Midgame Recubes part 2

In the last post we were looking at a five point match. White leads 1-0 and holds the cube and things have gone her way. There's a very strong resistance to recubing anything that has a strong gammon chance at this score for obvious reasons. Here's White on roll in the position immediately following the last position in the last post. Black is on the bar but has managed to grab the 20pt anchor.

Position One


White was too good last turn but with Black having anchored on the 20pt, she can't be too good now. Is she good enough to redouble? It's close, but XG says that she should. How might we arrive at that decision OTB? It's very tough. To me, just looking at it as it stands, it doesn't "look" like a cube. Black will certainly have a game if he can enter somewhere soon, but he can also enter awkwardly. White can lose her market when Black dances again and when she can make the ace point on Black's head. Can't blame anybody for not redoubling this one though.

Position Two


White didn't double and Black danced. Now the redouble is very clear, almost too good in fact. White should redouble and Black should pass, as he has dipped below 25%. He will have a game of some sort if he can enter quickly, but when he can't, his blot on the 24pt is very vulnerable, particularly to doubles that usefully switch points. Again White chooses to roll, a reasonable decision if a small mistake.

Position Three.


Black has entered on the 24pt and been forced off his 20pt. Although White can now attack him there and will, even if she has to hit loose, this isn't technically good enough for a redouble that puts the match on the line. However, it's very close to being a redouble and a 7% chance of a wrong pass is enough to make it correct? There must be that surely? White should recube and give Black the headache. It can't be stressed too often. Look at it from the opponent's point of view. Does Black really want to take this one on for the match? The bot says that he should, but players aren't bots and will be thinking things like "3-0 down isn't so bad" rather than the more useful "have I got 25%?"
Again White didn't double and rolled 3-1, played 10/6. Black rolled 5-1 and made a fatal error, playing 14/13, 6/1. Suddenly White has one more spare and Black has one less. He should have played the brave 14/8, duplicating fives and keeping his checkers in play. That leaves this last position.

Position Four


White hardly had a recube at all last turn, now she is too good again! A recube is a large error and Black mustn't be given the chance to pass. I'd really like to know whether White took these four decisions by assessing them well, or whether she just thought, "Haven't a clue, let's roll and see what happens next"! I think all of these are too tough to assess with a sufficient degree of accuracy to make a guaranteed correct decision. If I was White in these positions, given Black's strong board, I'd have been doubling to see whether he was prepared to play one of these for the match. As I said at the start, cube holders 4-away are very reluctant to redouble and in my experience the player 5-away is very reluctant to take, so it's a bit of a stand-off!
I would recommend playing Position One here for practice after doubling. If you can get this one right and redouble, you'll be spared all the other decisions! Let me know if you managed the theoretical equity of 71-72%.
Until the next time, enjoy the game!

More Backwoods v. Zyxtcba, part 5.

Today I received the results of David Escoffery's Fibs Fall Open, 2011. This long running tourney attracted 91 entrants at $50 a time and the results were as follows.

David McAllister (rastaman) from Australia defeated Matthias Krings (sita) from Germany in the final and the losing semi-finalists were Klaus Evers (Phoenix) from Germany and Roberto Litzenberger (roadkillbooks) from Argentina. Also in the money after reaching the last eight were the Americans Eddie Branch (eddiebranch), Scott Simpson (notverylucky), John Graas (jdg) and Chuck Bower (c_ray).
You can enter twice in this tournament and there is a consolation event as well. Thomas Lusk (USA) won this defeating Klaus Evers in the final and Michael Schoppmeyer (stoerchle) from Germany and Joe Potts (mrjoe) from the USA also cashed.
Entries are now being taken for the Spring Open, so go in the first instance to the rules page .
The draw is on March 25th. Lets give this excellent international tourney a full bracket of 128.

Two more cube positions from zyxtcba v. Backwoods. In this first one, Black (zyxtcba) is on roll leading 6-5 to 13. White is on the bar.



This is a very good reference position. Any time that you have made the 6-5-4 block, you should be alert to the possibilities of a double. I like to put a mental flag on it, saying "Cube Hot!" With a White checker on the bar Black should definitely be doubling this one. A lot of people would pass this, me included, but White has just enough going for him to eke out a highly marginal take. White's 5pt is made and Black has no active builders with which to press home the attack. His awkward stack on the midpoint will provide those, but that's a turn or two away yet. Does Black think that he is too good to double? We don't know, but there are two ways to approach this when you are not sure about that. The first is to play on and see what happens next, the second is just to double it anyway and I very definitely like the second approach. If it is slightly too good, well at least cashing puts a point on the scoresheet, but sometimes they take! If it isn't too good, then the double must be right. I guarantee that the last thing White wants to see here is the cube, so put it to him. I only play on when it is clearly right to do so.
A few turns later the game has progressed to something much more static. Black is on roll again.



A 22pt anchor behind a five prime is usually a pass, particularly when as here, the doubler has one man back against three back for White. Either of these factors can make for a double on their own, together they are very strong. Worth remembering, this is quite a common situation. Black failed to double this though and cashed next turn when he was even stronger.
Bold cube action wins matches. The first thing that you will notice if you watch world class players in action is that they double very aggressively, more so than at any time in the past. The first thing that you notice about beginners (not that either of these players is a beginner, I don't mean that) is that they miss lots of doubling opportunities. Use dorbel's rule. "If I was him, would I want to be doubled? If the answer is no, turn it!"

More from this match tomorrow. Until then, as ever, enjoy the game!

Fibs League 54 Play-off

In my last post I promised you some riveting "defending against a back-game" stuff. The game turned out to be even more difficult than I thought, with very long rollouts needed and a lot of thinking time, so rather than present you with something half-baked and not thought out, I'll postpone that for a bit.
Not much use trying to explain it to you if I don't know what's going on in the first place!

Instead, some reporting. Fibsleague, Franck del Rio's long running contest reached the end of session 54 with a 13 pt play-off between zyxtcba (USA) and Backwoods (Finland). Zyx had never won the title, while BW was going for his record setting sixth crown and was the favourite in most eyes. Zyxtcba was the winner in a thriller, 13-12 so let's take a look at some of the major decisions.

Game 1. 0-0 to 13. White (Backwoods) on roll.



White has caught Black cold and put two checkers on the bar and correctly doubles this. Can Black take it? Zyxtcba passed and you can hardly blame him. Nobody likes to take with two men on the bar, but it's a fairly clear take and the pass is a blunder. What are the features that might lead us to the right decision? White only has a two point board as yet and still has only eight checkers in the zone. The midpoint is still stacked and the back checkers haven't moved. The point that White has made is the 3pt, useful but not a killer and the point that White has slotted is the 2pt, not ideal.
How can we add all this up and get "take" at the end? If only there was some nice formula that took all these factors into account, but that is the Holy Grail of backgammon theorists. Keep this one in mind next time you get caught out by an early blitz. Improvements for White, a better point made, a better point slotted, split back checkers, another builder in the outfield, any of these can push this one into the drop box.

One of the biggest equity sinks for all of us is the missed double. There are a number of reasons for this. First and foremost, doubling any position gives your opponent the chance to make a mistake, passing when he should take or vice versa. Also, if taken you can't then make subsequent missed double mistakes! A string of four or five missed doubles hurts a lot. A third subtle feature is that most players below expert level handle redoubles poorly, often missing a chance for a correct 4 cube. Handing the cube to an opponent of this type eliminates your chance of a doubling error and opens the opportunity for your opponent to make a big mistake or two.

Game 2. Black 0 White 1 to 13. Black on roll.



Zyxtcba (Black) misses a chance to double here. He has a small lead (128-136), he has escaped his back checkers and his front position should develop nicely, giving him winning chances racing, priming or even blitzing in some variations. White can still take. His own front position is OK, so if he can hit a shot he can turn it round quickly and he might jump out and win the race too. Nevertheless, Black should double this and let White have a problem, even though he probably knows that White, a bold taker and redoubler, will make good use of the cube if he can.
Black missed two more chances to double this one before eventually cashing to level the scores.

Game 3. 1-1 to 13. White on roll.



Backwoods (White) missed a very rare doubling opportunity here. Usually doubling while you are still anchored like this is a no-no, race lead (127-148) or not. Here though Black is going to have a lot of difficulty unloading that grossly obese midpoint and building a board. White's plan is to run the blot off the 20pt and then wait for a double or a 5-x to clear his anchor. After that, the race should be easy to win. I can't see that I would have doubled here either though. I am just being wise after the event.
In the match Backwoods ran the blot with a 6-2, zyxtcba eschewed the hit after rolling 2-1 and we got to this position.

Game 3. 1-1 to 13. White on roll.



Here's why it was correct for White to double last turn. He has escaped his blot and gained in the race and this is now a pass! This is really exactly the same as when you double with the midpoint to clear against a bar anchor. White has plenty of time to roll the double that does the job and should only face one single shot when he doesn't. In the match he again chose to roll and rolled a 5-5 for mega market loss. Two very unusual positions and we'll go on and look at Game Four tomorrow. I do mean tomorrow I promise.
Until then, enjoy the game!

The Famous Five Find The Answers.

The "Famous Five" were the heroes of a series of British children's books written by Enid Blyton, although I doubt if they are much read today. On their holidays the intrepid and ageless quintet invariably encountered a criminal gang engaged in nefarious nocturnal pursuits and defeated them with skill and derring-do. What Julian, Dick, Ann, George and Timmy the dog would have made of our Croatian Quiz is anybody's guess, but thanks to the Famous Five readers who tackled them on behalf of everybody else.
Before we take another look and reveal the answers can I say that these are really tough problems that might well have beaten anybody, because they are not easily solveable without a lot of thought. Each is in one way or another unusual and not responsive to the "usually I do this" method" that gets us through most situations.
Here's Position One again.



White is "stripped and stacked", a hugely inflexible position to be in. If she can't throw a life-saving double there is going to be a blot somewhere that Black, trailing by 8 pips after the roll, will want to hit. To take advantage of that he needs to cover his blot and over the board I would have tried 13/9, 6/4. This got one vote from the panel and is the third best play. It does beat 22/16 which got two votes, but that play fails on two counts. It doesn't make the four point board and it runs into a race where Black trails. It is a blot that has no useful function when it isn't hit; a blot on the landscape one might say. Another panellist tried 23/21, 13/9, but again, it doesn't make a four point board. It threatens to make it, but that is never as good as actually making it! Lastly we had one vote for 23/21, 8/4, which was also my student's play.
And the right answer is...............22/20, 8/4!
There is an excellent YouTube video where Falafel, Bob Wachtel and Michy discuss a play from Bob's match. Then Bob reveals the right answer as per computer rollout. Falafel says, "Now I know the answer I understand the position", which is very funny and also very wise. So, what is happening here? White's position is the key, because she is really stuck for play with her outfield points all stripped and her 6pt stacked. What XG's play does, after it has crucially made the 4pt board, is to deny White safe landing points. Playing 23/21, 8/4, the second best play although still a large error, allows White to dump checkers onto her 1 and 2pts and even more importantly, denies her the option of playing to the 11pt in relative safety with most twos and fours.

Position Two was a lot easier with only three posible plays.



Bar/20/14 got no votes from the panel, another "blot on the landscape", trying to race into a race where Black trails, although not by much. Bar/20, 24/18 got two votes and is the usual answer to this sort of thing, forcing contact while Black has the better board, in the hope of exchanging hits or making a high anchor. Three of our panellists went for bar/20, 13/7 and that is clearly best. It beautifully illustrates two themes that recur constantly. The first is that Black has killed two checkers on the ace point and fatally wounded two more on the 3pt, which is the casualty list when a 5-5 blitz fails. This means that his remaining 11 checkers all have to work extra hard, doing the work that is usually done by 15 men. In bg terms, this means more risk. In addition, Black still has a stack on his midpoint, 5 out of his 11 playable men in one place. I always put a mental flag on a stack that says "Urgent: please unstack me". It's no use saying, "I can't do that now, I've got other problems". Positional flaws like this have to be dealt with immediately. Again, in the immortal words of Falafel (possibly quoting Italo Calvi), "If not now, when?" The handy duplication of threes makes the medicine go down a little more easily.

I bet you can't wait for the other answers, but you will have to wait until tomorrow, as I have a flat to tidy before mrs dorbel comes back from her holidays.

Mochizuki Masayuki v. Matvey Natanzon, a 13pt IIBGF Match.
Mochizuki Masayuki v. Carlo Melzi, a 13pt IIBGF Match.

The XG files for these matches, annotated by me and approved by Mochy are now available, $10 each. Gnu and Snowie users can have a Gnu file with the commentary on a separate word doc. Both matches show brilliant play at the highest level and are very instructive for students at all levels.

PayPal please and enquiries to dorbel(at)gmail(dot)com.

Until tomorrow, enjoy the game!

Moet or Veuve?

One of the most pleasant decisions that we face comes when we have to decide between playing on for the gammon or doubling and taking a point. This is rather like the man in the Marbella supermarket (true story this), who I heard call out to his wife, "They 'aint got no Moet luv, you want Veuve?" Here are three positions from a Gridgammon match. It's the first game of a 13 pointer and the Blue checkers are being played by the expert Italian Carlo Melzi. Take a look at these and decide what is the correct cube action for both sides. You'll do well to get all three right.

Position One. Blue on roll, 0-0 to 13. White is on the bar.

Position Two. Blue on roll. 0-0 to 13.

Position Three. Blue on roll, 0-0 to 13, Blue is on the bar.

Three positions of varying toughness, so take some time and decide what is the correct action for both sides. A large and well informed gallery of watchers got at least one of these wrong. How will you do? The comments section is there, so share your thoughts.
Until tomorrow, enjoy the game!

Mochy v. Birkhahn, Game 7,

...........and Mochy leads 7-6 to 11. As with all the other games, you will get much more from this if you set up the pieces on your board to follow the game.

1. RED: (4 3) 13/9 13/10

WHITE: (3 3) 24/21(2) 8/5(2)

This is a roll that bots play badly, both Snowie and Gnu prefer to play 24/15*, 13/10. Martin knows better and takes two key points with this fine roll. 24/21(2), 6/3(2) has more safety and unstacks the 6pt, but the awkwardness after the match play is a price worth paying to get the 5pt. The blot on the 8pt isn't a great liability, as if it is hit Red will have four blots while White has a high anchor and a better board. Excellent play which the rollouts back up.

2. RED: (1 1) 24/23 10/9 6/5(2)

WHITE: (2 1) 8/5

3. RED: (5 2) 24/22 13/8

Mochy makes a small error here. It is slightly better to run out with 24/17 and not give White the chance to use his stacks for a blitz.The spare on the midpoint is very useful and worth preserving too.

WHITE: (6 2) 13/5

Good play. 13/7, 5/3* is much too big and White waits for a better attacking chance.

4. RED: (3 1) 23/22 8/5

WHITE: (6 5) 13/2

5. RED: (5 2) 8/3 5/3

WHITE: (2 2) 13/11 6/4(3)

There is no gain from playing to the 11pt, it is just a blot on the landscape. That checker is just as useful and of course completely safe on the midpoint. 6/4(3), 4/2 is clearly best. Not sure what Martin was thinking of there. He has just made three consecutive safety-first plays, so perhaps he felt that he was going to have to take some risks to clear the midpoint. There has to be some gain from a risk though and there isn't any here. Best to lock down the four point board and wait and see what happens next.

6. RED: (3 1) 6/2

No risks for Mochy. White leads in the race by 9 pips after this roll, but is about even in the game because he is behind a four prime whereas Red's anchor is behind a three prime. Not only that, leading in the race is a double-edged sword in these mutual holding games, because you may be forced off your anchor first and get attacked. Mochy keeps all his assets in place and makes a quiet play.

WHITE: (2 2) 13/9(2)

This is the play that I would have made, along with many others I expect. The bots however prefer 13/9, 11/9, 4/2. They do this because it gives Red awkward ones next turn, being unable to hit and cover without breaking his midpoint and a point in his prime. If he chooses not to hit, he usually has to break his prime. This sort of play is hard to find over the board but it is worth noting the circumstances that have created the opportunity for it. Red has TMP, too many points and has to break a point with every roll this turn. You can see that this is a condition to be avoided where possible and look for opportunities to exploit it when you see your opponent in it. The match play is a very small error.

7. RED: (3 2) 13/8

WHITE: (1 1) 11/10 9/8(2) 5/4

Hard to sort these out and know what's best. Snowie prefers 9/7(2), a useful asset on later turns and easier to clear if White can escape and get into a bear-in. The match play is a very small error.

8. RED: (6 4) 13/9 8/2

WHITE: (4 4) 10/2 8/4(2)

I don't think may of us would look beyond this "forced" play, but Snowie finds something very slightly better, which is 10/6, 5/1(3)! This creates a gap and buries three checkers which is the downside but saves two sixes and a five on the upside. This buys some valuable time for White, so that he isn't forced off his anchor. It also makes it harder for Red to run out next turn. Note that Red still has TMP and will probably have to break something somewhere next turn.

9. RED: (6 3) 22/16 9/6

WHITE: (4 2) 5/1 4/2

10. RED: (6 3) 22/13

22/16, 6/3 is a very close alternate here, no indirect shots at he expense of losing a valuable builder.

WHITE: (4 2) 6/2 6/4

Not a very good play from Martin, 5/1, 5/3 is clearly better, retaining a four point board. His play also kills his last two fives inboard and will force him off his anchor with a 5-3 next. The match play is a blunder costing about 0.12 ppg.

11. RED: Double

Mochy alertly doubles. Although he trails in the race by 16 pips, White's board has crashed and Red will have powerful attacking chances against the last two checkers. Red wins about 72% here, including 15 match winning gammons so this is actually a pass. The take is a blunder costing about 0.24ppg.

WHITE: Take

12. RED: (6 1) 13/6

Mochy not tempted by thoughts of a trap play like 13/7, 8/7. In general you want your opponent to be completely crashed before trap plays become viable and White may yet manafacture a four or five point board.

WHITE: (3 1) 4/1 2/1

4/1, 4/3 is better. Slotting is the way to build a board. Martin may be thinking, "I don't want a blot inboard if I hit a shot next turn", but for that to hurt he needs to get a shot, hit it with a number that doesn't cover and then be hit back from the bar. That's a pretty long parley, certainly under 10% here, so the rest of the time he will gain or break even from slotting. The match play is a small but clear error.

13. RED: (4 2) 16/14 6/2

WHITE: (1 1) 4/3(3) 2/1

14. RED: (3 2) 14/9

WHITE: (3 2) 5/2 5/3

15. RED: (5 3) 9/1

Good play by Mochy. Clearing the 8pt and creating three spares (8/3, 8/5) looks very flexible but unblocking fours is a big downside. The match play may look less flexible but it still has three checkers aimed at the 4pt if White runs with one. The blot on the ace point isn't a big liability and may even turn into an asset.

WHITE: (5 2) 3/1

16. RED: (3 1) 6/3 2/1

WHITE: (5 2) 3/1

17. RED: (6 3) 8/2 8/5

This play provoked a lot of discussion in the whispering gallery. It is probably correct and 9/3, 9/6 is a very close alternate. The match play creates larger attacking chances at the cost of being harder to clear when the attack doesn't materialise. Toss a coin.

WHITE: (6 5) 21/10

18. RED: (6 5) 9/3 9/4*

WHITE: (4 1) bar/21* 10/9

19. RED: (6 1) bar/19 5/4*

WHITE: (4 3) bar/21* 9/6*

Perhaps the toughest play of the whole match. This play is right for money but wrong at this match score, even though it wins more often and wins more gammons! How can that be? The actual figures look like this. I have averaged Snowie's and Gnu's cubeless rollouts.

After the match play White will win 66.9% including 7.65% gammons. Red will win 33.1% including 3.8% gammons. These are cubeless figures, i.e. they assume no later cube action.

After bar/21*/18, White will win 66.85% including 4.8% gammons and Red will win 33.15% including 3.35% gammons. We can see that everything is pretty even except that the double hit wins nearly 3 more gammons. However the single hit is correct and by a large margin. To understand why, we have to know what White's correct cube strategy is holding the cube at this score. He wants to redouble any reasonable opportunity, to kill the cube, kill Red's gammons and activate his own gammons as match winners. He can redouble anything next turn as long as he isn't on the bar! He still leaves 7 fly shots after the single hit of course, but that is better than 11 after the double hit. Not being hit is the deciding factor here, to maximise his chances of shipping a strong cube next turn. Red needs a little more than 30% to take at this score.

What if Red dances again? With one man on the roof White will always have a cash. With two men on the roof he will be too good by a very small margin, so you can see that he doesn't need the extra risk of being hit.

20. RED: (3 1) Can't move

WHITE: Double

Technically this is just too good to double, but to decide that over the board is almost impossible and a cash now is a good practical play, even though a medium sized error costing 0.067 ppg.

20. RED: Pass

And now runnerup (White) leads 8-7 in this excellent match. The next game is the last of the match and I hope to post that next week. Until then, enjoy the game!