BGonline.org Forums

Summary of the “Isight Method”

Posted By: Taper_Mike
Date: Wednesday, 10 December 2014, at 7:47 p.m.

In Response To: Keith count fails this cube hard (Wolfgang)

The formulas below are given on page 28 of the Isight PDF.

Start by making these adjustments to the raw pip counts.

• Add 1 pip for each additional checker on the board compared to the opponent.
• Add 2 pips for each checker more than 2 on point 1.
• Add 1 pip for each checker more than 2 on point 2.
• Add 1 pip for each checker more than 3 on point 3.
• Add 1 pip for each empty space on points 4, 5, or 6 (only if the other player has a checker on his corresponding point). This penalty applies even if there are no checkers higher than the point with the “gap.”
• Add 1 pip for each additional crossover needed to get home compared to the opponent.

Next, calculate the Isight metric.

• T = adjusted count of opponent (the "trailer")
• D = T - L (difference, can be negative if T < L)
• Isight metric P = 80 - L / 3 + 2 * D

The Isight metric P purports to be the CPW (cubeless probability of winning), but it is better just to treat it as an arbitrary metric. Since it has an error of something like plus or minus 4%, it is best not to treat is as a percentage. Do not use it in match play (unless you make other adjustments first). In money games, it is very accurate.

Here are the doubling thresholds:

• P < 68: No double, take.
• 68 <= P < 70: Initial double, take
• 70 <= p <= 76: Redouble, take
• P > 76: Redouble, pass

Isight can also be used to generate an estimate of EPC, but the margin of error is too large to make it of much practical use.

Mike

Post Response

Subject:
Message: