Game 2, move 14: Herd 3-2

[diane]

Stiefnu rolled 6-2 moved the only sensible option, 7/5, 7/1, Herd to move

[Julia_H]

Spoiler

White has no blotting rolls next time, and even after the one roll that doesn't leave him even at the back (62) he's still got no blotting rolls the time after that (and he's not going to hit with a man on the ace uncovered; with 31 or 32 I'd consider ripping two rather than hit and cover the ace). By the time we get a shot, our board will be completely wasted. I think our best chance now is to run and hope for a quick set of boxes.

[close]

[ah_clem]

Spoiler

Pretty much like last time: gammons are few and don't mean much.  Stay and hope for a shot.

[close]

[ah_clem]

Spoiler

White has no blotting rolls next time, and even after the one roll that doesn't leave him even at the back (62) he's still got no blotting rolls the time after that (and he's not going to hit with a man on the ace uncovered; with 31 or 32 I'd consider ripping two rather than hit and cover the ace). By the time we get a shot, our board will be completely wasted. I think our best chance now is to run and hope for a quick set of boxes.

These are good points.  But I'd observe two things: 1) big doubles will be just as effective if we stay or go (assuming he doesn't put us on the bar).  And 2) by staying we give him the opportunity to make a mistake - say he rolls 21 and gets greedy. Or rolls a pick and pass number that covers the ace.  I'm not sure whether hitting is correct in either scenario, but it gives us a chance to gain equity if he chooses wrongly.  So I say stay, but I doubt there's much equity difference between the two moves (and of course, if we stand to gain equity from his hypothetical mistakes the bot won't recognize that).

[close]

[Julia_H]

Spoiler

These are good points.  But I'd observe two things: 1) big doubles will be just as effective if we stay or go (assuming he doesn't put us on the bar).  And 2) by staying we give him the opportunity to make a mistake - say he rolls 21 and gets greedy. Or rolls a pick and pass number that covers the ace.  I'm not sure whether hitting is correct in either scenario, but it gives us a chance to gain equity if he chooses wrongly.  So I say stay, but I doubt there's much equity difference between the two moves (and of course, if we stand to gain equity from his hypothetical mistakes the bot won't recognize that).

True, but we also give him the opportunity for a joker 11 or 22, which would erase what's left of our racing chances.

[close]

[ah_clem]

Spoiler

True, but we also give him the opportunity for a joker 11 or 22, which would erase what's left of our racing chances.

Are those even jokers?  With our full board we might have more equity on the bar than where we are now. The racing chances would diminished but the chances of a shot would go up. Not sure what  the net effect would be...
 
I'll roll some of these hypotheticals out once they become moot.  Interesting questions.

[close]

[ah_clem]

rollout

Spoiler

Here it is correct to run.  Not a big error, and it might have been worth it if stiefnu misplayed the 21.


Note that other running plays are better than our choice, but are not shown in the rollout results.


XGID=-BCBBBB----A-------ddAbda-:0:0:1:32:4:0:1:5:10


    1. Rollout¹    21/16                        eq:-0.735
      Player:   13.26% (G:0.00% B:0.00%)
      Opponent: 86.74% (G:0.00% B:0.00%)
      Confidence: ±0.000 (-0.735..-0.735) - [100.0%]
      Duration: 35.0 seconds

    2. Rollout¹    11/6                         eq:-0.761 (-0.027)
      Player:   11.95% (G:0.00% B:0.00%)
      Opponent: 88.05% (G:0.35% B:0.00%)
      Confidence: ±0.001 (-0.762..-0.760) - [0.0%]
      Duration: 17.0 seconds

   

¹  1296 Games rolled with Variance Reduction.
   Moves: 3-ply, cube decisions: XG Roller


eXtreme Gammon Version: 2.03, MET: Kazaross XG2

[close]

[ah_clem]

Are those even jokers?  With our full board we might have more equity on the bar than where we are now. The racing chances would diminished but the chances of a shot would go up. Not sure what  the net effect would be...
 
I'll roll some of these hypotheticals out once they become moot.  Interesting questions.

gnu puts 22 as +0.118 luck and 11 as +.087 luck.  That's good, but far short of the .3 threshold for "lucky" or .6 for "very lucky".

(I haven't figured out how to get this number out of XG)

And in case anyone is wondering, hitting and closing the board is the right play for both rolls, it's just not as much of a game-changer as it might seem.