Winning through willpower

Are you watching closely?

Every great chess game consists of three parts or acts. The first part is called, well, the opening. The players show you something ordinary: a Ruy, an English, or — in this case — a Grunfeld. Perhaps you check against the database to see if it is indeed real, unaltered, normal. Of course… it probably is. The second act is called the middlegame. The players take the ordinary something and make it do something extraordinary. Or sometimes they don’t:

PonoCarlsen1.png

Now you’re looking for the reason I’m subjecting you to this very very forced reference… but you won’t find it, because of course you’re not really looking. You don’t really want to know. You want to be… fooled.

Of course, you’re probably not impressed yet. That’s why every chess game has a third act, the hardest part, the part we call… “Magnus Carlsen popping off” (yes, this is the official term):

PonoCarlsen2.png

Ok, why am I showing you this nonsense? Well, every chess player has, at some point, witnessed a game that made a huge impression on them in some way. One of Kasparov’s many tactical melees, Tal’s sacrificial attacks, or Karpov’s positional dominations tend to be popular choices, as do games featuring incredible moves (e.g. Khismatullin’s brilliant 44. Kg1!! against Eljanov) or games with external importance (e.g. world championship-ending queen sacrifices). My formative game, displayed in two compelling images above, features nothing wild, nothing particularly counterintuitive, and was even a rapid (!) game. Yet the game was… magical. From one of the most boring positions you could possibly construct, Carlsen — cleanly (no blundering a rook or anything) — slowly outplayed a player (Ponomariov) known for their endgame ability over the course of 70+ moves!

Now, of course, while Carlsen is an incredible player it’s not like Ponomariov had any business losing this position. Personally, I don’t even think getting from point A to point B was a result of skill difference: rather, Carlsen just wanted to win badly enough that fate forced Ponomariov to comply. In doing so, Carlsen taught me (and us) two things: first, this Carlsen guy is just insane — at least for me, this was the first time I really realized that Carlsen was something special (and this was after he was #1 in the world). But more practically, that pretty much any position is winnable even against the best of opponents, and given that you and I aren’t exactly consistently playing against the best in the world (though if you are and you’re reading this, please sign my forehead) even the most apparently drawn positions are often very much alive.

With just this instance, you’d certainly be forgiven for thinking that the above was simply a once-in-a-lifetime fluke, but fortunately I’ve come armed with personal examples to convince you otherwise!

LADIES AND GENTLEMAN, FOR MY FIRST TRICK… let’s first set the stage a little. The event is the 2017 US Amateur Team East, one of the (if not the) largest events in the US. My opponent: FM Gregory Markzon, a local player who was about 200 points lower rated than me, but who I grew up playing against with… poor results. After some adventures in the middlegame involving both of us trying to force draws at different points for the match situation, eventually matters clarified to my team being down 2-1, forcing me to break out of a repetition into a losing position, and somehow surviving to the following almost symmetrical position with Black:

Q3vsQ3win.png

Obviously this position is very drawn, but actually things are not quite as simple as they appear. Black’s queen is slightly more centralized than its counterpart, and Black’s king is a single tempo more active than the enemy monarch. The latter is particularly important as it means many pawn endings would be winning for Black, allowing him to break out of potential perpetual checks by parrying with the queen. Thus, Black can at least hope for a situation where he slowly escorts his queen to the a-pawn, somehow corrals it while preserving the b-pawn in the process, then argues that the b-pawn is faster than the kingside connectors. In the game, White didn’t quite appreciate the danger of this idea until Black’s king had already managed to cross the halfway point, by which time some (mild) precision was actually required to survive. But psychologically speaking, going from a basically assured draw to suddenly having to find some accurate moves is often very tricky, and fortunately for us Markzon wasn’t up to the task. After well over 100 moves I managed to collect the match-tying point.

As a slight sidenote, if you take nothing else away from this article: **activate your king in queen endgames!!**. For some reason, basically all players under 2400 or so seem to be unwilling to activate their king and allow the possibility of checks, even when said checks are fairly harmless and can just be blocked. I understand the reluctance to risk blundering pawns to forks, but in general you really do need to play queen endgames like somewhat slower pawn endings (so the same principles apply: how far your pawns are is often more important than how many you have, and king activity is paramount).

Ok, this game wasn’t quite the same thing as Carlsen’s masterclass — there were some clear advantages in my position that made it reasonable to play on (even if the match situation didn’t mandate it). So let’s look at an even more blatant (and recent) example. Here, I am white against FM Arthur Guo, an 11-year-old 2300 who I fully expect to have a title by this time next year. The tournament is the Philadelphia Open, and this is late enough in the tournament that nothing but pride is on the line (we are both well out of money/norm contention. LADIES AND GENTLEMAN, FOR MY SECOND TRICK:

KatzGuo.png

This is about as symmetric a position as you can get, and if anything it can be argued that Black’s pawn structure is better off. But for reasons of personal pride (and annoyance at spoiling a close-to-winning middlegame), I badly wanted to find a way to at least try to press something. Of course, the only way to even pray for this is to engineer a way to get in h4-h5 under favorable circumstances, somehow compromise Black’s pawn structure, and then figure something else out from there, so this was necessarily my plan. In the meantime, I also need to find a way to get my queen onto a safe — but active — square, avoiding queen trades while simultaneously not allowing Black to come up with winning ideas of his own. This requires some dancing: 1. Qd4 Qb6 2. Qa4 Qb4 3. Qa7 Qe7 4. Qa8 Qc5 5. Qb7 Qe7 6. Qc6 Qc7 7. Qa4! (finally a “safe” square!!). This 7-move sequence allows White the time to get in a single tempo, so the “winning” plan becomes clear: use these 7 moves to get in g3, repeat the sequence to get in Kg2, repeat the sequence to get in h4, then finally repeat the sequence to get in h5. Oh, we also have to watch for Qc1-f4 ideas in the meantime. Phew.

In the game, this is more or less what happened, allowing me a glimmer of winning chances (though of course the position is still very drawn):

KatzGuo2.png

Here 1… Nxh5 is impossible due to 2. Qd7!. This forces 1… gxh5, which allows a cute trick: 2. Nd4 Qg5 3. Ne6!? fxe6 4. Qa7+ Nd7 5. Qxd7 when Black’s pawn structure is ruined and there is finally something to play for. Eventually, I was able to collect some of the pawns at the cost of a queen trade, leaving me with a still-drawn N+2 vs. N+1 ending. After some trickery involving a brilliant “bluffing face” (a subject for another day…), I managed to scam a win when Black, deceived by my fake confidence, erroneously resigned in the final position:

KatzGuo3.png

where 1… Nxf8 2. g7 Kf7! forces the draw immediately.

Of course I don’t mean to suggest playing on in every drawn position; sometimes, saving time and energy is the correct practical choice. But what I do suggest is “watching closely”: carefully considering when a position is truly dead vs. when it’s simply not easy to make progress. In particular, even just staying alert and focused throughout these endings where “anything” draws (often, not quite anything!) can earn you an extra half point here and there when your opponent goes to sleep (which is hardly rare). Even if this happens only one in 20 games, that’s an extra 3-4 rating points per event or so on average, which really adds up! Even full point swings are more than possible with such an approach.

AND NOW, LADIES AND GENTLEMEN… FOR MY FINAL TRICK, I’LL MAKE THE LAST WORD

Advertisements

Anatomy of a swindle (and how to avoid one)

It is often said that the first sentence of an article is the most difficult to write. I agree, so I’ll just skip it!

My name is IM Alexander Katz, probably best known for my uncanny ability to win games from massively worse positions. This is probably a quality earned by way too many hours of online bullet, but it’s a useful skill to have nonetheless; throughout my career, I’ve managed to win games with computer evalutions as low as -145. It seems logical, then, that I should write something on that subject.

Of course, it’s impossible to consistently execute these types of game-reversing swindles without help from the opponent, no matter how many practical traps and complications one manages to create. So while there is a certain art to constructing a masterful escape, there is a straightforward (but not simple!) way to avoid it: constant awareness.

The reasons for this are simple. As chess players, we are subject to certain attractions; to selection biases coming from two competing — but not unrelated — sources. The first is the pursuit of beauty; from the incredible feeling one gets when executing a beautiful sacrificial attack. Everyone loves to end a game in such a way, even to the point that some (the gambiteers among us) will structure their entire gameplan around the possibility of creating one. But it is dangerous to get attached to this desire for a simple reason: everyone remembers their beautiful sacrificial mating attacks, but nobody remembers all the times the opponent simply takes all the material and stifles a yawn.

Still, while we are attracted to the beauty of a stylish finish, we are also defined by our pursuit of simplicity: our desire to end games with as little risk and effort as possible. Sometimes this approach is justified, as there’s no reason to enter unnecessary complications when one has a safe edge. But all too often, it results in the exact opposite: by conceding something in exchange for apparently simplicity, suddenly the opponent may be in a much better position to create complicated counterplay.

Swindles are built off of these two principles: the chess player is both over- and under-confident at different points of the game. Determining which state the opponent is in — and how to best exploit it — is the hallmark of the swindler.


Let’s see some examples of these phenomena in action.

LarsonKatzUSCL

Larson – Katz, USCL 2014

Clearly Black has outplayed White from this position, and holds all the advantages one could hope for from this opening. The c5-square, normally a point of serious contention, is firmly under Black’s control. White’s pieces — especially his rooks — are staring into thin air while Black is preparing to start an assault on the White king. All in all, though the computer gives the cold-blooded 0.00 evaluation, practically speaking Black has excellent winning chances.

Unfortunately, I chose instead to lash out in the search of an aesthetic finish:

19… b5?? 20. bxc6 b4 21. cxd7 bxc3 22. Rxc3 Nxd7

and here I was forced to realize that my “combination” had resulted only a slightly worse position, where my b-file counterplay was barely sufficient to balance the pawn deficit. Shortly thereafter I overpressed and was quickly punished.

The cause for the loss here wasn’t a miscalculation, or time pressure, or any other of a number of excuses we tell ourselves to mask the true reason: the loss was due to a massive failure in judgement. One needs to retain objectivity in all positions — even the ones that are clearly in his or her favor — to maximize the score.


This last game an example of over-confidence; pressing the position too hard to force a “fun” finish. The opposite — playing too safe and allowing counterplay — is just as deadly:

KatzFranciscoUSCL.png

Katz – Francisco, USCL 2014

In this position I found a strong combination:

15. exd6 cxd6 16. Bxe7 Nxe7 17. d5! Bg4 18. Bb5

where White is simply up the exchange for no compensation.

KatzFranciscoUSCL2

Katz – Francisco, USCL 2014, position after 23… Bxb2

Just a few moves, White needs to decide whether to continue with 24. Rab1 or 24. Nxf5. Both options are still winning of course, but I was afraid of the complications that might result from 24. Rab1 Nxd6 25. Rxb2 Re8. Without thinking much more, I immediately played the “safe” option

24. Nxf5? Bxa1 25. Ne3 Bd7 26. Qxa7 Be5

when suddenly Black has some very annoying compensation for the pawns due to the two bishops aimed at the white king.

Had I taken the time and effort to actually analyze the alternative line, I probably would have found

24. Rab1 Nxd6 25. Rxb2 Re8 26. Nf6+! +-

which is, of course, a very simple win. Instead, as we both drifted into time pressure, Black’s counterplay became stronger and stronger until I wasn’t able to keep my grasp on the position to the point that I almost lost. However, in an amusing display of irony, my opponent also backed out into the “safe” option and forced a perpetual check when a win was available.


That about covers the pitfalls. Now let’s see how one can avoid them!

KatzKachieshviliNxd4

Katz – Kacheishvili, USCL 2016

In this position, Black — a strong grandmaster — has just sacrificed the exchange to combat the center. White can certainly take a “safe” approach, playing something like 16. Qh3 and trying to make use of the extra exchange later on, but this would give Black unnecessary counterplay against the e5-pawn. In fact, it turns out that Black would be no worse if White took this approach.

Instead, it is necessary to not be lazy and instead charge forward into the complications:

16. Qh5! g6 17. Bxg6 hxg6 18. Qxg6 Qg7 19. Qe8+ Qf8 20. Qg6+ Qg7 21. Qxg7+ Kxg7 22. Rc1 Nc6 23. O-O Ndxe5 24. Rc3!

Normally I would have included a diagram somewhere in the middle of that long line, but I choose not to this time because it is necessary to see the entire variation before entering into these complications. So much for lazy play!

KatzKachieshviliRc3.png

Katz – Kacheishvili, USCL 2016, position after 24. Rc3

The point is that Black’s pieces are simply too poorly placed despite the nominal material advantage, and Black’s king is in serious danger. White won shortly thereafter.


Now, an article about swindling wouldn’t be complete without a full-blown swindle, so let’s explore one now:

KatzGetzUSCL.png

Katz – Getz, USCL 2012, position after 51… Rf8

It’s not hard to see that White is completely busted: he’s down a pawn (and a key one at that), his bishop is horribly placed, and there is no real counterplay to speak of. In fact, the position is close to resignable. Well, close.

Instead of rolling over and awaiting Black to figure out a way to crash through, it’s important to get something of one’s own going. After all, things can’t get too much worse! This motivates White’s next move:

52. g4!?

Positionally suicidal of course, but practically necessary to even hope to achieve anything.

52… hxg4 53. hxg4 g5?

The first mistake, caused by lazy play. Black’s intention is clear: to prevent White from playing g5 himself. But had Black taken a bit more time here, he would have realized that g5 is not half as dangerous as the threats Black can make with a move like 53… Qb5! (threatening Rh8 and Ne2+, d3, etc.). Of course, the game move is winning as well, but now White has a plan he can put together:

54. Kg3!?

Another seemingly silly move, but necessary to get some counterplay going. The idea is to use the h-file in conjunction with the weaknesses Black has just created (on the light squares) to at least annoy the opponent.

54… Qb5 55. Rh1

KatzGetzUSCL2

Katz – Getz, USCL 2012, position after 55. Rh1

Here, Black panics in view of the upcoming Qf5, though it is simply not dangerous. A line such as 55… Rf7 56. Qf5 Ne2+ 57. Kf3 Nf4 illustrates this well: White is simply getting mated and cannot generate real threats. But it scared Black nonetheless, causing him to prevent it with

55… Ne2?! 56. Kf3 d3??

which finally allows White to complete the swindle:

57. Qa7+ Rf7 58. Rh7+!!

KatzGetzUSCL3.png

Katz – Getz, USCL 2012, position after 58. Rh7+!

and Black was forced to settle for the draw.


Of course, it is impossible to perfectly avoid and execute swindles, and it is generally a fruitless pursuit to try. But there are basic principles that are immediately applicable to one’s own play: setting maximum problems for the opponent, remaining alert at all stages of the game, and so on. These help even when not hopelessly behind!

In any event, it is important to be aware of one’s own selection biases, because it applies to many more cases than a simple style of play. It is easy to ignore the real reasons behind a loss, and thus never contend with them, while remembering all the flashy victories one accumulates over a career. This is a very dangerous approach to the game, and perhaps the most prominent reason for stagnation.

Anyway, it is often said that the last sentence of an article is the most difficult to write. I agree with this one too, so I’ll just skip it as well. Thanks for reading!