Friday, January 27, 2012

Chess960: Team Chess and Seconds?

In the latest Wilk aan Zee tournament 2012, we saw Aronion play the masterpiece exchange sacrifice (incidently Levon is a superb Chess960 player and supporter). In the video commentary he notes that his team of analyzers and seconds helped him to find this exchange sacrifice in home preparation.

That got me thinking. Having teams of seconds helping the super grandmasters in world competition is good isn't it? After all it turns Chess into a bit of a team sport. Great!

What about Chess960? If you don't know what the start position is going to be, then there is no point having teams of analyzers working for you in the background is there?

IS there?......

Chess960 is actually better for team chess! Imagine this scenario. In the future when all the old traditional chess players have moved on and Chess960 is being played more regularly, FIDE will create tournaments where the next day's Chess960 position is revealed at the very end of the current day's play (when the last game has concluded).

What will that mean for team Chess960? Basically the top GM's will have one night to prepare their position for the next day's play. It will be the same for all players because the next day's start is revealed to all players and media at the same time. That in itself would be a bit of a spectacle and make the tournament more interesting (note the King's vs Queens tournament roulette wheel idea) and everyone will wait until the last game has been played to find out the next day's start position.

It actually benefits GM's to have seconds and teams of analyzers working for them all through the night looking for the best lines in the next day's start position. The GM simply has dinner after the current day's play, get's involved with the team over the next day's position for an hour or two after dinner then goes to bed and has a restful nights sleep. Next morning he or she gets up, has breakfast and the team explain to him or her the very best opening lines that they have found overnight. The GM commits them to memory and off we go for the day's play!

The media also get the Chess960 position a day before the next day's scheduled play, and so they can also make informed commentary for the next day's play. We get out of this terrible rut of today's play where we play the same traditional start over and over again. The chess will be creative and still have all of the home preparation players want. Even the spectators can go home and think about the next day's start position to make it more enjoyable for them too. There is plenty of scope for chess seconds and team efforts and the GM's still have to have excellent memories because they alone will have to face the Chess960 start.

These kind of Chess960 super tournaments would be a bit like Formula One Racing. The Chess960 start position is for all effective purposes the car that the chess player will drive. The team of analyzers and seconds are his pit crew who work tirelessly through the night making sure that for the next day's start, his or her car is in pristine condition.

Heck, why not call it "Formula-1 Chess"!
Enjoy 960

Chess960: The Great King's Pawns Opening Research Project

Played an SP141 Chess960 game the other day. So often when playing quickly as white, I'll play 1.e4 as a knee jerk reaction! After all it is the famous King's Pawn Opening in traditional chess isn't it? The problem is that in Chess960, 1.e4 can have a bit of a nasty sting in tail for white! Here is an example:

SP141 White to play: 1. e4

The classic scenario simply goes like this:
  1. e4!?  ... f5!? (White screams doohhh! because of Bxa2)
  2. Nb3!? ... fxe4!?
  3. f3!?  ... Nb6!?
I cannot tell you how many times I have played e4 to find this scenario or similar cropping up! So here are some stats for the scenario and how often to expect it. Notice that the theme crops up in batches of three consecutive SP's in a row? I've included quick computer checks of exactly the same King's Pawn Opening opening sequence from above, noting their approximate severity:

| 141 | NRQBNKBR ok
| 142 | NRQNKBBR good
| 143 | NRQNKRBB dubious

| 157 | NRNBQKBR good
| 158 | NRNQKBBR good
| 159 | NRNQKRBB good

| 173 | NRNBKQBR good
| 174 | NRNKQBBR good
| 175 | NRNKQRBB good

| 189 | NRNBKRBQ dubious
| 190 | NRNKRBBQ dubious
| 191 | NRNKRQBB dubious

| 237 | NRQBKNBR ok
| 238 | NRQKNBBR good
| 239 | NRQKNRBB dubious

| 253 | NRKBQNBR good because of Qe1
| 254 | NRKQNBBR good
| 255 | NRKQNRBB dubious

| 269 | NRKBNQBR dubious because of f1
| 270 | NRKNQBBR good?
| 271 | NRKNQRBB dubious

| 285 | NRKBNRBQ dubious
| 286 | NRKNRBBQ good because fxe4
| 287 | NRKNRQBB dubious

| 333 | NRQBKRBN dubious
| 334 | NRQKRBBN good
| 335 | NRQKRNBB good?

| 349 | NRKBQRBN good?
| 350 | NRKQRBBN good?
| 351 | NRKQRNBB good?

| 365 | NRKBRQBN dubious
| 366 | NRKRQBBN good?
| 367 | NRKRQNBB good?

| 381 | NRKBRNBQ good?
| 382 | NRKRNBBQ ok because e4 blocks Qh1
| 383 | NRKRNQBB dubious

Basically the underlying pattern is this:
  1. If a rook is on f1 it is dubious for white because of Rf8xRf1
  2. If a queen is on f1 same as above
  3. If a queen or rook is on e1 then usually good
  4. If a knight is protecting a7, then it is dubious for white
  5. If a knight is on f1 then good
I actually quite like the Chess960 numbering scheme because of the way that themes like this play out in consecutive batches of SP numbers. Makes for interesting studies by consecutive SP numbers.

Chess960 insiders tip!:
If you are playing white and you knee jerk 1.e4 only to realize that black can counter attack on the edge pawn and your rook on b1, don't panic! Just continue playing the opening calmly, because out of all Chess960 positions, the situation crops up only 4% of the time, and when it does, 1.e4 is only dubious roughly 33% of the time and it is never fatal!

Why do people still insist on playing traditional chess?? Analyzing the patterns in Chess960 is a lot of fun.

Enjoy 960!

Thursday, January 26, 2012

Chess960: Knights out early

Got into another drama with the Chivalry Knights the other day. In the diagram below, see if you can predict where black can stuff up here because of white's early developed Chivalry Knights. If you don't know what they are check out Naming The Knight Pairs.

Basically, the problem is that I neglected the reality that not only can the Chivalry Knights quickly move forward and attack a weak seventh rank square if the opponent is not wary, they can also quickly move forward and switch to attack a weak sixth rank square! Notice that in the diagram below, the knights actually have powerful forks if they can get to the highlighted red and yellow squares and black has fallen asleep:

SP603 black to move: be a bit careful to avoid a trap on the 6th rank

I know a few people have said that the chivalry knights are a problem because they can't easily move forward being on the same colour squares and thus blocked more efficiently by a single pawn. But boy can they come to life when the opponent is not watching out for them.

Enjoy 960

Sunday, January 22, 2012

Chess960: Queen moves Qf1-Qh3 or Qc1-Qa3

Was reading Mark Week's blog An Early Queen Sortie and got a bit interested in how often this theme actually happens in Chess960:

White's option of Qh3:

The stats on it's regularity are actually pretty easy to work out:
xxxxxQNN = 06
xxxxxQRN = 18
xxxxxQRB = 18
xxxxxQNB = 12

Therefore total probability = 54*2/960*100 ~ 11% ~ 10%!

But how often is Qh3 actually worth it, out of that total approximate possibility of 10% for Qh3 showing up on the board? Here is a subdivision of the above 10% of possibilities for Qh3, with a pin of the king on c8:

xxKxxQNN = 2
xxKxxQRN = 4
xxKxxQRB = 3
xxKxxQNB = 4

Therefore total probability = 13*2/960*100 ~ 3%!

The basic problem with Qh3 is not just that it is strategically dubious, but that it only ever shows up in one in ten games, and when it does, it probably is not a good idea for these reasons:

  • To play Qf1-Qh3, g3/g4 must be played, but most likely, the only time that g3 is worth it is when a bishop is in the corner, but that only ever shows up in practice in roughly 6% of games. Even then, out of that 6%, there are probably no occasions where it is worth playing (see below).
  • The queen can threaten to capture the edge pawn on h7, but that seriously misplaces the queen. Yes it is true that if the h7 pawn falls, black cannot castle g-side into safety and therefore black's strategy is disturbed, but the black king will be no closer to the area than e8 (otherwise h7 cannot be undefended), and therefore black will probably castle c-side into safety while white has wasted a lot of effort on the h7 pawn.
  • The queen can threaten a pin on a king, but that is almost always parried. The only justification for Qh3 is if by pinning the king, black's development is disturbed. The question is though, will black be disturbed more than white sidelining the queen on h3?
  • The queen can move from Qh3 across the third rank into a better position, but that costs a third tempo and usually the queens lateral movement is impeded because other pieces are sitting on the the third rank to block her.
  • If a rook is on g1, then a pawn move g4 is a possibility, but again this only ever happens in roughly 7% of all Chess960 games. That said, pawn to g4 could well be good, but when would Qh3 ever be good to follow it with?

Chess960 insiders tip: If you want to play the opening quickly, you can almost totally disregard the move Qf1-Qh3 if it ever shows up as a possibility. This probably also applies to it's partner Qc1-Qa3.

Enjoy 960

Tuesday, January 17, 2012

Chess960: Develop the Centre Bishop with Bf3?

With Chess960, opening themes are learned according to their probability of showing up over the board across a number of start positions. These themes can be small chunks or over arching piece placement themes. Here is an interesting theme that goes part of the way into answering Irina Krush views on the Center Bishops that Mark Weeks has discussed over at his blog.

Should I play Bd1-f3 to fianchetto the bishop?

The problem with Bf3 in these situations, is that a knight on f8/h8 can annoy the bishop on f3 by moving forward to e5/h4, threatening to exchange the bishop and isolate a rook pawn. To do this black only has to expend one extra tempo (because the tempo taken up from initially developing the knight is good). On the other hand, white has to expend one tempo moving the bishop to f3, then possibly another tempo moving the bishop away from f3 to avoid the exchange, and possibly another tempo to allow Bf3-g2 if white wants to keep the bishop on the long diagonal (but a bishop on g2 is still being attacked by the same knight on h4!)

So instead of answering the question with yes or no, here are the stats for how often the question arises:

Bd1 = 240 SPs (with mirror 480 SP's)
Bd1+Nf8 and/or Nh8 = 176 (with mirror 352 SP's)

If you punch the maths, the question Bd1-f3? and it's mirror Be1-c3? arises in roughly 33% (1 in 3) of all games! 

Even more interesting is that since 50% of all Chess960 games will have a bishop on d1 or e1, and if there are 2x176 games with opposing knights, then out of all the Bd1/e1 games, the question Bf3/c3? occurs in 75% of those starting positions!

So what about the question "Should I play Bd1-f3 to fianchetto my bishop?"
  1. It is unlikely that NxBf3 is good for the bishop player
  2. It is unlikely that a bishop on f3 can find a stable outpost on the long diagonal by moving into the center
  3. It is unlikely that playing Bf3/g3/Bg2 can work because the enemy knight arrives on h4 to attack Bf3, then biffed by g3 and white cannot retreat the bishop to g2 because of NxB.
  4. It is unlikely that if there is a queen in the corner, that the Bf3/Qh1 combination is a good idea because it blocks the queen and only rarely do the two pieces working together present a crushing tactic on the long diagonal that is worth keeping the queen locked away on h1.
  5. It is unlikely that a bishop on f3 will find a compensating tactic for the fact that it probably will have to move again for a variety of reasons.
  6. It is likely that Nh4 attacking a bishop on f3 is a useful move not just that it results in an exchange NxB, but that Nh4 attacks g2. Black ...Nh4 is really a big pain in the ass for white in these situations.
  7. A bishop on f3 blocks the f2 pawn.
  8. Only 25% of bishops in the center do not have to deal with the opposition of the knight question
  9. It is unlikely that Bf3/g4/Bg2 is a viable opening idea because it has cost two tempi to implement and the pawn on g4 is going to need serious attack options to compensate for the broken kingside pawn structure and that the g4 pawn could well be undefended and not easy to defend without cost.
  10. The hope with Bf3 (or Bc3) is to build a start setup where the NxBf3/c3 exchange is actually favourable for white. But when would that be???
Therefore, with either one bishop in the center or two, by far the most likely development squares for the bishops will either be on the second or forth ranks with Bd1-e2 or Bd1-g4.

And so here is the big insiders information tip for Chess960 fans:

Since 50% of all Chess960 games have one or two bishop in the center, and since developing a bishop to the third rank is probably not good, save time thinking about the bishop with the plan to develop it to either the second or forth ranks. In the case of a bishop on d1, this means Bd1-e2 or Bd1-g4. If either of those options are not pleasing, give a lot of thought to Bd1-c2.

The other question is whether to develop the bishop in the other direction. For example Bd1 to c2 or b3. That is reserved for another topic except to say that a bishop on c2 can be very powerful if it is backed up by a queen that is on the back rank! Bc2 only takes two tempo and the c2-c4 pawn is probably good to move as well as per the English opening in traditional chess to counter in the center. Choosing instead Bb3 is usually ineffective (I think), because a bishop on b3 often blocks a knight or a queen and it also blocks the b-pawn. It only rarely get's involved in favourable tactics when it sits on b3 as well. That said, Bd1-b3 is worth considering alongside Bd1-e2 and Bd1-g4, because Bb3 does directly attack a castled king.

So from most likely to least likely we get:
  1. Bd1-e2
  2. Bd1-g4
  3. Bd1-c2
  4. Bd1-b3
  5. Bd1-h5 (when there is no where else to put the bishop and there is a knight on g6)
And a final note! Here is an example where Bc3 actually works very well, which appears to contradict everything that I have already said about it being no good. However note that black cannot exchange off the bishop on c3 with a knight and note that if black plays BxB white get's Nc3 for no cost.

SP619 white to play: Bc3 works!

White's plan is possibly Bc3/Nd2/O-O something like that.

Enjoy 960

Friday, January 13, 2012

Chess960: Odd midgames or not?

As for the question whether with absolute best play, does Chess960 produce lots of unusual mid-games relative to traditional chess, I cannot answer. The question will be resolved after lots of Chess960 has been played and then compared with the huge traditional chess database. Another possible way to go, is to compare the CCRL FRC database with the CCRL traditional database.....The idea would be to search both databases for unusual midgame piece combinations that mathematically balance. For example, there are over 100,000 games in the CCRL-FRC database now, what are the proportions in each database of 960/traditional games that feature unusual midgames say two knights verses a rook and pawn? The problem is that the CCRL-FRC database is polluted by engines that do not know how to play the Chess960 opening at anywhere near best practice.

Anyway, here is a somewhat unusual midgame, but it cannot be said that Chess960 created this because the scenario below was not arrived at with best play!

SP425 black to play: find the only drawing move!

Here is the same idea but arrived at from SP009:
Again, black to play for a draw
Answer given below......

Puzzle 1:
1. ...h3! which nicely blocks white's h2 pawn which will be swept away by black's g1 rook good enough for a draw. And now to feature the fantastic tactic:
2. g3? ... Rg2+!

Puzzle 2:
1. ...h3!

Wednesday, January 11, 2012

Chess960: Queens in the Corner

Queens in the corner are a challenge. To make it a bit simpler, I gathered some statistics for queens in the corner with respect to adjacent knights. I did that because I think that the position of at least one knight with respect to the queen in the corner, forms the main theme of how the queen will develop. Why is that?

Take this combination:

When the knight develops we most likely get:

The point is that with a knight on c3, the queen in the corner is sheltered from her only main threat, fianchetto bishops on the long diagonal opposing the queen.

The other main combination that produces the same pattern is:

If you do the maths on these two patterns here are the results:
Number of positions with Q in corner (and it's mirror) = 240
Number of positions with QN or QxxN (and it's mirror) = 144
Therefore this theme occurs 60% of the time with queens in the corner.

Now if you take the other main theme of queens in the corner with respect to a knight:

If you do the maths on this pattern here are the results:
Number of positions with Q in corner (and it's mirror) = 240
Number of positions with QxN (and it's mirror) = 64
Therefore this theme occurs ~25% of the time with queens in the corner.

With so many things about Chess960, learning to play it comes down to working with the likelihood of certain start configurations turning up over the board. If you learn how to develop the queens in the corner with respect to only the three patterns above, you have covered  roughly 85% of all queens in the corner scenarios!

The suggestion is that knowing these three patterns will allow you to think more clearly about how to develop the queen in the corner horizontally or vertically because the structure of the knight and queen in the corner has a nice tension between protecting the queen but also impeding the queen.

Enjoy 960