Containment 4

[Tim Chow]

XGID=-BBCaBC---A----A-------dfA:1:1:1:54:0:0:0:0:10

X:Player 1 O:Player 2
Score is X:0 O:0. Unlimited Game
+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O |
| | | O O |
| | | O O |
| | | O O |
| | X | 6 |
| |BAR| |
| | | |
| | | |
| | | X X | +---+
| | | X X X X X | | 2 |
| X | | X X O X X X | +---+
+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 93 O: 35 X-O: 0-0
Cube: 2, X own cube
X to play 54

---
Tim Chow

[Bradley K. Sherman]

Tim Chow <tchow...@yahoo.com> wrote:
>Score is X:0 O:0. Unlimited Game
> +13-14-15-16-17-18------19-20-21-22-23-24-+
> | X | | O O |
> | | | O O |
> | | | O O |
> | | | O O |
> | | X | 6 |
> | |BAR| |
> | | | |
> | | | |
> | | | X X | +---+
> | | | X X X X X | | 2 |
> | X | | X X O X X X | +---+
> +12-11-10--9--8--7-------6--5--4--3--2--1-+

> X:93 O:35, X to play 54

Bar/21,10/5

--bks

[badgolferman]

I get these wrong all the time. Let's try Bar/21, 10/5.

[smcrtorchs]

I have been experimenting a bit with "wild" containment lately and although this is not "wild" from what I have observed bar/21 10/5 should be auto. This is because I want to have my last checker as back as possible and because I do not want to leave shots for O to hit, especially shots shots that bring O near to home even if they are indirect.

So to leave a direct shot here would seem weird, because O would have a good chance to roll this home after he hits at the 10 point. This is true especially with 2 more blots laying around for O to hit next turn if X does not hit immediately back and after bar/16 if O hits once, X will have only a little bit above 8 numbers that hit back on average.

At the same time, the checker on the 21 point has good coverage and will most likely give us another shot on that checker. There is no reason to rush things and try to get this shot next turn.

[Tim Chow]

Like other respondents, I played 10/5, but the computer clearly prefers 15/10. Let's review the three main strategic goals for outfield containment:

1. Flood the outfield for as much coverage as possible.

2. Block specific numbers, e.g., high doublets.

3. If getting hit is bad, then avoid getting hit.

Intermediate players tend to overvalue blocking high doublets, but in this position they would probably play 15/10 for that reason alone, and would get the computer's nod of approval. 15/10 not only blocks 66, but also 33 and (less importantly but not completely negligibly) 61 and 62. The reason that blocking high doublets is often the wrong play is not that blocking doublets is bad, but that often there is an alternative that spreads the checkers out much better. Here, though, 10/5 doesn't really do that. Either way, O has a few rolls that give X no direct shot, while most rolls give X a single direct shot. If you're not getting significantly better overall coverage, then the big swing on 66 starts to make its presence felt.

For your amusement, I've also rolled out the position at DMP, where the wild-looking bar/16 is a top contender.

1. Rollout¹ Bar/21 15/10 eq:-0.175
Player: 35.34% (G:0.00% B:0.00%)
Opponent: 64.66% (G:4.34% B:0.02%)
Confidence: ±0.004 (-0.179..-0.170) - [100.0%]

2. Rollout¹ Bar/21 10/5 eq:-0.248 (-0.074)
Player: 32.75% (G:0.00% B:0.00%)
Opponent: 67.25% (G:4.96% B:0.02%)
Confidence: ±0.004 (-0.252..-0.244) - [0.0%]

3. Rollout¹ Bar/20 10/6 eq:-0.253 (-0.078)
Player: 32.41% (G:0.00% B:0.00%)
Opponent: 67.59% (G:4.87% B:0.02%)
Confidence: ±0.004 (-0.257..-0.249) - [0.0%]

4. Rollout¹ Bar/16 eq:-0.321 (-0.147)
Player: 34.87% (G:0.00% B:0.00%)
Opponent: 65.13% (G:18.61% B:0.14%)
Confidence: ±0.005 (-0.326..-0.317) - [0.0%]

===
DMP
===

XGID=-BBCaBC---A----A-------dfA:0:0:1:54:0:0:0:1:10

X:Player 1 O:Player 2

Score is X:0 O:0 1 pt.(s) match.

+13-14-15-16-17-18------19-20-21-22-23-24-+
| X | | O O |
| | | O O |
| | | O O |
| | | O O |
| | X | 6 |
| |BAR| |
| | | |
| | | |
| | | X X |

| | | X X X X X |

| X | | X X O X X X |

+12-11-10--9--8--7-------6--5--4--3--2--1-+
Pip count X: 93 O: 35 X-O: 0-0/1
Cube: 1
X to play 54

1. Rollout¹ Bar/21 15/10 eq:-0.291
Player: 35.46% (G:0.00% B:0.00%)
Opponent: 64.54% (G:5.50% B:0.04%)
Confidence: ±0.002 (-0.292..-0.289) - [99.3%]

2. Rollout¹ Bar/16 eq:-0.294 (-0.004)
Player: 35.29% (G:0.00% B:0.00%)
Opponent: 64.71% (G:21.09% B:0.16%)
Confidence: ±0.002 (-0.296..-0.292) - [0.7%]

3. Rollout¹ Bar/21 10/5 eq:-0.342 (-0.051)
Player: 32.92% (G:0.00% B:0.00%)
Opponent: 67.08% (G:6.55% B:0.04%)
Confidence: ±0.002 (-0.343..-0.340) - [0.0%]

4. Rollout¹ Bar/20 10/6 eq:-0.348 (-0.057)
Player: 32.60% (G:0.00% B:0.00%)
Opponent: 67.40% (G:6.16% B:0.03%)
Confidence: ±0.002 (-0.350..-0.346) - [0.0%]

¹ 1296 Games rolled with Variance Reduction.
Dice Seed: 271828
Moves: 3-ply, cube decisions: XG Roller

eXtreme Gammon Version: 2.10, MET: Kazaross XG2

---
Tim Chow

[Tim Chow]

Oh, I meant to ask if Paul would classify this as a joker avoidance problem?

---
Tim Chow

[smcrtorchs]

On Saturday, February 8, 2014 2:31:58 AM UTC+2, Tim Chow wrote:

Although I am drunk at the moment and cannot think on the RO results, thanks for rolling this out on DMP as well. It is interesting to know that it is a different case.

[Walt]

X can make the 10 point and block sixes, but any non-small roll other
than 66, 62, or 61 will allow O to scoot past with only a single checker
with an indirect shot between her and victory.

I like spreading them around with bar/16. If hit, I'll just enter and
have 3 guards impeding her progress.

[Tim Chow]

On Sunday, February 9, 2014 3:14:19 PM UTC-5, I wrote:
> Either way, O has a few rolls that give X no direct shot, while most rolls
> give X a single direct shot.

I had some doubts about this statement after reading Walt's post, so I decided to count it out to make sure. After 10/5, O avoids a direct shot with 31 41 65 55 66. After 15/10, O avoids a direct shot with 43 52 54 63 64 55. So O does have 3 more shot-avoiders after 15/10, but that's not that many.

After bar/16, 66 55 and 33 are killers, but everything else gives X almost a direct shot (or better), even though X may be on the bar. The gammons from getting hit are too costly at a normal score, though.

---
Tim Chow