Can you double according to score?

[blitzxz]

Okey, here is a hard one 

It's a 5-point match black on roll. What's the correct cube action in these different scores?

0-0
1-1
2-2
2-1, black leads
1-2, white leads

And to make it bit of a challenge: I'm almost 100 % sure that no one will make all of these right without assistance.  If you however make it, you are truly a master of using match equity tables and evaluating this position. I will post the answers later. Please don't spoil the fun with bot analyzes. (By the way you can only see the pic if you're logged in)

[PersianLord]

0-0: D/T
1-1: D/T
2-2: D/T
2-1: ND/T
1-2: D/P

[dorbel]

OK, in the spirit of the thing, I haven't (yet) put this through Snowie! I always double this. I think I pass at 0-0, not sure about that though, as you can take more deeply than usual at that score.  I suspect that this one is a very marginal take/pass. I pass at 1-1 and even more quickly at 2-2. Leading 2-1 it's a very big pass but trailing it's quite an easy take.
Nice quiz, thanks blitz. One thing, how did they get to this position? It looks a bit manafactured.

[socksey]

Having recently had a setup exactly like this one, I would double and take on all of them except for the 2-1 which I would not double since I doubled this and lost the match with a gammon.  

socksey



"Long ago when men cursed and beat the ground with sticks, it was called witchcraft. Today it's called golf." – Will Rogers

[blitzxz]

One thing, how did they get to this position? It looks a bit manafactured.

The game went like this: (white starts and the score was actually 2-1 for black which was me)

21: 24/23 13/11  44: 8/4(2) 6/2*(2)
15: bar/24 13/8   11: 8/5 6/5
51: 13/8 11/10

I've seen something similar to this before but not very often. I'll hold on to the rollout answers still for a awhile. 

[boomslang]

Now this is an interesting post!


My cube decisions would be:

0-0 : DT
1-1:  DT
2-2:  DT
2-1:  No double + Take
1-2:  DT

Since I have my own table I just CAN'T be wrong

PS: What about the other non-trivial scores?

[PersianLord]

Now this is an interesting post!


My cube decisions would be:

0-0 : DT
1-1:  DT
2-2:  DT
2-1:  No double + Take
1-2:  DT

Since I have my own table I just CAN'T be wrong

PS: What about the other non-trivial scores?

Seems that your sophisticated table is in agreement with my gut feelings and you should be proud of that 

BTW, have you set up the positions and got the bot analysis?! If so, which bot did you consult, Snowie or GNUBG or even BGBlitz? I would be surprised to see that Snowie takes the cube when leading 2-1.

Regards -- PL of FIBS

[boomslang]

BTW, have you set up the positions and got the bot analysis?! If so, which bot did you consult, Snowie or GNUBG or even BGBlitz? I would be surprised to see that Snowie takes the cube when leading 2-1.

No I haven't...

[blitzxz]

Spoiler

0-0
correct answer: double, take

The surprising thing in this position is that white wins here about 33% of the games. Because of the anchor black wins only 25 % gammons. This makes it very clear money take. In 5-away 5-away gammons value more because doubled gammons lead to crawford-game. However it's still not enough to make it a pass.

1-1
correct answer: double, pass
Now doubled gammons win the whole match which changes everything. It's a huge pass.

2-2
correct answer:double, pass

Let's work with the numbers a bit. First lets make simple and an unrealistic assumption that white will be able to redouble perfectly. I mean white will always redouble when he wins and never redouble when black wins single. So the match winning chance would be: white wins + black single wins x white wins from 3-away-1-away. 0,33 + 0,42x0,25=0,435. This is bigger then match winning chance if white drops which is at 2-away-3-away about 0,40. So it comes to how well white can redouble here. If I assume that white will never miss markets but 10 % of games he redoubles after which the games turns for black that would give 0,33 + 0,32x0,25= 0,41. Still higher then pass equity. The actual difference in pass and take according to gnu is well under 0,5 % and it's a borderline pass. This is surprising because normally you drop easily in this score. But high winning chances and redouble possibility in this score can sometimes make correct takes which might look desperate.

2-1
correct answer: no double, take

White can take in this score with very low winning chances because he will have monster redouble for match. And white should never ever miss his markets for redouble by doubling too late. Black can however lose the markets here. If he hits, rolls big doubles or white rolls something horrible like 5-4. Then again white could also improve his position with important points. Because white's low take point black should still hold his horses and wait just a move or two. Because white has an anchor black is likely to get almost perfect double later and risk of losing markets isn't just big enough compared to white's powerful cube ownership. I doubled here in this score. And you can justify it if there is a chance that white will drop.

1-2
correct answer: double, pass

This is enourmous pass. Doubled gammon would take black from the losing side to straight winner. It's even close to too good. But black's 25 % gammon chances is not enough to compensate for white's superjokers 1-1, 3-3, and 6-6.

Roll-outs are in the text-file. These are 2-ply but very short. Though the differences are big enough and standard error small enough (at least for me) to call it a result.

And as a side note now you could also add the possibility that opponent makes wrong takes/passes and asymmetric match equity tables and things would get really messy.   

[close]

[dorbel]

Thank you Blitz, very instructive and some surprises, for me at any rate. I'm not surprised to see that number one is a take, I just under-estimated White's wins here. I am however surprised to see that the pass is so marginal at 2-2. I would have thought that it would have been bigger than 1-1, not smaller! Clearly I need to rethink what I am doing at these scores. Incidentally, 1-1 to 5 is a very rare score for some reason; I had to search through dozens of matches before finding an example. Wonder why that is?
When leading 2-1, Black is just in the window according to Snowie, but you can't criticise blitz's marking this as "better to wait". As he says, you should get a more efficient double in a turn or two, as this position isn't very volatile. Easy take for White of course. Her window on a redouble opens at 30% and closes at 60%. Nice to be able to offer an underdog double of course, but you do often see people wait too long.