Position # 53

[PersianLord]

Position information:

Pip-count:   Red 109-88 White
Score:        Red    0-4   White (Crawford game)
Match:                 5-pointer

Red to play 6-6

[PersianLord]

Spoiler

Position ID: tt0LACBs2wZDAA , Match ID: 8Am7AEAAAAAA

This is a position wherein I disagree with GNUBG's move. Obviously the most simple move is to close-out the White's straggler which the bot seems to favor. But I think we need more vision and foresight regarding this position. I took into account the following considerations during the actual match and avoided hitting:

1- White has 13 men in his home thus closing-out would not give Red a huge chance to win a gammon. Note that even a gammon at this score is not so tasty. With 1-away,4-away White can have a free drop and with 1-away,3-away he can't and this is all the difference. I won't call this a bargain, if we consider the risks involved.

2- What will happen if Red close-out White? Yes, Red will conduct his bear-ins with more ease, BUT what about bear-offs? If White get a late shot he will have a great chance to close-out Red, considering his excellent board and this might change the fate of this game.

3- What will happen if Red doesn't close-out White? Well it's obvious that, considering Red's 6-pt prime, White will have to move inward his board and since Red has plenty of time, White's fearsome board will crash. Now, if during bear-offs White get a late shot, his chances to win the race and game will be much less because he can't close-out Red.

4- White has little chances to win the race even if: i) Red doesn't close out him now and letting him making a speed, bear-off wise board AND ii) Red get some bad rolls during his bear-off stage, letting the White's man run early.

My move was 17/11-13/7-8/2(2).

[close]

[dorbel]

Spoiler

PersianLord's recommended play isn't bad, but completing the close out now is slightly better. There are several reasons for this. If Red doesn't complete the board now he isn't certain to do so in the future. He may roll awkwardly and never close that last checker out. He may also have to hit loose at a time when his sixprime has broken, which always carries the risk of a joker. It is true that allowing White to keep playing for the moment may force White to crash, but that isn't certain either. White will of course kill sixes on the next roll and she then has a lot of fives and fours to play before the 6pt has to go. When White is forced to crash, that will increase her racing chances at about the same rate as her hitting chances decrease, so her win ratio will remain fairly constant. PL's choice is imaginative, but slightly worse than the obvious closeout, as rollouts clearly show.

[close]

[blitzxz]

Spoiler

Simple will usually maximaze single wins and because score, like PL said, there's no reason to go for gammons. I would propably try to take the extra gammons if in gammon go but not sure for money play. Differences can't be big though.

[close]

[dorbel]

 

Spoiler

It doesn't appear to be correct to eschew the closeout at any score. If Red tries 13/1*, 8/2(2), he rarely picks up a second checker and most often of course has to complete the board soon anyway.

[close]

[playBunny]

Spoiler

Rollout 1296 trials, 2-ply for 12 plies then 0-ply. 

#    Move                  Equity    (Diff)      Win     WinG    WinBg --- Lose    LoseG   LoseBg
1    8/2(2) 7/1*(2)        +0.950              97.4%     2.3%     0.0% --- 2.6%     0.0%     0.0%
2    13/1* 8/2(2)          +0.918 (-0.033)     95.6%     6.2%     0.1% --- 4.4%     0.1%     0.0%
3    17/5 8/2(2)           +0.906 (-0.045)     95.1%     4.4%     0.1% --- 4.9%     0.2%     0.0%
4    17/11 13/7 8/2(2)     +0.905 (-0.045)     95.0%     5.0%     0.1% --- 5.0%     0.1%     0.0%
5    13/7(2) 8/2(2)        +0.897 (-0.054)     94.6%     4.7%     0.1% --- 5.4%     0.2%     0.0%
Std Err                     0.001               0.1%     0.1%     0.0% --- 0.1%     0.0%     0.0%

[close]

[blitzxz]

It doesn't appear to be correct to eschew the closeout at any score. If Red tries 13/1*, 8/2(2), he rarely picks up a second checker and most often of course has to complete the board soon anyway.

2-away 1-away crawford at least gives not closing better in gnu roll-out.

[dorbel]

Absolutely correct yes, I should have said that PL's play isn't correct at any score. At 2-away, 1-away it would be correct to play 13/1*, 8/2(2) and that might even be as good as the closeout for money. However, it is a play that requires more skill than just closing out, notably when you face the "Do I close the board now, or do I give it another roll to try and recycle a checker?" It may also be correct later to break the 6 prime before closing out, another tough play that you give yourself. White's play remains trivial throughout of course.

[lewscannon]

Yeah, I close out the board and take the point. The next game, at 4-1 non crawford and cubed on the first roll, is the one to get fancy in, if you get the shot.

[zare]

What will happen if Red close-out White? Yes, Red will conduct his bear-ins with more ease, BUT what about bear-offs?

Spoiler

An early closeout means Red will have a better distribution of spares, and will leave fewer shots in the bearoff, and those shots will come with more checkers off.

If you delay closing White out, and White crashes, and then you successfully close White out, you have not gained. You have traded losses from unavoidable late hits for some losses without getting hit, and some losses from early, avoidable shots you risk because the race is in danger. For the same distribution of Red's spares, White will typically win slightly more with a speed board than with the current position.

Think of how the position will look if White rolls 6-6 as soon as you break your board. With a speed board, White may be a roll or two behind, but most doubles will work and you might miss, so White may have 20%-40% racing chances. With the current position, only a few doubles will give White 4 crossovers, so White would win the race only a few percent of the time.

The main benefit from delaying the closeout is the chance to pick up a second checker for the gammon. That is close to worthless at Crawford odd-away.

The main cost is that a closeout is not guaranteed. If you reach a closeout about 95% of the time later, you sorely regret your fancy play 5% of the time. That's a large risk for this position, which should be very safe. If a later closeout were certain, then the difference between the plays would be small. Since you might roll an awkward 5-5, or you might get hit and then crunch while trying to escape, or White may anchor (worth a few wins more than a closeout), the closeout is clearly right.

[close]