Bloody Mary "Sunday Special" finals position

[don]

In case the .png file doesn't show up here,

    GNU Backgammon  Position ID: c9sEAyA2O4wHAA
                    Match ID   : UQmgACAACAAA
    +24-23-22-21-20-19------18-17-16-15-14-13-+  O: adrian
    | O     O  O  O  O |   |    O           X |  2 points
    | O     O  O  O  O |   |                X |  
    |       O          |   |                X |  
    |                  |   |                X |  
    |                  |   |                  |  
    |                  |BAR|                  |v 5 point match
    |                  |   |                  |  
    |                  |   |                  |  
    |                X |   |                  |  
    |    X  X     X  X |   |          X     O |  On roll
    | O  X  X     X  X |   |          X     O |  1 point
    +-1--2--3--4--5--6-------7--8--9-10-11-12-+  X: WildThing (Cube: 2)

This occurred during today's tourney finals, and here's the whispers:

scrat whispers: take
PersianLord whispers: sure
PersianLord whispers: time for redouble
scrat whispers: exactly
Sixta_II whispers: redouble?
spielberg whispers: tricky - she only gets half the added value
> You whisper: yea
spielberg whispers: oops - stupid me - she gets perfect value - yes redouble
PersianLord whispers: what do u mean, steve?
> You whisper: I'd say no redouble till gammon is out of question
PersianLord whispers: gammon is out
> You whisper: no it isn't, it's like 5-10 percent
> You whisper: and no redouble now!
PersianLord whispers: sure
> You whisper: that's the other reason for no redouble, jokers
PersianLord whispers: not redoubling was a bad mistake
> You whisper: i disagree
> You whisper: mebbe I'll ask gnu later
PersianLord whispers: we can wager a bet, don
Sixta_II whispers: no mistake

Here's gnu's opinion:

Spoiler

Cube analysis
2-ply cubeless equity  +0.281 (Money:  +0.222)
    Win  W(g) W(bg) -  Lose  L(g) L(bg)
  0.610 0.055 0.001 - 0.390 0.052 0.001

Cubeful equities:
1. Double, pass         +1.000
2. Double, take         +1.076  ( +0.076)
3. No double            +0.851  ( -0.149)
Proper cube action: Redouble, pass

[close]

[PersianLord]

Spoiler

Many thanks for posting this nice position, don  and be happy we didn't wager a bet 

I think even adrian's double to 2 earlier in the same game was a bit pre-mature as he hadn't a solid position. Another interesting point is that you were exactly ture in estimating the gammon chances for WildThing to 5-10%. 

But a thing I may need more explanation is this statement of yours:

don whisper: that's the other reason for no redouble, jokers

I'm relatively a novice in studying the western backgammon, but I'd like to modify your words into :

"that's  exactly the very reason to redouble, avoiding jokers" !

I mean when WT had a clear re-double for match, adrian would have definitly passed as he's not a defiant beginner. So WT would have gained 2 valuable pts and there would be no joker threat for him. Even if WT doubled and adrian took, still WT did the right thing, as BG is the game of probabilities and you just can't rely your plans on your OR your opp's possible jokers.

Againt thanks for the good work.

[close]

[don]

Yeah, but notice there is still a gammon chance, so we'd come out even on the bet!

[blitzxz]

Spoiler

What a surprising result! I was 100 % sure that this was no double. And actually it is big no double... in money game or if the score is for example 0-0 (double would be huge blunder then). If the score is 1-1 then it is double, take. But in 2-1 it truly turns out to be weird looking double, pass. White is still leeding the race, he might easily escape or hit as opponent will propable leave indirect shots. White has just under 40 % chances to win... and pass?? Yes, in this match score the cash point for redouble is very low as white will have just over 40 % chances in score 3-away 2-away if he will pass.

[close]

[PersianLord]

Let's review this position once more, this time with more attention on theory.

Pip-counts : orange 110 - 99 white
Score :        orange  1-2       white
Match:         5-pointer

Consulting Kit's eauity table, the MWCs are as follows: (4 away- 3 away)

orange's MWC: 41%
white's   MWC: 59%

Now :

1- If orange doubles and wins, his MWC will be 100%
2- If orange doesn't double and wins, his MWC will be 59%
3- If orange doubles and loses, his MWC will be 0%
4- If orange doesn't double and loses, his MWC will be 17%

Risk = 17%

Gain = 41%

Take point = 17/(17+41) = 29%

Orange's GWC is 61% > 29%

So far it's a clear double for orange.

Now for white:

1- If he takes and wins, his MWC will be 100%
2- If he  passes, his MWC will be 41%
3- If he takes and loses, his MWC will be 0%


Risk = 41%

Gain = 59%

Take point = 41/(41+59) = 41%

White's GWC is 39% < 41%

So it's a clear double/pass as the gain/loss equity is heavily favores orange. So our friend who has voted fot double/take is very wrong. But those who have voted for "no double" might have probably thought like don : it's too good to double as orange may win a gammon.

But as you see, gammon win chances for orange is just 5.5%. In the rest of possibilties, orange will win 55% of the times, but will lose 39% of the times that would be very costly. So again, the gain/loss is heavily favored for doubling now. In GNUBG's words, not doubling is a "bad mistake" : a whopping equity loss of 0.149

[blitzxz]

White's GWC is 39% < 41%

So it's a clear double/pass as the gain/loss equity is heavily favores orange.

2 procent difference in winning chances is very small to me. Even gnu makes that kind of errors time to time. If you can evaluate game winning chances that well you must be expert player. And actually white will have 40 % if he will pass (3-away 2-away not 4-away 3-away) and quick rollout gives only 37 % game winning chances (that is just that 2 procent error but to the other way this time).

[PersianLord]

2 procent difference in winning chances is very small to me. Even gnu makes that kind of errors time to time. If you can evaluate game winning chances that well you must be expert player. And actually white will have 40 % if he will pass (3-away 2-away not 4-away 3-away) and quick rollout gives only 37 % game winning chances (that is just that 2 procent error but to the other way this time).

I think we don't need to calculate the GWC exactly to decide whether it's a take or not, we just need to take a look at risk/gain to conclude. Here orange risks 17% of MWC to gain 41%, whille white risks 41% to gain 59%. The conclsuion is quite clear I think.

[spielberg]

So it's a clear double/pass as the gain/loss equity is heavily favores orange. So our friend who has voted fot double/take is very wrong.

It was me who voted for double/take - wrong I admit. I don't think that's "very wrong" tho' - at a cost of 2% MWC it's considerably less wrong than no double which costs 15%.

[PersianLord]

It was me who voted for double/take - wrong I admit. I don't think that's "very wrong" tho' - at a cost of 2% MWC it's considerably less wrong than no double which costs 15%.

you're right

[blitzxz]

I think we don't need to calculate the GWC exactly to decide whether it's a take or not, we just need to take a look at risk/gain to conclude. Here orange risks 17% of MWC to gain 41%, whille white risks 41% to gain 59%. The conclsuion is quite clear I think.

There seems to be some confusion. This not the correct way. Take point calculation is only for take point. Let's not mix double decision and take decision. So here white can choose play for the match in this game or pass and continue in 3-away 2-away down. According to woolsey table in that score white would have 40 % chances. So if white has bigger chances to win this game (and the match) he should take. And white should know the exact game winning chances to make the desicion. It turns that white has couple procents less so take is blunder. But evaluating chances with that accuracy is still hard to do.

Now doubling decision is another thing. There is no easy way to calculate that. What doubler wants do is to be as close as possible to opponents take point so he can practically win the game with the cube and get maximum effectivity for double. Here orange has already missed the point so question is should he cash or go for gammon. The risk for orange is huge. He could be in sure 3-2 lead or the game could easily turnaround complitely with only one roll and he would end up in 1-4 crawford. The risk is just too much for just 5 % gammons.