by Douglas Zare
25 March 2004

Backgammon is a race with obstacles. It is relatively easy to understand the race. The hard part is understanding the obstacles, and how they affect the race. This is not the most common view of
backgammon, but it is particularly useful in holding games. You may
find this perspective refreshing even if you don't adopt it at all times.
Usually, you try to break contact when you are ahead in the race,
and preserve contact when behind. "When ahead in the race, race." This is not universal
| The rest of this article (42.63 K) is premium content. |
Article text Copyright © 1999-2009 Douglas Zare and GammonVillage Inc.
Doug,
I am confused. Under position 2 you say "Red has about the same advantage here...". This is somewhat confusing. Same advantage as in Position 1? This is hardly so as the racing lead is gone. Same equity as White? Then there is no advantage.
The benefit of this article to me was realizing the holding role reversal when the midpoint is cleared. Good point.
Ilia
In position 2, the "same advantage" is supposed to mean about half of the advantage needed to double, just as in a race of that length. Both position 2 and a race of the same length and wastage are worth roughly 62% wins. There are slightly fewer wins in position 2, but a few gammons, too. 72% wins would correspond to a borderline double in a race.
Douglas Zare
Excellent article again, Douglas. I've always felt there were a few moves that don't work as well as they should, and some that have worked better than they should. Maybe these will make more sence to me now.
In Position 6, what do you mean by "too many checkers in the outfield"? The position in queation has eight in the outfield, which seems quite a lot. Is the risk an increased gammon danger when your straggler gets blitzed?
Good point, and I should have been more explicit about the gammon danger.
In position 6, there are a lot of checkers in the outfield (12 extra crossovers) at the moment. That means an early 5- 5 for Red will win many gammons. However, unless Red rolls 5-5, it will be some time before Red is prepared to attack. While Red is making points and bringing ammo into range, White will bear in, lowering the probability of a gammon when Red attacks.
The current count of extra crossovers is more important if White is likely to get attacked immediately. This can be a big concern if Red were able to blitz with all high doubles, e.g., where Red would want to play 8/2(2) 7/1*(2) with 6-6, or if Red already played 6/1*(2) with half of an escaping 5-5, or if Red had awkward spares on low points that may be used to hit loose immediately.
Douglas Zare
In position 6 you speak about crossovers while in position 11 you speak of pips both positions referring to gammon chances. Since there is something like epc shouldn't we have someting like egc ( effective gammon count) to? Perhaps there is even a more ore less simple way to get it from epc?
Thanks for all that article btw. It helps my understanding backgammon a lot.
Artur Barth
You are welcome.
I'm working on a gammon count formula for estimating the probability of a gammon in case you get closed out immediately. Francois Tardieu already has one based on the total pips and crossovers outside, but I think something like the epc can be more accurate. However, an important factor is the strength of your offense. You lose more gammons when your opponent doesn't worry as much about getting hit in the bearoff.
In position 11, the reason the decision to run is so sensitive to the race is that the probability of winning if Red stays back does not depend much on the race. Of course, if Red runs, each pip is worth another 2%, or 0.100 after double/take. Extra gammons from an early attack are not a large factor there.
Douglas Zare
Red has 14 checkers in #5. Borne off one?
Hi ekw,
Yes, in position 5 Red has one checker off already. You can actually see it on the left side (it's very small though) and there is a number "1" indicating it is off.
Best regards,
Michael Strato GV Editor
"Relative equities determine who is holding" Good decision rule Douglas.
What I mean with egc (effective gammon count) compared with epc was more like this: to get your odds of being gammoned if you decide to break contact and start running against it you advance all checkers six pips, take the epc of that and adjust for half a roll and adjust something for the position in your homeboard (gaps on low points)before the shift. Now the wins of opponent in the shifted position should be equal to your gammon losses in the actual position. Of course this will be only one half of the medal and you will have to decide whether you get enough shots, whether your board will hold a shot checker and about the danger loosing further ground by hits of your opponent etc.
One real big advantage of an egc would be that you wouldn’t have to learn much more if you already know epc. For example running with some checkers on your mid- and eightpoint might be compared wit stack and straggler positions in races.
Best regards Artur
Has anyone got a specific reference to Tardieus formula for the probability of getting gammoned which is mentioned above ?
Regards
You must be signed in to post comments.

