I am a non math. This is a serious flaw in my performance, espacially in match. Often i decide by experience and “gut-feeling”. So a closer view, which is possible without bots (Stick or Bill will do this over the board, we fishes can do only an educated guess).
Five point match 3away, 3away
“Play is close to normal backgammon at this score, but you should tend to drop quicker than usual. A look at the numbers illustrates why. Let's assume gammons aren't an issue, and you are being doubled.
If you pass: You are behind 3-2, 40% equity.
If you take and win: You are ahead 4-2 (Crawford), 75% equity.
If you take and lose: You are behind 4-2, 25% equity.
As can be seen the taker is risking 15% to gain 35%. This is worse than the normal money 3 to 1 odds -- you need 30% winning chances to justify the take. If there were anything special about the recube value it wouldn't be so bad, but the recube is normal. If your opponent passes, he will be behind 4-2 (Crawford) for 25% equity. If he takes, it is for the match. Therefore your opponent will have the normal 3 to 1 odds on his take; his only disadvantage will be that the cube will now be dead so he will have to play the game to conclusion.
Since the drops come quicker at this score, the doubles come quicker also since the doubler doesn't want to lose his market. Otherwise, play is fairly normal.”
(
http://www.bkgm.com/articles/GOL/Aug99/fivept.htm)
“Gammon price
The relative value of winning a gammon compared with the value of winning a single game. Gammon price is computed as GP = (WG - W) / (W - L), where WG = value of winning a gammon, W = value of winning a single game, and L = value of losing a single game. In money play, the gammon price is 50%. In match play, the gammon price depends on the score of the match and the level of the doubling cube. See posts by David Montgomery and Ron Karr." And don't forget different gammon rates. See below.
(
http://www.bkgm.com/gloss/lookup.cgi?gammon+price)
So this are two major adjustments. The takepoint may be higher, and the gammon price may be higher then in cash game. If you know this, then your “gut” will tell you, that as this score, with higher gammon rates, the doubling window will significantly change. Also, the point of first and last give will be lower. It is fine, when we know it exactly, but it is better then nothing if we know this approximately.
I will show, to get a real feeling, the exact numbers(Kazaross/Rockwell MET). Non gammon adjusted, for both players:
Take point dead/alive 30,29% / 22,76%
Double Point dead/alive 50,00% / 50,00%
Cash Point dead/alive 69,71% / 77,24%
Too Good Point: dead/alive 100% /100%
Gammon adjusted, this changes dramatically (Gammon Price: 0,769).
I take the exact GR and BGR from the RO ( Skara 45,99% / 7,84%; Wachtel 31,67% / 4,88%).
Market Windows:
Take point: dead/alive; Skara/Wachtel 33,43% / 30,78%; 39,34% / 33,78%
Double Point: dead/alive; Skara/Wachtel 48,70% / 48,70%; 51,30% / 51,30%
Cash Point: dead/alive; Skara/Wachtel 60,66% / 66,22%; 66,57% / 69,22%
Too Good Point: dead/alive; Skara/Wachtel 78,36% /84,42%; 84,42% / 78,36%
So the point of first give is for Scara below even! And he can double in only up to 66,22%. Regarding 66 55 11 52 65 61 51 64 as market losers, we have a whopping 13 throws, or 36,1%, to lose our market in next turn. And if Wilcox Snellings comes the way, putting out his wallet to give one point for the cube, i would even for money decline his offer.
Because of the confidence interval, i have to crank up the rollout assuming cash game. If there is interest, i will post XG results, 3ply/3ply variance reduction, 1296 trials for match, ???? for cash game.
Last edited by higonefive; 11-10-2010 at 04:35 PM.