What it Takes to Become a Grandmaster

Introduction

After the 2014 world championship match, former champ Garry Kasparov offered a striking insight into what he had seen.

Seventy percent of the moves could have been made by any competent player, he said.

Now, you can dispute the exact number. But Kasparov’s main point was valid: Many of the moves played by remarkable players are not very remarkable.

He had more to say. Twenty-five percent of the match’s moves could have been played by any grandmaster, he added.

Kasparov was leading up to the punch line:

The five percent that remained were world-champion quality moves, he said. Only players like the match contestants – Magnus Carlsen and Vishy Anand – could find them and be confident enough to play them on the board.

Many fans wondered which moves comprised that five percent. But in this book we’re concerned with another question that Kasparov’s insight inspires.

What characterizes the twenty-five percent? That is:

What are ‘grandmaster moves’?

What distinguishes them from the moves that average players can find? Are they the result of superior calculating ability? Are they the fruit of years of experience against very strong opponents? Do they stem from clicking through thousands of database games?

If these are the necessary factors, then a non-GM may have little chance of ever playing a grandmaster move. He will have trouble even understanding one. He can only play over a GM game, stand back and marvel.

But there’s another answer to the question:

Grandmaster moves are not beyond the understanding of average players. Many of those moves are based on principles, on the positional techniques we call priyomes or on different ways of thinking that are unfamiliar to non-GMs.

We see these move-motivators repeatedly in grandmaster games, like this:

Vitiugov – Bukavshin, Chita 2015

1 d4 d5 2 c4 e6 3 ♘c3 c6 4 e3 ♘f6 5 b3 ♗d6 6 ♗b2 0-0 7 ♗d3 e5 8 dxe5 ♗xe5 9 ♘f3 ♗g4

While these moves were being played, the world-class computers Houdini and Stockfish had already concluded that Black was better. They didn’t consider 10 ♕c2 to be among White’s four best options.

Black tried to punish White for taking liberties, with 10 ... ♗xf3 11 gxf3 d4.

Then 12 exd4 ♕xd4 would leave White with serious weaknesses on dark squares and weak pawns (13 0-0-0 ♕f4+ 14 ♔b1 ♖e8 and ... ♘a6).

Moreover, after 12 ♘e2 ♕a5+ he had to move his king (because of 13 ♕d2? ♕xd2+ 14 ♔xd2 dxe3+ 15 fxe3?? ♗xb2).

What Black – and the the super-engines – failed to appreciate at first was 13 b4! ♕xb4+ 14 ♔f1.

White has violated all sorts of basic principles in just 14 moves. But there were valid reasons for White’s play.

Yes, the doubled f-pawns are a weakness.

But control of a file – provided it’s a half-open file – is often much more important. There’s a hierarchy of positional advantages and disadvantages that grandmasters understand well and they justify 10 ♕c2!.

Yes, White failed to castle.

But early castling is often a wasted move, which can be delayed in favor of something that takes a much higher priority, as Mikhail Botvinnik liked to say.

And, yes, White’s king looks highly vulnerable.

But without a light-squared bishop, Black has little chance to attack it. A grandmaster’s acquired sense, called king feeling, tells him that it is Black’s king that should be in greater danger. A GM may not be able to verify this by calculating. He just feels it.

White got his pawn back with 14 ... c5 15 f4 ♗d6 16 exd4. The power of his rooks and bishops became evident after 16 ... ♘bd7 17 ♖g1 g6 18 ♖b1 ♕a5 19 f5!.

For example, 19 ... ♔h8 20 fxg6 fxg6 21 ♗xg6! (21 ... hxg6 22 ♕xg6 and mates or 21 ... ♖g8 22 ♗xh7! ♘xh7 23 dxc5+ ♗e5 24 ♗xe5+ ♘xe5 25 ♖xb7 ♘f6 26 ♕f5 and wins).

Black tried to neutralize the b1-h7 diagonal by shedding a pawn, 19 ... g5 (20 ♖xg5+ ♔h8).

But he was losing after 20 ♕c1! ♔h8 21 dxc5! in view of 21 ... ♗xc5? 22 ♕xg5 and mates.

The game went 21 ... ♗e7 22 ♗d4 ♕c7 23 ♗e4! ♖ab8 24 ♘c3 h6 25 ♘d5 ♕d8 26 h4 ♗xc5:

And 27 hxg5 ♗xd4 28 gxf6 resigns (28 ... ♘xf6 29 ♕xh6+ ♘h7 30 f6 and mates).

Did you notice any grandmaster moves?

Yes, 27 hxg5, many amateurs would say.

And they would be wrong.

Any competent player, to use Kasparov’s words, could have found the final combination. The star moves, the ones that made a game-changing difference, began much earlier, with 10 ♕c2! and 12 ♘e2!.

These kinds of moves are difficult to appreciate. But with the proper approach and study you can understand them. You can learn to apply the reasoning behind them – and enjoy the rare satisfaction of playing a grandmaster move in your own games. That’s what this book is about.

Mysterious

There are so many different kinds of moves that can be called grandmasterly. But what characterizes many, if not most, of them is they are counter-intuitive. Sometimes they are downright mysterious.

Smyslov – Lombardy, Monte Carlo 1969

1 c4 ♘f6 2 ♘c3 d6 3 d4 e5 4 dxe5 dxe5 5 ♕xd8+ ♔xd8 6 ♘f3

Black retreated his only developed piece, 6 ... ♘fd7!. It looks bizarre. But it is quite logical:

Black wants to play ... f6. That would solve the long-term problem of defending his e5- and f7-pawns. It would also prevent tactics based on ♘g5. Black could continue ... c6, ... ♔c7, ... a5, ... ♘a6 and ... ♘dc5 with smooth development, a safe king and no real weaknesses.

White’s reply, 7 g4!, is more of a head-turner. But once again, it’s a move whose strengths become clearer upon inspection:

White expects to see ... f6 and already is planning to exploit it with a timely g4-g5 and gxf6. If he follows up with ♖g1 and ♗h3 he may have a serious advantage because he would control key lines like the g-file and h3-c8 diagonal.

Black was one of the 100 top players in the world when this game was played. Yet he admitted he was stunned by 7 g4. After 29 minutes of thought he concluded that his position was if not actually lost then very bad. He eventually lost.

In truth, he was far from lost. But White would simply have a position any GM would be delighted to play after, say, 7 ... f6 8 b3 ♗b4 9 ♗b2 ♖e8 10 0-0-0.

Another GM game went 10 ... c6 11 g5! ♔c7 12 ♗h3 ♘a6 13 ♖hg1 ♗f8 14 ♗f5 h6 15 gxh6 gxh6 16 ♗xd7 ♗xd7 17 ♘e4 ♗e7 18 ♖g7 and White won.

What helped make 6 ... ♘fd7 and 7 g4 grandmasterly is that they were counter-intuitive.

Intuition is a refined sense of what the best move in a position is likely to be. It is what enables good, experienced players to play good, experience-based chess moves. If those players develop this sense further they might become masters. They will be able to make master moves.

But to go further, beyond master, a player needs to appreciate the moves that his intuition rejects out of hand:

He needs to consider developing a rook on a closed, rather than open, file. He must resist occupying an appealing – but actually worthless – outpost square. He should learn to make an I pass move or even a retreat. He must be willing to create a backward pawn that supports a strong center. He needs to learn when a bad bishop is really good and when an equal position isn’t.

In short, to get beyond master he needs to go beyond master moves.

Grandmaster Mystique

Before we go any further, let’s make something clear: This book will not make you a grandmaster.

There is much more to learn, many other skills to master than can be described in any one book. It takes years to achieve that level of insight, precision and understanding. Even if they invest years, very few students will reach the GM level.

Nevertheless, I believe that almost any serious student – or competent player – can understand grandmaster moves if they take the time. But first, they have to get past the feeling that the task is impossible.

Pachman – Fischer, Havana 1966

1 d4 ♘f6 2 ♘f3 c5 3 c4 cxd4 4 ♘xd4 e6 5 e3 ♘c6 6 ♗e2 d5 7 ♘c3 ♗c5 8 0-0 ♗xd4!? 9 exd4 dxc4 10 ♗e3 ♘a5 11 ♗xc4 ♘xc4 12 ♕a4+ ♗d7 13 ♕xc4 ♗c6 14 ♗g5 ♕a5!? 15 ♕c5 ♕xc5 16 dxc5

Black’s eighth, ninth and 15th moves looked odd. They brought about an endgame with bishops of opposite color. Those bishops are notorious for making endgames hard to win. Yet Bobby Fischer was not playing for a draw. Far from it.

But now the game began to look bizarre. Black chose 16 ... a5. After 17 ♖fd1 he switched to the other wing with 17 ... h5. That prompted 18 h4. What’s going on?

Moves like these help create a grandmaster mystique that leaves an aspiring player muttering, I’ll never be able to understand this. No matter how much I study.

The mysterious RP move is actually a common feature of grandmaster games. And it isn’t that mysterious. Black played 16 ... a5! to stop a queenside initiative based on b2-b4-b5. His move also artificially isolated the c5-pawn. That pawn is a potential target because it can’t be easily defended by another White pawn.

Then 17 ♖fd1 h5! threatened to gain space for his KR with ... h4 (and possibly for his bishop with ... h3). White stopped that with 18 h4.

When you look at it that way, these moves weren’t so bizarre after all. The next stage of the game was easier to grasp: 18 ... ♘d7 19 ♗e3 ♘e5 20 ♗d4 ♘d7 21 b3 ♖g8 22 ♗e3 (otherwise 22 ... g5) ♘e5 23 f3 ♘g6 24 ♗f2 ♘f4 25 ♗e3 ♘d5.

After 26 ♘xd5 ♗xd5 27 ♖d4 ♔d7! 28 ♖c1 ♔c6 29 ♖c3 f6

... the possibility of ... g7-g5-g4 prompted 30 f4?!. Black had the edge after 30 ... ♖gd8 31 ♔f2 a4! 32 ♖xa4 ♖xa4 33 bxa4 ♗xa2 and eventually won.

In the following pages I’ve compiled 50 attributes of grandmaster thinking. They help GMs find grandmaster moves. These are not the only ones worth knowing. Becoming a grandmaster isn’t that easy. But these are a good start.

There is no prioritizing. The uber-luft (1) is not necessarily more important than, for example, the endgame anchor (11) or the technique of tacking (32) or the wing chain (41). This was just a convenient way to organize this book. Each chapter will end with a series of quiz positions based on the move-motivators discussed in it. Don’t expect to solve all of them, or even most of them. After all, even grandmasters fail to find some grandmaster moves.

Chapter One

When a grandmaster tries to explain a strange-looking move that he played, his annotation may consist solely of long varations of move analysis.

That analysis may be technically accurate. It will likely be based on what a computer found. But it may be useless to the vast majority of aspiring students.

Let’s try to do better, by explaining grandmaster moves in words. We will start with one of the recurring mystery moves that never fails to surprise fans.

1. Uber-Luft

Experienced players know what luft is. Many of us learned the hard way, as beginners:

An enemy rook or queen landed on our undefended first rank. Our castled king had no escape. Checkmate.

The traditional way to avoid this fate is to create luft (German for air). You push your g-pawn or h-pawn one square to make a flight square at g2 or h2. In desperate times, f2-f3 or f2-f4 will do.

Grandmasters appreciate the benefits of advancing the pawn two squares. This is uber-luft.

White would love to take on c5. But 1 ♕xc5? runs into 1 ... ♕xe4!. White is betrayed by his vulnerable first rank (2 ♖xe4? ♖d1+).

The other capture, 1 ♘xc5, allows 1 ... ♗xe4!, since 2 ♘xe4 ♕xe4! 3 ♖xe4? ♖d1+ is another mate.

It shouldn’t take long to realize that the lack of luft is crucial in this position. But 1 h3? again allows 1 ... ♗xe4! – while 1 g3?? ♖d1! is disastrous.

The position calls for 1 h4!. It creates a flight square at h2 and – what’s equally important – it does it with tempo. The g5-bishop is attacked and if it moves, White has time to capture strongly on c5.

Relatively best was 1 ... ♗xe4 but thanks to 1 h4 White can play 2 hxg5 ♗xf5. Then 3 ♖be2 would give White the better winning chances.

Black preferred 1 ... gxh3. But White’s first rank was no longer toxic. He could play 2 ♘xc5!.

Suddenly the tactics – which were running in Black’s favor in the previous diagram – are smiling on White. For instance, 2 ... ♗xe4? 3 ♘xe4 ♕xe4 doesn’t work because of 4 ♖xe4 ♖d1+ 5 ♔h2.

Neither does 2 ... ♗c8? in view of 3 ♘e3! and 3 ... ♗xe3 4 ♕xd8+, and White won. (Better was 2 ... ♗c6 but after 3 gxh3 and ♕b4! White is still on top.)

Uber-luft is the rare case of a move that is both safety-minded and super-aggressive. It need not be forcing, as 1 h4! was in that example. But it should do more for you than a mere one-square pawn advance.

Lastin – Loginov, Krasnodar 2002

1 e4 c5 2 ♘f3 d6 3 d4 cxd4 4 ♘xd4 ♘f6 5 ♘c3 a6 6 ♗g5 e6 7 f4 ♕b6 8 ♕d3 ♘c6 9 0-0-0 ♘g4 10 ♕g3 ♘xd4 11 ♕xg4 ♗d7 12 f5 ♖c8 13 ♕g3 ♘c6 14 ♗c4 ♘e5 15 ♗b3 ♕c7 16 ♖hf1 b5 17 fxe6 ♗xe6 18 ♘d5 ♗xd5 19 ♖xd5 f6 20 ♗f4 ♘c4 21 ♔b1 g6 22 ♖fd1 ♗e7

White might look at solid candidate moves such as 23 ♕h3 and 23 ♖5d4 (so that 23 ... 0-0? 24 ♗xd6). But his position is promising enough for him to be more ambitious.

The sacrifice 23 ♗xd6 ♗xd6 24 ♖xd6 ♘xd6 25 ♖xd6 seems to fit the bill. But after 25 ... ♕c5! White has to take time to meet the mating threat of 26 ... ♕g1+. He may be winning after 26 a3 ♖f8. But it isn’t clear.

When you look at variations like that, they may give you hints about the best move. These hints point to 23 a4!.

White threatens to win the b-pawn with 24 axb5. But 23 a4 also makes ♗xd6 much more dangerous because White has luft at a2.

If Black protects the b5-pawn with 23 ... ♕c6, White can pull the trigger on 24 ♗xd6! ♗xd6 25 ♖xd6 ♘xd6 26 ♖xd6.

The absence of a last-rank mate means that 26 ... ♕c5 is harmless. White could play 26 ♖xf6, followed by ♖e6+ and/or ♖xa6 followed by the decisive entrance of his queen.

Instead, Black managed to get his king to a degree of safety with 23 ... bxa4 24 ♗xa4+ ♔f7. But after 25 ♕h3 h5 26 ♗b3 ♔g7, White had a choice of promising lines, including 27 ♕e6 ♖he8 28 ♖5d4.

Masters often make luft when they have nothing more important to do. Timing isn’t critical in such a position. But uber-luft should be prompted by a sense of urgency:

Black has no time for 1 ... ♕f3? because of 2 ♖xd4 ♖e2?? 3 ♖d8+.

He saw the right idea – but one move too late, 1 ... axb4? 2 ♕xb4 h5.

White defended with 2 ♖xd4 ♕f3 3 ♖e4!. The tactics that might have helped Black had evaporated. He lost the queen ending, 3 ... ♖d8 4 ♖e3 ♕d1+ 5 ♔g2 ♖xd3 6 ♖xd3 ♕xd3 7 h4.

Yet Black could have saved the game with 1 ... h5!. Then ... ♕f3 is more of a threat because he has no last-rank problem.

There’s more. With 1 ... h5! Black prepares ... h4-h3! with an eye to ... ♕g2 mate. For example, 2 b5 h4! 3 ♕xa5? h3!.

White can neutralize the kingside with 2 h4.

But then 2 ... ♕f3! 3 ♖xd4 ♖e2 leads to another safe rook ending (4 ♖f4 ♖xd2 5 ♖xf3 axb4).

The main alternative is 2 b5 h4! 3 gxh4!. But 3 ... ♖e6 and 4 ... ♖g6+ prompts 4 ♖c6 ♖xc6 5 bxc6 ♕xc6, with a drawish queen ending (6 ♕xa5 ♕g6+ and ... ♕xd3).

The h-pawn is the most likely candidate for uber-luft, particularly in the endgame. We’ll examine cases of that later, in (26). Advancing the g-pawn two squares can be a bit more sophisticated:

If White could trade one pair of rooks, her b-pawn offers real winning chances. But with four rooks on the board Black is not worse. She has several options but 1 ... g5! is a grandmasterly claim to counterplay.

It makes the h2-pawn a target for a rook raid such as ... ♖a1+-h1. It also stops White from solidifying her pawn structure with f2-f4.

Black had enough play after 2 ♖d7 ♖a1+ 3 ♔c2 ♖c8+ 4 ♔d2 ♖cc1 5 ♖b7 ♖d1+ 6 ♔c2 ♖ac1+ 7 ♔b3 ♖d2.

Now 8 ♖f3 is met by 8 ... ♖cc2! (9 ♖bxf7 ♖xb2+ or 9 ♔a3 ♖xf2). A draw was agreed after 8 f4 gxf4 9 gxf4 ♖xh2 10 f5 exf5 11 ♖ee7 ♖h6 12 ♖xf7.

Of all the GM mystery moves, uber-luft is the one that amateurs should have the easiest time applying to their own games:

Whenever you are about to to push your h-pawn or g-pawn one square to create luft, consider advancing two squares. In the vast majority of cases, one square will be better. But in that minority of positions, the benefits of two squares are considerable.

Berliner – Fischer, Bay City 1963

1 d4 ♘f6 2 c4 e6 3 ♘c3 d5 4 cxd5 ♘xd5 5 e4 ♘xc3 6 bxc3 c5 7 ♘f3 cxd4 8 cxd4 ♗b4+ 9 ♗d2 ♗xd2+ 10 ♕xd2 0-0 11 ♗d3 b6 12 0-0 ♗b7 13 ♖fd1 ♘c6 14 ♕b2 ♕f6! 15 ♖ac1 ♖fd8 16 ♗b5 ♖ac8 17 ♘e5?! ♘xe5 18 dxe5 ♕f4 19 ♖xc8 ♖xc8

Black threatens the e4-pawn and might win the e5-pawn with ... ♖c5.

A natural defense is 20 f3 because of Black’s lack of luft (20 ... ♖c5?? 21 ♖d8 mate).

Black’s advantage would disappear after 20 f3 h6 21 ♗e2. For example, 21 ... ♖c5 22 g3 ♕xe5? is bad (23 ♕xe5 ♖xe5 24 ♖d8+ ♔h7 25 ♖d7).

However, Black does much better with 20 ... h5!. His idea is ... h4, to discourage g2-g3 and to soften up the White kingside with ... h3. After 21 ♗e2 h4 Black would retain good winning chances.

White chose 20 ♕d4, which threatens both the mating 21 ♕d8+ and the win of material with 21 ♕d7.

This time the quiet 20 ... h6 works well because 21 ♕d7? allows 21 ... ♖c1 or 21 ... ♖c2. White would have a difficult game after 21 f3 ♖c5.

But 20 ... g5! was stronger. Black prepared 21 ... ♔g7 and ... ♖c5xe5.

The difference between 20 ... g5! and routine luft, 21 ... g6?, was revealed after 21 f3?! g4!. Then 22 fxg4 ♕xg4 was bound to cost White a pawn. The same is true of 22 ♗e2 gxf3 23 ♗xf3 ♔g7.

White prefered 22 ♗e2 gxf3 23 gxf3 and now 23 ... ♖c2 was better than the game’s 23 ... ♔h8 24 ♔h1 ♗a6! 25 ♕f2? ♗xe2 26 ♕xe2 ♕xe5 because White could have defended better with 25 ♕d2!.

To repeat, a one-square advance is usually the better move when you need luft. It’s one of the seventy percent of moves that any competent player would find. Uber-luft is the grandmasterly exception.

2. Bad Pawns Support Good Centers

Young players can be terrified by scary stories about bad bishops. But as they gain experience they learn that often a bad piece can do very good things.

Masters know Mihai Suba’s sage observation, Bad bishops defend good pawns. We’ll see another version of the benefits – bad bishops build blockades, in (48).

Grandmasters carry this logic one step further: Bad pawns can support a good center.

Torre – van der Wiel, Thessaloniki 1988

1 e4 c5 2 ♘f3 ♘c6 3 ♗b5 g6 4 0-0 ♗g7 5 c3 ♘f6 6 ♖e1 0-0 7 d4 cxd4 8 cxd4 d5 9 e5 ♘e4 10 ♘c3! ♘xc3 11 bxc3 ♕a5 12 a4

White’s c3-pawn looks like the poster child of positional weaknesses. It is backward and lies on a half-open file. That makes it a prime target for a Black rook at c8. If Black can trade queens, that pawn alone may give him excellent winning chances.

But while queens are on the board, that c-pawn is a hero. It is the bedrock of the d4/e5 chain. That chain gives White a kingside space edge and attacking chances.

And it is not easy to win that pawn, e.g. 12 ... ♕xc3? lands the queen in trouble after 13 ♗d2 ♕b2 14 ♖b1 ♕a2 15 ♖e2.

Black completed development 12 ... ♗g4 13 ♖e3! ♖fc8. Then came 14 h3 ♗xf3 15 ♖xf3.

The c-pawn remains secure (15 ... ♘d8 16 ♗a3 ♖xc3 invites 17 ♕d2 or 17 ♕e1 with advantage).

Black realized he needed to get rid of that pawn and that it was safer to trade it off than try to capture it. After 15 ... e6 16 h4! a6 17 ♗f1 b5 18 h5 b4 19 cxb4 ♕xb4 he had exposed the d4-pawn.

But White had made enough progress on the kingside that after 20 ♗e3 his attack had become unstoppable, 20 ... ♘a5 21 ♖h3 ♘c4 22 hxg6 fxg6 23 ♕g4 ♕e7 24 ♗g5 ♕f7 25 ♕h4 ♗h8 26 ♗d3!, e.g. 26 ... ♖ab8 27 ♗xg6! ♕xg6 28 ♖g3.

If Black had won the d4-pawn or invaded with heavy pieces along the c-file, a different result was highly likely. But the c3-pawn denied him.

A similar situation arises in the Winawer French (1 e4 e6 2 d4 d5 3 ♘c3 ♗b4 4 e5 c5 5 a3 ♗xc3+ 6 bxc3). Black may open his side of the c-file with ... cxd4. The recapture cxd4 undoubles White’s pawn but exposes the c2-pawn to attack.

Then c2-c3 by White would create another backward pawn on a half-open c-file. It would also make White’s dark-square bishop a bit worse. But experience has found that supporting the d4-pawn is more important.

What about other ugly center pawns? If Black plays ... e5 in the open variations of the Sicilian Defense, his d-pawn becomes backward on a half-open file. That was once considered a near-fatal weakness.

But that was before openings like the Boleslavsky Variation (1 e4 c5 2 ♘f3 d6 3 d4 cxd4 4 ♘xd4 ♘f6 5 ♘c3 ♘c6 6 ♗e2 e5!?), the Najdorf Variation (1 e4 c5 2 ♘f3 d6 3 d4 cxd4 4 ♘xd4 ♘f6 5 ♘c3 a6 and 6 ... e5) and the Sveshnikov Variation (1 e4 c5 2 ♘f3 ♘c6 3 d4 cxd4 4 ♘xd4 ♘f6 5 ♘c3 e5) proved their value. That strong pawn on e5 gets key support from its d6-brother.

Another common example of this occurs when Black accepts a backward e6-pawn in the French Defense.

Wieczorek – Fedoseev, Warsaw 2012

1 e4 e6 2 d4 d5 3 ♘d2 ♘f6 4 e5 ♘fd7 5 ♗d3 c5 6