May 11, 2011

Playing 54, to split or not?

Last time I promised a shorter analysis, so here it is.
Today we'll discuss only one problem, also from the opening phase of the game, but it is regularly misplayed.

I encourage you to write your opinions about positions, questions about my analysis, etc.
By writing your opinion before seeing analysis you will sharpen your game; posting questions about analysis will help me to know on what should I concentrate while writing, where to put more or less effort in explanation, etc.

So here we go:



As some of you noticed, our position is reached after you played 31: 8/5 6/5 and your opponent played 55: 13/3(2).

is Player 2

score: 0
pip: 147
                         
Unlimited Game
Beaver
                          pip: 163
score: 0

is Player 1
XGID=-b---BD-B---cE---c-e--b-B-:0:0:1:54:0:0:2:0:10
to play 54


It is important to notice this position is different than sequence where your opponent had 53 to play on previous roll.

Let's look at the key differences:




First thing (1) to notice is the difference in pip count. In the left position, black has advanced for 20 pips, in the right position for only 8 pips. In the left position you're more behind in the race.
Second thing (2) to notice is the difference in spare distribution. In the right position, black has stripped 8pt; in the left position he has a spare checker on 8pt. If he attacks from 8pt, he will still retain his 8pt in the left position.
Third thing (3) to notice is ten checkers in the attacking zone with stacked 6pt screaming to get unstacked in the left position; versus only eight checkers in the zone and already unstacked 6pt in the right position.

Putting (2) and (3) together - black is eager to attack your back checkers if you split them, converting this game to a blitz.
Having (1) in mind - when behind in the race, you have more chance for a successful priming game, trying to contain black's back checkers.

The play that both doesn't give your opponent chance to attack and raises your priming potential is 13/8 13/9 and it is correct by a wide margin:
1.3-ply13/9 13/8eq: +0,168
Player:
Opponent:
54,38% (G:16,91% B:0,74%)
45,62% (G:13,74% B:0,80%)
2.3-ply24/20 13/8eq: +0,083 (-0,084)
Player:
Opponent:
53,05% (G:14,98% B:0,77%)
46,95% (G:15,50% B:0,49%)
3.3-ply13/4eq: +0,046 (-0,122)
Player:
Opponent:
51,75% (G:15,48% B:0,65%)
48,25% (G:15,49% B:1,06%)
4.3-ply24/15eq: +0,024 (-0,143)
Player:
Opponent:
52,36% (G:13,41% B:0,67%)
47,64% (G:17,07% B:0,51%)


In the right position, splitting with 24/20 13/8 is correct. You're no longer under such attacking threat and you're not behind in the race, so making an advanced anchor on 20pt should be one of your game plans.

When your opponent uses his spare checker from 8pt to make home board point on his first roll, it is a right time to split, while his 8pt is stripped (without spare checkers).
When he makes a home board point with 55 or 44 and brings new attackers from the midpoint while keeping the spare checker on the 8pt - splitting is too dangerous and you should avoid it.




This week I had a question "How should I know when to double?" so next couple of posts will be about doubling - we'll start from some simple positions covering doubling basics and we'll continue to some more complex positions.

The next time we'll discuss this position:
is Player 2

score: 0
pip: 111
                         
Unlimited Game
Beaver
                          pip: 100
score: 0

is Player 1
XGID=----BCD-CB-Ac---ab-dcb----:0:0:1:00:0:0:2:0:10
on roll, cube action?

Would you double here? Would you take as black? How sure are you in your decisions?

For more experienced players: You're leading 1-0 in a 3pt match. Would you double this and would you take this as black? How sure are you in your decisions?

(Please say in a comment if you're answering to the main question or the question for more experienced players. Or you can try to answer both questions)


13 comments:

  1. 1a) I would double as a white
    1b) I wouldn't take doubling as a black
    2a) I wouldn't double, but would accept doubling from black
    2b) I wouldn't take it as a black
    2c) 1a) 80% 1b) 70% 2a) 60% 2b) 70%

    ReplyDelete
  2. Money play:

    Using the Trice count, it's double/take. Black's point-of-last-take in computed as 10% of white's pipcount + 2 pips = 10+2 = 12. Black's point-of-last-take is exactly what it sounds like - if he's more than 12 pips behind he should drop, and if he's within 12 pips he should take. White should double if he's within 3 pips of the point-of-last-take and redouble if within 2. Here, White's 11 pip lead is square in the middle so it's double/take.

    Note that the Trice count is an approximation - you have to take wastage and gaps into account and there are other more sophisticated methods out there that account for these differences, but the Trice count is the simplest and easiest way to arrive at a doubling decision in non-contact positions.

    I'm 100% sure this is a double for money. I'm not completely sure about the take, since black has three more crossovers to make than white, but I'll still say take with about a 70% confidence level.

    AtS (2-away 3-away)

    Here, if white doubles and black takes, black has a mandatory redouble and we're playing for the match. Since doubling here essentially throws away White's lead, White should be careful about doubling. I often get these wrong, and err on the side of holding. Here, I think the double is correct, because of the extra crossovers. Without those extra crossovers tipping the balance I'd hold.

    What about the take decision? Black is almost at the point of last take for money, which is somewhere around 22% GWC (deeper than the oft-cited 25% because of cube vig) If Black passes, then he'll need to win the next two games in a row, and assuming they players are at equal strength that gives him 25% MWC. So it looks like passing gives him more MWC than taking.

    double/pass but much less certain of this than for money.

    ReplyDelete
  3. 1a, I would cube as white.
    1b, I would pass as black.
    2a, I would not cube as white
    2b, I would probably take and redouble as black
    I am not sure in any of the situations..
    traxxy

    ReplyDelete
  4. 1a double - fairly sure
    1b no take - fairly sure
    2a double - no clue so i dont change anything
    2b no take - no clue so i dont change anything

    The method I use is the one where you add 2 pips for each checker more than 1 on 1pt, 1 pip for each checker more than 1 on 2pt, 1 pip for each checker more than 3 on 3pt, 1 pip for each empty space on 4,5,6, and then add 1/7th (rounded down) for player on roll. in this case, most of that doesn't apply, just the added 1/7th which gives us 111 to 114.

    then the rule is: double if my count exceeds opponents by no more than 4. (it exceeds by 3 so i double). redouble if my count exceed opponent's by no more than 3 (not applicable). take if doubler's count exceeds mine by at least 2 (it does).

    ReplyDelete
  5. @pthalo - that method is called "Keith count" and it is very good in medium-long races (bearing in, like here) with some wastage on low points.

    If Keith count gives doubler 3 more pips - it is still a take (as you noticed), so why did you choose to pass this? :)

    ReplyDelete
  6. There were some problems with blogger service yesterday.
    Unfortunately, some very good comments are deleted. I hope blogger will return them.

    ReplyDelete
  7. Here is the explanation of the problem:
    http://buzz.blogger.com/2011/05/blogger-is-back.html

    It seems we just have to wait and those deleted comments will return during the weekend.

    ReplyDelete
  8. Main question:

    Double/drop. I have a lower pip count and am fewer rolls from starting to bear-off than black is. I think of this as a gambling position: Dice rolling is much more important than checker play from now on.

    I'm never very sure about doubling decisions, but I'm pretty confident about this one.

    On the other hand, (because of your second question) this could be a double/redouble. I'm not "more experienced" but it is obvious, that if white doubles at 2away 3away, then THIS is the issue. So, at 2away 3away I wouldn't double, because black is then just one 66 from winning the match.

    ReplyDelete
  9. Part 1. Double/take. The home boards have the same distributions. White takes three possibly four rolls to get in and start to bear off, black takes four or five rolls to get in and start to bear off. close enough for just one set of doubles to put black back in the driving seat.

    Part 2. No double/take. If white passes the cube over and black throws a joker double four, five or six he will close enough to re-double white out of the game if white throws badly, its then the crawford game and white has to win two games in a row to win the match. If he keeps calm and tries to win the match without the cube HE will then be in a crawford game in which black needs to win two games in a row.
    If black gets any sets of big doubles and turns the cube white can drop and will only lose the one point, equalising the match.

    ReplyDelete
  10. As comments are not returned automatically, I will copy-paste them:

    Bojan:

    1a) I would double as a white
    1b) I wouldn't take doubling as a black
    2a) I wouldn't double, but would accept doubling from black
    2b) I wouldn't take it as a black
    2c) 1a) 80% 1b) 70% 2a) 60% 2b) 70%

    ReplyDelete
  11. ah_clem

    Money play:

    Using the Trice count, it's double/take. Black's point-of-last-take in computed as 10% of white's pipcount + 2 pips = 10+2 = 12. Black's point-of-last-take is exactly what it sounds like - if he's more than 12 pips behind he should drop, and if he's within 12 pips he should take. White should double if he's within 3 pips of the point-of-last-take and redouble if within 2. Here, White's 11 pip lead is square in the middle so it's double/take.

    Note that the Trice count is an approximation - you have to take wastage and gaps into account and there are other more sophisticated methods out there that account for these differences, but the Trice count is the simplest and easiest way to arrive at a doubling decision in non-contact positions.

    I'm 100% sure this is a double for money. I'm not completely sure about the take, since black has three more crossovers to make than white, but I'll still say take with about a 70% confidence level.

    AtS (2-away 3-away)

    Here, if white doubles and black takes, black has a mandatory redouble and we're playing for the match. Since doubling here essentially throws away White's lead, White should be careful about doubling. I often get these wrong, and err on the side of holding. Here, I think the double is correct, because of the extra crossovers. Without those extra crossovers tipping the balance I'd hold.

    What about the take decision? Black is almost at the point of last take for money, which is somewhere around 22% GWC (deeper than the oft-cited 25% because of cube vig) If Black passes, then he'll need to win the next two games in a row, and assuming they players are at equal strength that gives him 25% MWC. So it looks like passing gives him more MWC than taking.

    double/pass but much less certain of this than for money.

    ReplyDelete
  12. pthalo

    1a double - fairly sure
    1b no take - fairly sure
    2a double - no clue so i dont change anything
    2b no take - no clue so i dont change anything

    The method I use is the one where you add 2 pips for each checker more than 1 on 1pt, 1 pip for each checker more than 1 on 2pt, 1 pip for each checker more than 3 on 3pt, 1 pip for each empty space on 4,5,6, and then add 1/7th (rounded down) for player on roll. in this case, most of that doesn't apply, just the added 1/7th which gives us 111 to 114.

    then the rule is: double if my count exceeds opponents by no more than 4. (it exceeds by 3 so i double). redouble if my count exceed opponent's by no more than 3 (not applicable). take if doubler's count exceeds mine by at least 2 (it does).

    ReplyDelete
  13. traxxy

    1a, I would cube as white.
    1b, I would pass as black.
    2a, I would not cube as white
    2b, I would probably take and redouble as black
    I am not sure in any of the situations..

    ReplyDelete