 49.Rb7

White is just biding his time, waiting for Black to advance the b-pawn to the second rank, at the same time staying on the 7th rank to keep the Black king out of the picture. Black next tries to bring the king in the other way, but White will have none of that!

49...Kh7 50.Rb5 Kg6 51.Rb6

What does Black do now? If he moves the king back to h7 he makes no progress. If he moves the rook to the second rank he has to watch again for stalemates and blocks his b-pawn from advancing. If he moves the Rook laterally, White takes the b-pawn, except in the case of 51...Rh1+ 52.Kg2, but then Black has to move back to the b-file and White just moves the king back to h3. So that leaves only one possibility, but it doesn't work either.

51...b2 52.Rxb2!

Once again, if Black takes the rook, it's stalemate.

52...Rh1+ 53.Kg2 Re1 54.Kh3 Kxf6 55.Rf2+ Ke6??

Now Black finally buckles and walks into the draw. He had to endure the complicated ending by moving his King to the g-file. If he advances the f- or h-pawn, looking to eliminate the stalemate cage, many 2-on-1 positions with a rook each are drawn, and so Black would have to be very careful if he wants any chance to win.

56.Re2+!! Rxe2 STALEMATE!

Creating Chaos

Sometimes simple methods like perpetual check or stalemate isn't immediately available, and one or the other can only be created through creating chaos in the position. Similar to the stalemate scenario above, this often may entail surrendering the best move for the more complicated one that features a better shot at human error. I should mention that those of you that are Diamond members on, Ivan Sokolov has 3 excellent videos on this exact topic, and I recommend observing all of them. For now, let's take a look at the following position:

Patrick McCartney - Karthik Rangarajan, 2005

Here we have a case of recognizing White's problems. The obvious issue is that he is down a piece for two pawns, but sometimes those extra pawns can mean something. However, in this case, Black has a major threat. He is looking to check the White king on b1 followed by capturing the b-pawn with the queen. After that, all squares will be protected for the black c-pawn to walk down and promote itself. The critical thing to observe is that White's pieces are being somewhat overworked. If White plays 29.Qe2?!, the queen no longer guards the bishop, and after 29...Bd6, Black would be threatening to win a piece by removing the guard on f4 with the bishop and capturing the then hanging bishop on h3. If White guards the bishop with a move like 30.Kg2, then after 30...Bxf4 31.exf4 Qg6+ 32.Kf1 Bd7, White hasn't lost any additional material, but just look at the pawns. They are a complete wreck and White is positionally busted. It also doesn't help that Black was able to eliminate another set of pieces. White needs to come up with something more chaotic, leaving room for error by Black. What other plan might White try for if he can't do anything about the b-pawn without train-wrecking the rest of his position? Well, the first thing to do is look at the geometry of the White pawns. When Black checks the White king, it will move up to the second rank, which is important. The capture of the b-pawn results in the Black queen landing on a dark square. The promotion square for the pawn is also dark. White has all the dark squared diagonals blocked by his own pawns, and with the king not on the back rank, the promotion will not be with check. So if White can maybe eliminate the minor pieces around the Black king, White can maybe draw via perpetual check down an entire queen because both queens will be on dark squares if Black takes the most rapid approach to promoting the c-pawn. Therefore, White abandons the b-pawn and tries to clean house around the Black king. Once again, what happens is not totally forced, but it goes to show that even a 2400 player can be tricked.

29.Nxd5 Qb1+ 30.Kg2 exd5

Believe it or not, this move is actually a mistake and the game is probably already drawn for White. Instead, Black should play 30...Qg6+ 31.Kf1 Bb7!, winning. There may be other lines with winning chances, like the immediate 30...Bb7 or 30...Bd7, but the queen check and pin is the simplest. Sometimes it's just amazing how much of a difference a move can make, and what appears to be the most obvious turns out to be the ultimate error!

31.Bxc8 Qxb2

So now if Black goes for the second queen, it will take him 3 moves to execute it. Therefore, White has 3 free moves, and then all subsequent moves will need to be with check or else White is lost. Keep in mind that moves that gain a tempo on Black, like checks, don't count. He can make 3 non-forcing moves before he is confined to checking the Black king. In the game, Black does go for the queen, but even if he tries to spend time moving his queen to a light square before trying to advance the pawn, it's too late.


Free move number one.

32...c3 33.Bxd5+

Check, gaining a tempo. This move doesn't count towards White's allotted free moves.

33...Kf8 34.Be6!

Free move number two. 34.Qf4 also draws, but all other moves lose.

34...c2 35.Qa8+ Kg7 36.Qb7

Now it's Black that has to be careful!


The only move that draws! 36...c1=Q?? leads to mate in ten starting with 37.Qxe7+.

37.Bxg8 c1=Q 38.Qxe7+ Kxg8 39.Qe8+

And with neither of the Black queens able to come to the rescue as both are blocked by White pawns, there is no way for Black to get out of check and a draw was agreed here.

An Example Where the Winning Side Wins

In this example, I am going to display the entire game because, to this date, this is the highest rated opponent I have ever beaten. I have drawn a couple of players higher than this, but as of right now, this game is what I would consider "McCartney's Immortal", similar to Kasparov's game against Topalov in 1999.

Alexander Matros (2447) - Patrick McCartney (1999), Columbia Open, 2010, Larsen's Opening


It is not unusual for IM's to play an opening like this against a player over 400 points down. The idea is to avoid theory, and simply trusting your skill to outplay your opponent through a simple game of chess. The only problem here is that he ran into a buzz saw in the form of my actually knowing Larsen theory.

1...e5 2.Bb2 Nc6 3.e3 d5 4.Bb5 Bd6 5.f4 Qh4 6.g3 Qe7 7.Nf3 f6 8.fxe5 fxe5 9.Bxc6 bxc6 10.Nxe5 Nf6 11.Nd3?!

This move is dubious. The main line is 11.Nxc6 Qe4 12.O-O Bh3 13.Rf2 Ng4, which is theoretically equal.

11...Qe4 12.Nf2 Qg2!

White is already lost here. He still had to castle on move 12 and Black would follow up the same way as in the main line, but with one more pawn than if White had taken on c6.

13.Qe2 Ng4 14.Qf1 Qxf1+ 15.Rxf1

15.Kxf1?? loses even faster to 15...O-O -+.

15...Nxh2 16.Rh1 Bxg3 17.Kd1 O-O 18.Nd3 Bg4 19.Kc1

So after White's fatal error on move 12, both sides have since made the best move every time. The critical thing for Black here, when playing up almost 450 points, is not to relinquish the advantage. Not every move will feature "only moves", and you don't always have to play the best move, but you can't afford to slip back into an equal position because unlike facing an 1800 player, you aren't going to get a second chance if you slip up.


According to artificial intelligence, 19...h5 is a tish better, but the move played is fine and doesn't destroy Black's winning advantage.

20.Rxf1 Nxf1 21.Ne5 Be6

Once again the computers prefer 21...h5.

22.Nxc6 h5

Now Black gets the ball rolling, and through careful analysis, one can realize that there is no way to stop the h-pawn from promoting.

23.Nc3 h4 24.Ne2 h3 25.Nxg3 Nxg3 26.Be5 h2

So now that we have the position that goes along with our topic, let's have a look at a diagram.

Here is another case where computer numerical assessment and human aspect don't match. Even after the best moves according to a computer, the position is, at best for White, minus 3. At this point, what is the difference between minus 3 and minus 5? Nothing really. Therefore, White is going to play the move that creates the most chaos on the board, and makes matters the most complicated for Black, rather than a move that the computer deems best but results in a cake walk win for Black.


Of course computers are going to say that White should play 27.Kb2 h1=Q 28.Rxh1 Nxh1. But all this does is lead to a position where Black has a rook for a pawn. There is no other imbalance in the position. By taking the Knight instead, Black does get to keep his queen, but White then has a piece that can function in a way that no Black piece can. The knight. White will try to use that, the extra pawns, and the opposite colored bishops against Black in return for his queen. This may be a bigger advantage for Black, but it's one that still requires some level of accuracy.

27...h1=Q 28.Kb2 Qg2 29.Be5 Rf8 30.Nxa7 Rf1 31.Rxf1 Qxf1 32.Nc6 Bd7 33.Nb8 Qb5 34.Bxc7

So now we have a very odd endgame. White has a knight and three pawns for the queen and the bishops are of opposite color. If Black can ever break up the White pawn chain, Black should win. White does all he can to try to force Black into making an error.

34...Bf5 35.Be5 g5 36.a4 Qc5 37.c3 Qb6 38.d4 g4 39.a5 Qh6 40.Bf4 Qh1

Thus far, Black has succeeded in not relinquishing the advantage, but he has paid a major toll for it in the form of time.


41.a6 should be answered by 41...Qe1 42.a7 Qd2 43.Ka3 Qxc3 and the a-pawn will fall. The square the knight would need to go to in order to cover the pawn is guarded by the queen, which when the square the queen sits on would cover the promotion square and the pawn can be rounded up. Otherwise, Black will get a skewer or a fork, depending on what White does, and round up the a-pawn. So White holds off on advancing it.

41...g3 42.Bxg3 Qg2 43.Ka3 Qxg3 44.Nc5 Qxe3 45.Kb4 Qe7 46.a6 Bc8 47.Kb5 Qa7 48.Kc6 Bxa6 49.Kxd5

I remember at this juncture I was down around (possibly below) ten minutes for the rest of the game. While it is still winning for Black, it is not easy with no Black pawns left.

49...Qf7 50.Kc6 Qg6 51.Kc7 Be2 52.c4 Qg3 53.Kc6 Bd1 54.b4

Mission accomplished! Black forced White to advance the b-pawn, forming a lateral pattern and making the center pawn weak. Break up the pawn chain and Black wins.

54...Qc3 55.d5 Qxc4 56.d6 Bf3 57.Kc7 Qxb4 58.Nd7 Kg7 59.Kd8 Qxd6 and White threw in the towel.

Well, this concludes this article on The Art of the Miracle Draw. We talked about various concepts such as switching gears towards playing what leads to the most complications for the winning side, not necessarily the theoretically best moves, and some of the various ways to pull off such a draw, whether that be perpetual check, stalemate, or simply confusing the opponent to no end until he hopefully buckles, something that we saw in three of the four examples in this article.

Until next time, good luck in your games.

ADDENDUM - You may have noticed that both articles thus far have used static diagrams with moves listed out rather than boards with the moves embedded within. This is intentional, and likely to be the format of the vast majority of my articles. The reason is two-fold. 1) Many of the games I analyze will be analyzed deeper than a mere sentence per move. Scrolling through annotations in the small, tiny window can be very annoying for the reader to have to deal with, and 2) I am of the firm belief that physically making the moves yourself forces the mind to absorb more than merely clicking a button to go through moves. Analysis on an actual board is closer to actually playing a game than clicking through a 2-dimensional board at a rapid rate.

No comments:

Post a Comment