While we couldnt' see Neil's #s from his GNU rollout he said 24/20 was only worse by about .035. I felt it was a much larger error to try to get away with this split. It's too big, end of story. If I make the 5 pt. you'll be sorry either way of course, but if I make it without you there, without POH, then you have a roll to hit any indirects I will leave or throw some of your anchoring up numbers to anchor up on the 3 or bar pt. Coming up to the 5 pt. also gives me numbers that while they may not make the 5, pick n pass or hit both.
Also, if you come up to the 5 pt. I'd have no hesitation about spinning the cube. Doesn't necessarily make the play wrong, but it certainly helps build a case against it. (how would you feel having to take a cube after you made that play?)
| 148 |                          |
| 153 | |
| Board image courtesy of GO-Figure | |
| # | Ply | Move | Equity | |
| 1 | R | 24/23 6/3 | -0.807 | |
| 0.2% 5.3% 31.5% 68.5% 19.6% 0.8% | ||||
| Live cube rollout: -0.811 | ||||
| 95% confidence interval: - money cubeless eq.: -0.519 ±0.016, - live cube: -0.811 ±0.037. Rollout settings: Full rollout, 648 games (equiv. 22008 games), played 3-ply (precise), cube 3-ply, settlement 0.550 at 16 pts, random seed, with race database. | ||||
| 2 | R | 24/20 | -0.896 (-0.089) | |
| 0.2% 5.6% 30.9% 69.1% 23.0% 0.7% | ||||
| Live cube rollout: -0.889 | ||||
| 95% confidence interval: - money cubeless eq.: -0.561 ±0.015, - live cube: -0.889 ±0.040. Rollout settings: Full rollout, 648 games (equiv. 22632 games), played 3-ply (precise), cube 3-ply, settlement 0.550 at 16 pts, random seed, with race database. | ||||