July 24, 2009 21:09

The Phoenix Attack: a rare line gives the classic Colle System a new lease on life

When most people hear of the “Colle System,” they think of the modest, solid, setup where White plays d4, Nf3, e3, Bd3, c3, Nbd2, and then looks up to see what Black has done with his moves. Guest author David Rudel proves that the system can actually be quite dangerous by introducing a new idea in the Colle.

This article is adapted from an excerpt of David Rudel's second book on the Colle System: The Moment of Zuke: Critical Positions and Pivotal Decisions for Colle System Players. The first book was published last year: Zuke ‘Em: The Colle-Zukertort Revolutionized.

When most people hear of the “Colle System,” they think of the modest, solid, setup where White plays d4, Nf3, e3, Bd3, c3, Nbd2, and then looks up to see what Black has done with his moves. This configuration is actually the “Colle-Koltanowski,” whereas playing b3 instead of c3 yields the “Colle-Zukertort,” considered a more ambitious opening. This latter option, the C-Z, scores quite well in practice, even at higher levels. It’s surprising how few people use it given that its mainline (1.d4 Nf6 2.Nf3 d5 3.e3 e6 4.Bd3 c5 5.b3) performs much better in practice (~58% for White) than many more common openings. This article shows an idea that should allow C-K players achieve similarly strong results.

C-Z setup C-K setup

The great Cecil Purdy (editor of Australia’s premier chess magazine for several decades and the 1st world champion of correspondence chess) suggested the C-K to beginners as a way of limiting the importance of opening preparation, allowing them to spend more time training their tactical abilities. It is a rugged system with, as Jeremy Silman puts it, “just a drop of poison.” Indeed, it begets more Bxh7+ sacrificial kingside assaults than any other opening, and there are plenty of opportunities for attacking chess when the center opens later.

If the C-K has all this going for it, why don’t more people play it?

Many players are put off by the Colle’s quiet initial moves, and it has engendered a sour-grapes reaction from players who find it annoying to play against. Several major writers, including Watson, Schiller, and Palliser, have remarked how Black has to work hard to achieve any meaningful play early on. If you have Black and are pitted against someone 100 points below you, the last thing you want to see is your opponent playing the Colle.

A Little History

Several world champions used the Colle System in the early 20th century, but the Colle-Koltanowski’s modest beginning has caused modern titled players to avoid it, playing sharper opening in the hope of gaining greater advantage. This means theory has grown very slowly. Indeed, hindsight has shown us that Black players had a rather wrong-headed understanding of the Colle for over half a century. Richard Palliser, in his excellent book on the opening, shows a good deal of astonishment at what used to the mainline: 1.d4 d5 2.Nf3 Nf6 3.e3 e6 4.Bd3 c5 5.c3 Nc6 6.Nbd2 Bd6 7.0-0 0-0 8.dxc5 Bxc5 9.e4 Qc7 10.Qe2 Bd6?!, which gives White a “healthy advantage” as long as he knows how to play against an IQP after 11.Re1 Ng4 12.h3 Nge5 13.Nxe5! Nxe5 14.exd5 exd5 15.Nf3.

White has a comfortable advantage in the old mainline.

Now that Black has figured things out a little better, I believe it is becoming clear that White players have also been approaching the opening with the wrong goals in mind (at least in the mainline). However, White can hardly be blamed for aiming at a quick, freeing e4 all those years given that Black was getting a poor position without any counter-play to speak of.

Problems in the current mainline

Let’s look at that push – 9.e4; it frees the bishop on c1 (kinda) and threatens to play e5 (kinda).

After a "freeing" 9.e4.

What do you mean by “kinda”?

Well, White “kinda” frees his Q-bishop because the e-pawn is no longer in its way, but the knight still is. And that knight is currently stuck defending the e4-pawn. Secondly, he only “kinda” threatens e5 because when Black responds …Ng4, he is actually adding two attackers to the battle for e5, not just one.

What are you talking about?

When the knight jumps to g4, he is clearly in a position to hit e5. A deeper point is that the knight on g4 allows the bishop on c5 to have a say in things because now White cannot add a defender to e5 via Re1 due to the threat of …Bxf2! Consider the diagram below (occurring after 9…Qc7 10.Qe2 Re8?! 11.e5 Ng4):

Re1 not an option for White.

So, White’s “freeing” move has really done nothing at all to free his position. In fact, if anything, it is even more restricted since the pawn on e4 blocks the prize bishop on d3. No wonder Colle himself enjoyed the simple 10.exd5?! once upon a time. Unfortunately, that move gives Black too much activity.

Furthermore, that bishop on c5 ends up foiling the obvious solution to this mess. White would like to just play h3 to put an end to the …Ng4 concern, but then the pin on the f2 pawn becomes an issue. Black can play a well-timed …Nh5 threatening to hop to the g3-square (quite possibly forking a rook on f1 and a queen on e2). Kh1 is no solution either since the other knight hop (…Nf4) can also be rather painful.

But can’t White take advantage of the knight’s vulnerability on g4? Doesn’t that just mean the “Colle Sacrifice” (Bxh7+) is no sacrifice at all?

Quite so. White does get Bxh7+ “for free.” For example, going back to that last diagram, White has 12.Bxh7+! Kxh7 13.Qxg4 Qxe5. He gets an attack on Black’s king without having to pay the bishop as a fee. And that is why prepared players began playing 10…h6!, a defense first analyzed by Reynolds and later popularized by Silman. Theoretically, Black gets equality here, but in practice he has been crushing White. Among games played by 2000+ players, Black has scored 64% and won outright over half! (It must be admitted, though, that often practical results are unfair to the Colle side since more often than not the Colle player is the lower-rated one.)

...h6 takes the wind out of White's sails.

From Bad to Worse

If this were the whole story, things might not be so bad. The Colle System, especially the Colle-Koltanowski, is often billed as a system where White sometimes settles for comfortable equality in a position they understand. Unfortunately, it turns out that 10…h6! (a good move to be sure) is not even needed. White’s attack is hampered by three things:

  • Black’s queen coming immediately to e5
  • The lack of an e5-pawn or rook on the e?file (generally a requirement for Bxh7+ attacks to work)
  • White’s queen first must take a move to recapture the knight (Qxg4) rather than going immediately to h5.

In particular, Black can get away with simply playing 10…b6!!, and White is already worse. In the obvious line, 11. e5 Ng4 12.Bxh7+ Kxh7 13.Ng5+ Kg8 14.Qxg5 Qxe5, Black’s defenses are adequate. Palliser quotes analysis by Bronznik that goes 15.Qh5 Qf5 16.Ndf3 Ba6! (only available due to 10…b6) 17.Rd1 (attempting to remove a bunch of squares from Black’s queen and then play …g4, forcing an exchange on g6 that leaves the e5-pawn vulnerable) 17…Be2 18.Re1 Bxf3 19.Qxf3 Qxf3 20.Nxf3, saying the position is “about even.” That might be so, but I think White is on the wrong side of “about even,“ and Black has better ways of responding to the threat of g4. For example, 17…Bd3 threatens Bc2 while allowing for the bishop to retake on g6 instead of the pawn. 17…Rae8 may be best since then Black would look forward to 18.g4?! Qg6 19.Qxg6 fxg6, when …Be2! is a real threat.

For this reason, Bronznik has suggested White consider 11.b3 instead, reasoning that after the natural 11…Bb7 12.Bb2, he is prepared for whatever Black may attempt. This would at least put White back in the “comfortable equality” category. The problem is that Black does not need to play nice. Instead of 11…Bb7, he could play 11…a5!

11...a5 stops b4 killing White's hope for central play.

Threatening to open the a-file with …a4?

Nah, that’s a threat, of course. But 11…a5 causes a more profound problem. It comes back to that blasted bishop on c5 again. In the lines that would naturally arise after the more tame 11…Bb7 12.Bb2, White depends on the disrupting b4?b5! pawn incursion to displace Black’s well-positioned Q-side pieces. This can be rather powerful if White does it after getting a pawn on e5, removing the d6?square from the bishop. So, not only does White have to consider a possible …a4, but Black’s a-pawn keeps him from executing a key b-pawn lunge and severely constricts his play in the coming middle game.

A New Hope — Enter the Phoenix

In retrospect, it should not be so surprising that White runs into a brick wall here. Knowing what we know now, one has to wonder whether 9.e4 + 10.Qe2 really makes much sense. After all, White has played a “freeing” pawn move that winds up restricting him. And then it turns out there is no advantage to playing e5 — indeed, a savvy opponent can punish him for doing so. Finally, Qe2 represents a loss of a tempo if White is banking on the Bxh7+ attack, for the queen can reach g4 or h5 just as easily from her original square as she can from e2.

I believe it is time for Colle Players to put 9.e4 on the burn pile, allowing a new Colle Attack to rise from its ashes.

What do you have in mind?

Well, I’ve hinted a bit at what I see as an under-estimated facet of Black’s setup. I believe players have not given due respect to the power of that bishop on c5. I’ve given some reasons so far, and there are more I haven’t mentioned, as to why that bishop is really a thorn in White’s side. The key is for White to put the question to this bishop before he plays e4 and commits his queen. In fact, in the attack I’m proposing White delays developing his queen, which can be rather well posted on her home square. Furthermore, if Black chooses to play his queen to c7, the ability for White to play f4 (supported by the pawn still on e3) can be pretty critical.

Let’s go back to the position after 8…Bxc5:

Instead of 9.e4, I’m suggesting White instead play…


White plans on turning the position into something closer to a reversed Meran. It is actually not too far off from the solution I’ve proposed elsewhere to fixing the Colle-Zukertort mainline. Rather than play solely for an e4-break, White will put his bishop on b2 and threaten both c4 and e4. Black can make it difficult to pull c4 off any time soon, and he can match White’s threat of e4 with his own e-pawn march. However, putting the two together favors White because the dissolution of the center makes the c4-break (when it finally comes) more deadly.

I've written a book devoted to this new attack, but I wanted to put it out there early for the Colle community to try it out a bit. To give some idea of the latent ferocity of this attack, let’s step through what has typically occurred in practice in the handful of games where this move has hitherto been tried:


I have not found a single game where a strong player put his bishop anywhere else. 9…Bb6 allows White to cramp Black’s position with 9.b5 while 9…Be7 looks passive (but should not be under-estimated).

10.Bb2 e5

This natural move was chosen by GM Sakaev, GM Karlsson, IM Sax, IM Raif, and the strong German Dirk Sebastian all within the last decade. It is more frequent than all other moves here combined.

11.e4 dxe4

11…Bg4, the other move seen in practice, is met well with 12.h3.

12.Nxe4 nxe4 13.Bxe4 f5

Black's king is vulnerable.

In general, White should respect the strength of Black’s e/f-pawn phalanx. However, in this particular situation, Black is kept too much on his heels to do much with it.

White has a couple promising possibilities after 14.Bd5+ Kh8:

  • Simply playing 15.c4 is good, presenting problems on both sides of Black’s board.
  • 15.Bc1!? is deeper than it looks. On the surface it threatens 16.Ng5, but the real point is that Black does not have a good way of stopping it. 15…h6 does not stop 16.Ng5 (16…hxg6?? 17.Qh5#) and after 15…f5, White may be quite happy pulling his bishop back with 16.Bb2!, figuring Black has now robbed himself of …e4, which is generally a major source of counterplay here. White can target the backward e-pawn easily.
  • But likely more formidable than either of those is 15.b5!, giving White a furious attack. Arndt-Ralf, 2004-2005 Bundeslinga continued 15…Ne7 16.c4 Ng6 17.h4! Nxh4?! (17…e4! 18.Ng5 Bf4 makes White work harder for his win) 18.Nxe5 Qg4?! (18…Bxe5 removes the dangerous knight and allows Black practical attacking chances after 19.Bxe5 f4! with Qg5 next) 19.f4 Bc5+ 20.Bd4 Bxd4 21.Qxd4 Qf6 22.Rad1, and material is equal but White is completely dominating.

It is worth pointing out that 15…e4!, likely Black’s best chance against this last option, is met by 16.bxc6 bxc6 17.Bxc6 Rb8 18.Ng5! Rxb2 19.Qh5, and it does not appear that Black can escape material loss.

Of course, there are plenty of other ideas and plans in this line, but I sincerely believe that 9.b4!!, which I’m dubbing The Phoenix Attack, is the battlefield upon which the C-K players must wage war in the coming decades. Let the ashes of 9.e4 fuel the next century of Colle games!

If you want to experiment with this new Colle idea, you should take a look at the Quick-Start guide I’ve placed online. A viewable database of all the Phoenix Attack games I know of (assembled with the help of Vincent Casasnovas) is available on the Colle System Players Portal. One of the games in that database is Colle’s own nail-biting victory over World Champion Max Euwe, an entertaining, double-edged shoot-out.

Editors's picture
Author: Editors


Felix's picture

The problem for the whole Colle thing is that your opponent rarely play like that. After 1.d4 d5 2.Nf3 Nf6 3.e3 c5! you have nothing. I know the idea 4.dxc e6 5.Qd4, but 5...Be7 and castle kills that line.

Arne Moll's picture

Not having the books either, I was wondering the same thing, Felix. Why play ...e6 so quickly when it seems much more flexible to keep the c8-bishop uncaged until White has shown his plans?

David Rudel's picture

I'm curious where you got the impression that 4.dxc5 e6 5.Qd4?! was considered a suggested refutation to this. Summerscale does not recommend this, Palliser does not recommend this, and neither do I.

In fact, I can only find one game in hugebase where White tried this (out of over 2000 games where 3...c5 was played).

In any event, White has antidotes to this line, including a rather dangerous reversed Abrahams and a solid transposition into a good Caro-Kann. Unfortunately, many of the older Colle books simply do not contain these refutations.

David Rudel's picture


In the abstract, you are correct. 3...c5 does give Black greater flexibility. However, it also gives White greater flexibility as well.

After 3...e6 [played twice as often as anything else, but there are certainly plenty of decent tries here for Black], White really only has two paths: the b3-Colle and c3-Colle. As a practical matter, a Black player can be reasonably confident of what he will face.

However, after 3...c5, White has several directions to take the game. In addition to the two I mentioned above [which are very different], White can also simply proceed with normal Colle Play, press for an IQP, or play into a Nimzo-Larsen Attack (Classical Variation).

As a practical matter, Black has to not only be pretty familiar with a wide range of theory but also be comfortable with a wide range of game styles to really be confident in 3...c5, whereas after 3...e6, he only really needs to be comfortable against both Colles.

It is also worth pointing out that Black often ends up playing ...e6 anyway, and in those games where he doesn't [playing his Bishop out to f5 or g4, he almost never is in a line where the extra move pair helps him.

So, it is true that 3...c5 puts White under a bit more pressure in committing to a kind of play, but Black very well may not like (or feel lost in) the play that ensues. An un-prepared Black player can get totally torn up in the reversed-Abraham's line, and the Caro-Kann reply is both solid and requires a completely different set of theoretical knowledge.

3...f5 allows White less flexibility in terms of how the game goes, but it turns out that the Slav setups that White gets there are actually pretty good!

Felix's picture

I read somewhere about this Qd4 idea with a note that this should be good for white. Anyway, the positions you get after other moves are not good either. Not critical at all. After c3 you get an exchange Caro Kann (oh, I feel lost ;) ), after c4 we can get quite a drawish position with cxd exd Nc6 Nc3 Bg4 and so on (I like that position :) ). I only see that Colle players don't like those positions.

You can always argue that your opponent may not know the line and such things, but on a good level you won't come far with this. I stopped playing the Colle some time ago, I didn't get it on the board frequently and when I had it on the board I got a drawish position (some games with b6 setup and such things).

Also I feel that for learning chess the Colle is a bad choice. You learn much more when playing main lines, you find GM games more interesting since you can look what they play in those positions and so on. Also you get more different types of positions.

Felix's picture

Btw., after c3 you can also play Qc7, Nc6, e5 as black with an advantage.
And after dxc (instead of c3) e6 c4 Bxc5 a3 you have dxc with a draw.

David Rudel's picture

Once again, who is claiming that you play 4.c3? This is simply not the sort of thing Colle books suggest. It certainly isn't the Caro-Kann setup I was referring to. I don't know of any modern, quality Colle book that suggests such a thing. It is the sort of thing an unprepared Colle player plays because he doesn't know better (or because he is just looking for a comfortable position...in which case he will still get what he wants).

While the 4.c4 line can be taken into drawish waters, those drawish waters certainly still favor White in practice and give Black little in the way of winning chances. For example, even if Black knows exactly what to play [5...Nc6 6.Nc3 Bg4 7.cxd5 Nxd5 8.Qb3 Bxf3 9.gxf3 e6 10.Qxb7 Nxd4 11.Bb5+ Nxb5 12.Qc6! Ke7 13.Qxb5 Qd7 14.Nxd5 Qxd5 15.Qxd5 exd5], practical play has shown Black having difficulty winning after either 16.Bf4 or 16.O-O. In practice, for example, that latter line has gone 9-34-2 between 2200+, suggesting it is easier for White to play. Those wins include many between strong players [2500+], including a rather severe upset between Reinderman and Khenkin and Ivanov's victory of Seirwan.

A prepared Colle player actually has a pleasant practical choice when confronted with 3...c5. If he is playing a weaker player or needs a win, he can use the reversed-Abraham. If he is playing someone higher or could live with a draw, he can go in for the 4.c4 lines and still have chances to outplay his opponent. This is especially true if his opponent is also playing for a win and chooses not to go for the drawish line you mention because an improvement has been found in the other critical line there, where Black plays ...e6.

Sure, there are lots of Colle System players who are a bit lazy in their preparation, and you'll be able to catch them more often than not by playing 3...c5 or some other play, but the Colle books put out in the last few years offer far better lines against the pet defenses than what you have evidently seen.

Indeed, many of the books put out earlier simply dodged some of these problem lines. Those books tended to try to show both kinds of Colle books in one volume. More recent books which focus on just one or the other [b3 or c3] have better coverage.

Tim's picture

The disparagement of the Colle-system / Player's

"I play the Colle. I like the Colle and will continue to play the Colle. Both variations."

A statement that sets one up for ridicule from many club level and above players. In the past the Colle player would walk into the local chess club and feel like the guy that left toilet paper stuck to back of his trousers. He received snickers from "the table", you know the players that can quote and raddle off the latest 30 move variation in the Sicilian or whatever latest line is most popular from a tournament somewhere that can't be pronounced.
Try going to your local chess forum and leave a post that you want to play the Colle. First, you'll be given advise not to waste your time and effort because it is all but busted and besides you will not be able to get into the lines you want because of early anti-colle deviations. They'll sat that Real chess players don't waste their time with such nonsense. Second, you may get quotes like "everyone knows that Bf5 or Bg4 are THE solution the Colle or at least doing something in the Kings or Queens Indian style. Black gets equality on move 4 or 5 at the latest."
I recently set up a poll on the Chess Publishing website (01/18/09) and the results are interesting 41% stated they never play the Colle and considered it a waste of time. Only 14% said they played it a majority of the time. A total of 134 votes since January.
Some of the comments:
- I do not play the Colle and probably never will do but I must say it has something. I hate to see my opponent playing it with white and my record against it is quite bad (35-40%). Maybe it is to quiet a type of play for me?!
- colle is essentially a drawing weapon. I usually only seen it trotted out when a draw will do. It is very hard for white to win ever. I have never lost to the Colle but also have few wins (5 wins 3 draws and no losses I think) all against higher rated.
- Colle is an excellent blitz opening, you can whip out the moves mindlessly and quickly without having to pay much attention to what Black is doing.
- I've often wondered if I, as a club player, shouldn't take up one of this mindless systems like the Colle or the Stonewall attack Just playing the same moves and following the same plans over and over again could probably make my life easier and score some simple textbook wins.

This just doesn't add up! In the last 2 years there have been 4 good books on Colle. (Both Colle-Zukertort and Colle-Koltanowski) and addressing the early deviations. The Zukem books by Rudel and the two by Palliser (Starting out The Colle and Starting out D-Pawn Attacks.) In my opinion there is allot more interest in the Colle than some players would have you believe .

I have the thought that us Colle players can hold our heads high, sniker back and know we have something for 'those' players at 'that table'.

John's picture

I must say I have never come across sniggering or aloof comments about the Colle at any OTB chess club, other than on internet discussions where they are pretty much invited. The Colle system has also been played at high level.

It is not a drawing opening for lower graded players either. I have beaten players graded higher and lower than myself using the colle.

The nice thing about the opening is that it is easy to learn for players without much time for study, yet as you progress there is more and more to the opening than meets the eye.

Where black is the higher graded player, there does seem to be a sense of "oh no, not the Colle" as they know they have to break it down and maybe take some risks to win. In any game or sport the underdog will be wise not to meet the favourite with their own preferred weapons.

Arne Moll's picture

One thing I've noticed myself is that Colle players often claim the Colle really isn't such a bad or boring opening at all - and then make quick draws with it anyway. (For instance, this happened at least once per team match when I played in the Belgian league.) I can imagine this leads to irritation (and bias) by players who tend to be open about their intentions. Why can't Colle players who like a quiet game just admit it, rather than hide their intentions by claiming they play it for 'attacking' reasons or whatever? Anyway, of course the setup is completely sound and can't be bad. It's more a psychological thing, I for one can't imagine Colle players aren't aware of this aspect of their choice of opening.

Lonefrog615's picture

Concerning the part of the article the reads:

(Starting with the bold text just after the diagram with the caption
"Re1 not an option for White.")

"But can’t White take advantage of the knight’s vulnerability on g4? Doesn’t that just mean the “Colle Sacrifice” (Bxh7+) is no sacrifice at all?

Quite so. White does get Bxh7+ “for free.” For example, going back to that last diagram, White has 12.Bxh7+! Kxh7 13.Qxg4 Qxe5. He gets an attack on Black’s king without having to pay the bishop as a fee. And that is why prepared players began playing 10…h6!, "

How does 13.Qxg4 ? Doesn't the knight on f3 prevent this?

Greg's picture

Hi my name is Greg and I am a big fan of the colle System.The best one out of all of them is the Colle-Zukertort it is by fair on my A-List.I have the book also by David Rudel called "Zuke Em" I read the Zukertort system it rules and its the only thing I play for white.Here is the opening of the Zukertort:

12.Rf3 to H3

There is the Zukertort for white!!!!!!!!!!!!

Last you can also play B-e2 as well but the R is by fair the way to go I have played this so much I now this system in my sleep. Try it and you will see.

Latest articles