Code:
1. 20/17 20/15 Eq.: +0,100
0,548 0,073 0,003 - 0,452 0,243 0,052 CL -0,124 CF +0,100
2. 20/17 12/7 Eq.: +0,063 ( -0,036)
0,544 0,066 0,003 - 0,456 0,254 0,058 CL -0,155 CF +0,063
3. 20/15 12/9 Eq.: +0,062 ( -0,038)
0,544 0,064 0,002 - 0,456 0,252 0,056 CL -0,153 CF +0,062
4. 20/12 Eq.: +0,052 ( -0,048)
0,540 0,063 0,002 - 0,460 0,252 0,055 CL -0,162 CF +0,052
5. 12/4 Eq.: -0,066 ( -0,166)
0,505 0,054 0,002 - 0,495 0,270 0,053 CL -0,256 CF -0,066
Rollouts have shown me that playing from the 20-point is only really profitable when white occupies the 15 or 16-point, and just arbitrary when white occupies the 14, 17 and 18 points. In case of the 14-point it is ofcourse unwise to play 20/17 20/15, and if the left checker on the 20-point is hit there are less indirect shots from the bar.
Also moving the white checker from 16w to 19 makes departing from the 20-point arbitrary, and this drop in relative value is mainly because the two checkers can be played without leaving a third behind.
Even when 16w and 19w both contain several builders, departing from the 20-point is a lot better:
Code:
1. 20/15 12/9 Eq.: -0,168
0,481 0,035 0,001 - 0,519 0,279 0,060 CL -0,340 CF -0,168
2. 12/4 Eq.: -0,259 ( -0,091)
0,450 0,028 0,001 - 0,550 0,293 0,056 CL -0,420 CF -0,259
This is ofcourse not because leaving the 20-point is so strong, but because moving up the lonely checker is so weak. The value of departing from the 20-point diminishes when the lonely checker has to move less up:
checker on 12-point:
Code:
1. 20/17 20/15 Eq.: +0,100
0,548 0,073 0,003 - 0,452 0,243 0,052 CL -0,124 CF +0,100
2. 12/4 Eq.: -0,066 ( -0,166)
0,505 0,054 0,002 - 0,495 0,270 0,053 CL -0,256 CF -0,066
checker on 13-point:
Code:
1. 20/17 20/15 Eq.: +0,100
0,551 0,074 0,003 - 0,449 0,246 0,052 CL -0,119 CF +0,100
2. 13/5 Eq.: -0,030 ( -0,130)
0,515 0,060 0,002 - 0,485 0,265 0,055 CL -0,229 CF -0,030
Position ID: vJ9xAABmGxAwNg Match ID: UokOABAAAAAE
checker on 14-point:
Code:
1. 20/17 20/15 Eq.: +0,088
0,545 0,072 0,003 - 0,455 0,246 0,053 CL -0,134 CF +0,088
2. 14/6 Eq.: -0,052 ( -0,139)
0,511 0,057 0,002 - 0,489 0,267 0,056 CL -0,241 CF -0,052
Position ID: vJ9xAABmGyAwNg Match ID: UokOABAAAAAE
checker on 15-point:
Code:
1. 20/17 20/15 Eq.: +0,097
0,551 0,071 0,003 - 0,449 0,246 0,054 CL -0,124 CF +0,097
2. 15/7 Eq.: +0,016 ( -0,081)
0,532 0,057 0,002 - 0,468 0,262 0,056 CL -0,195 CF +0,016
Position ID: vJ9xAABmG0AwNg Match ID: UokOABAAAAAE
There is a 50% drop in the relative value of 20/17 20/15 (0,081 vs 0,166) when the checker can be played outside the homeboard in comparison to the original position. In his book “Modern Backgammon”, Bill Robertie warns against semi-killing a checker, but in an open field like in this position, I can't see so much harm. I also doubt Taper_Mike's perspective about the bad dices, because white will be next turn, and it is not likely that much will be left of white's outside point. 3-point backgames are not the bot's strongest point, but for convenience we presume that the bot is right. The suggestion is maybe that the option of creating a point in direct range of the 3point-gap is important. And another thing that is important is to have two free checkers close together in order for better containment of a hit white checker.
In order to save energy and not to make too much fuss about little I propose to play always from the 20-point in these kinds of positions.
PS Commentary and criticism are welcome