The Thorp count is based on the pip count, adjusted for the distribution of checkers. Thorp counts are only displayed during the endgame (all checkers for both players are on their respective home sides).

Thorp counts are calculated by following these steps for each player:

1. Start with the pip count.
2. Add 2 for each checker still on the board.
3. Add 1 for each checker on the 1 point.
4. Subtract 1 for each occupied home board point.
5. If the player on roll's count is greater than 30, multiply by 1.1.

You have an initial double if the Thorp difference is greater than or equal -2. You have a redouble if the difference is greater than or equal -1. You opponent has a take if the difference is less than or equal 2.

For example, a straight pip count of this example:

yields Blue:23 and Red:22, which would suggest that Blue (on roll) has a clear double, and Red a questionable take. The extreme stacking should indicate this is not a double (and a trivial take), and the Thorp count (Blue:38.5, Red:32) confirms this.

The Thorp count was originally intended for money play, and is designed to factor stacking and gapping into bearoff (or near-bearoff) races so that apparent pip count advantages aren't misevaluated.

The Thorp metric was described in THE MATHEMATICS OF GAMBLING, by Dr. Edward O. Thorp, otherwise famous for his book BEAT THE DEALER. It was popularized by Bill Robertie in ADVANCED BACKGAMMON, VOL. 2, pp. 189-195.

A rec.games.backgammon article by Chuck Bower describes how to use the Thorp count to calculate cubeless game winning chances.