Louisville Backgammon Club

Frank’s Position Analysis — May 2005

by Frank Frigo

This month’s problem comes from the recent 2005 ABT Midwest Championships in Chicago. Every March director Bill Davis produces a backgammon extravaganza of high-level competitions and entertaining side-events. This year’s tournament was no exception. A few years back, Bill added the prestigious Grand Crystal Beaver event. This is a higher stakes single elimination tournament with the use of chess clocks. Players may enter in the round of 64, 32 or 16. With a first prize of about $7000 and a Swarovsky crystal at stake, the competition is typically very strong.

On Friday evening, I found myself facing Steve Sax of Los Angeles, CA in the quarter finals. Steve is a very skilled and accomplished player. Aside from his many major tournament victories over the years, Steve was crowned the 2002 American Backgammon Tour Champion.

In the April 2005 edition of Carol Cole’s “Flint Area Backgammon News,” Chuck Bower featured several of our match positions in his article entitled Lessons from the Midwest. One of the more interesting problems Chuck discusses (see diagram above) was a decision I faced when trailing 10 to 8 in a match to 11. Chuck applies both his expert commentary and the help of gnubg (a neural net computer program) to provide some insight into the position. For those of you less familiar with backgammon software, gnubg can play a backgammon match at a World Class level and also has the capability of playing any position to conclusion 1000s of times in a matter of seconds. Not only does such a simulation provide objective analysis, but it is also simply the best method for uncovering the optimum play in almost any backgammon decision. In the problem at hand, gnubg charges me with a significant error for my play of 18/11. In fact, the neural net simulation suggests that the preferred play (11/9, 6/1) wins the match more than 4% more often than my choice does. Often, such a discrepancy can become clear when applying some fundamental backgammon principles. But as Chuck points out in his article, this one is not so clear. Chuck goes on to challenge the readers to “find the answer” for why my chosen play could be so wrong.

Since I committed this error (and since I have subsequently had some time to reflect upon the roll-out results), I thought I might take a stab at explaining why the choice of 18/11 could be such a blunder.

Lets first take an inventory of the key features of this challenging decision.

1. I lead by a whopping 27 pips in the race after my 5-2 roll.
2. My position is getting awkward. I have a large stack of checkers on my 6-point and very few spares to play elsewhere.
3. My opponent’s home board is on the verge of being dangerous for me, but at the moment he has a blot on his 5-point.
4. All legal plays leave my opponent with a direct shot at a blot.
5. Gammons don’t matter for either side. At this score, any loss is lights out for me, and unless I win a backgammon (very unlikely) I will have to win two games to capture the match regardless of whether this game is a gammon win or not.

Generally, when you have a large racing lead you are content to break contact with your opponent and bring the position home in the most simple and direct way possible. Unfortunately, I cannot accomplish this immediately. However, because of my opponent’s blot in his home board, I thought this might be a good time to try to exploit my racing advantage in an otherwise awkward position. Where did I go wrong in my logic?

First of all, my play leaves an immediate 12 out of 36 numbers that hit and cover the blot on the 5-point. That means 33% of the time I will find myself on the bar facing a strong four point board. Secondly, I can be hit with several other rolls that don’t cover his 5-point. Often I will get a return hit, but many times my opponent will subsequently make his 5-point on the next roll while I spend time on the bar or land on the 1- or 2-points. Finally, even if I don’t get hit immediately, my checker is by no means home free. In fact, Steve is still a substantial favorite to get another shot on his next roll.

Steve has his own problems. Not only is he trying to attack my checker (after my 18/11) and close his board, but he also has to extricate his back checkers over my 4-prime. This may seem like a lot to do, but remember, he doesn’t have to do it all at once. He will be happy to defer the escape of his back checkers until he finishes with me in the frontcourt. (Which is exactly what he did by the way!)

So why does 11/9, 6/1 now seem so clear?

1. It leaves the fewest shots in a forced blot situation.
2. It maintains my defensive bar point (18-point) which inhibits my opponent’s ability to close me out or build a full 6-prime, should he successfully hit. My racing lead is large enough that I can probably survive a hit as long as I don’t get closed out or primed.
3. If I am missed, I can now patiently break down my prime, build my board behind my opponent’s anchor, and wait for a more opportunistic time to jump from the 18-point. Either a shot at an enemy blot or a set of doubles will provide such an opportunity.

It may seem odd to make a play that relies upon the prospects of rolling a set of doubles or of getting a future shot as you are an underdog for this to occur on any individual roll of the dice. Remember, you are trying to win the war, not just the battle. When you have several consecutive opportunities to roll an advantageous number, it can become far more probable than one might imagine. As long as you have reasonable alternatives, patiently waiting to clear your defensive anchor while preserving your racing advantage can be a very solid strategy. Such is the case here.

– Frank Frigo

Return to Features