Strategy · Part 4

Pip counting

The pip count is the total distance your fifteen checkers still have to travel. It decides every race cube, most run-or-stay choices, and whether that “safe” play just lost you the game. Good players count in under ten seconds — here's how.

The numbers to know

  • Starting pip count: 167 per side (rules §3).
  • An average roll moves 8.17 pips — the expected value including doubles. A normal turn is ~8 pips; doubles average 16-plus.
  • Only the difference matters in a race — that, plus who is on roll. Being on roll is worth roughly half a roll (~4 pips).

Cluster counting

Count checkers in clusters — same-point stacks multiply, mirrored formations cancel. The opening position in clusters:

ClusterArithmeticPips
5 on the midpoint5 × 1365
5 on the 6-point5 × 630
3 on the 8-point3 × 824
2 on the 24-point2 × 2448
Total167

Two speed-ups compound it: symmetry cancellation — regions where you and the opponent mirror each other contribute zero to the difference, so skip them — and reference shifts: count a messy stack as “all on the 5-point, plus 3, minus 1” rather than point by point.

The half-crossover method

Jack Kissane's tournament technique: count crossovers — quadrant boundaries each checker must still cross — instead of pips. Each crossover is worth ~3 pips (a half-crossover rounds to 1.5), giving a difference estimate accurate to 2–3 pips in a fraction of the time. Ideal for confirming “is this cube even close?” before doing an exact count.

Race formulas: turning counts into cubes

  • Thorp-style rule of thumb: with a lead of about 8% of your count + 2 pips, double; the taker should be within about 12% + 2 pips of the leader.
  • The 8-9-12 rule (Trice): in a medium race, take your own count as the base — a lead reaching +8% doubles, +9% redoubles, +12% is a pass.
  • On-roll bonus: remember the roller effectively banks half a roll; formulas above assume you count before rolling.

When the raw count lies

Two positions with equal counts can have very different winning chances. The refinements:

  • Keith count: penalizes stacked, gappy distributions before applying race thresholds — the practical standard for over-the-board race cubes.
  • EPC (effective pip count): adds expected wastage to the raw count — essential in the bear-off, where a 3-stack on the ace-point “wastes” most of every big roll.
  • Ward's formula: a refined Thorp variant for close redouble decisions.

The theme is always the same: smooth, even distributions race better than tall stacks — which is also why stacking the 6-point is a beginner leak.


Related: cube theory for what to do with the count, and the running game for playing the race itself. Boardgammon shows a live pip count at the table — practice estimating before you peek.