If you know one of the good counting formulas, you can do a calculation over the board and get a pretty good answer. The best formulas right now are the Keith Count and the ISight Count. I use the Keith Count, so I'll describe how the process works.

The Keith Count works for money games, so we'll start by pretending it's a money game and then adjusting at the end for the match score.

First we calculate a Keith number for the side on roll. We take the pip count (39) and perform a series of corrections.

1) Add 2 for each checker more than 1 on the ace point. Doesn't apply here.

2) Add 1 for each checker more than 1 on the 2-point. Applies, so we add 1 and we're at 40.

3) Add 1 for each checker more than 3 on the 3-point. Doesn't apply.

4) Add 1 for each gap on the 4, 5, and 6-points. Doesn't apply.

5) Add 1/7 of the pip count and round down. Pip count is still 39, so we divide by 7 and throw away the remainder, giving us 5. Adding that to 40 gives us 45, our adjusted pip count for the player on roll.

We repeat the exercise for the opponent, but we don't add 1/7 of the pip count. None of the adjustments apply to White, so his adjusted pip count is still just his pip count, 39.

Now we subtract the player not on roll from the player on roll, which is 45 - 39, or 6. This is the Keith Number for the player on roll.

We compare that number to the following table:

Keith Number = 5 or greater, no double and take.

Keith Number = 4, initial double only and take.

Keith Number = 2 or 3, redouble and take.

Keith Number = 1 or less, redouble and drop.

So for money we get a solid no double and take. Now we look at the match score.

Black trails 5-6 to 11, so he should be more eager to double or redouble than for money. But the Keith Number puts him pretty far away from a redouble for money, so over the board I would guesstimate and wait on a cube here.

What does XG say? Well, it agrees on the no redouble for money by a big margin (by 0.065), and its XGR++ analysis says no redouble at this match score. But I did a long rollout which indicated that redoubling at the match score was narrowly correct (by 0.01).

In short, clear no redouble for money, and at the match score whatever you do won't be a serious error.

I agree, I was quite sure no redouble for money. The match score is what made the situation interesting. I also did a rollout and found very slight redouble. Thanks for carefully presenting Keith count.

I do very much the same but with the iSight but make a little adjustment to make it simpler at the end.

The iSight count goes like this:
Add 1 for each additional checker your opponent has off more than you.
Add 2 for each checker more than 2 on 1-point.
Add 1 for each checker more than 2 on 2-point.
Add 1 for each checker more than 3 on 3-point.
Add 1 for each empty space on 4,5 & 6 points if opponent has that point made.
Add 1 for each crossover more than the opponent.

In this case there are no adjustments to make on either side.

For the correct cube action in a cash game I follow Bill Robertie's advice in 501 Problems: If the pip count is more than 30 follow the 8-9-12 rule. As the score is even it's no double/take (I might have paraphrased here).

In this match at 5away-6away the take point is 23% which is about the same as for cash so I know this must also be no double/take.

iSight does have a formula to workout the cube action but it is only really doing the 8-9-12 rule.

Don't use the obsolete 8-9-12 rule. Nack 58 is a much better modern alternative. The more difficult question is how to adjust the pip counts before you use a formula.

Could you explain what the Nack 58 count is? I've never heard of it.

Sure, Bill. Once you have adjusted pip counts (using your favorite adjustments), the formula gives proper cube action.

If the leader's pip count is 57 or less, than use Trice's "Rule 62" from Backgammon Bootcamp: subtract 5 from the leader's pipcount, divide by 7, and round down. This gives the point of last take when added to the leader's pipcount.

If the leader's pipcount is 58 or more, then subtract 32, multiply by 2, and find the closest integral square root of that number. Once again, it gives the point of last take.

Here is the link to Nack explaining this:

http://www.bgonline.org/forums/webbb...mes;read=87795

Bill - Is the Keith Count still your ‘Adjusted Count’ of choice ?

Yes. It's very good and easier to use than some of the others, which matters when you're playing a tournament match and the clock is running. For races with some checkers in the outer boards I use the 8-9-12 Rule, again because it's simple, easy, and highly accurate.

Bill

Until what point in the Bear Off do you continue to use the Keith Count ? Is there
a point at which you switch to just manual rolls estimation or another system ?



Regards


Hagadol

Until what point in the Bear Off do you continue to use the Keith Count (or other
adjusteded pip count mehtod) ?

Is there a point at which you switch to just manual rolls estimation or another system, as the checkers still to be taken off decreases ?

Interested to hear thoughts on this.

I use the Keith Count as long as there are more than 5 or 6 checkers and they're spread out across the inner board.

Once the checkers are jammed on the low points I'll try to estimate the number of rolls left and go from there.