What exactly is PR?
After a game, an engine (gnubg or XG) replays every checker play and cube decision you made, compares each to the best available play, and totals the equity you gave away. PR is that error averaged per decision and scaled ×500 by convention — so a player who gives up 0.010 equity per decision plays at PR 5.0. Because it grades decisions rather than results, PR is immune to luck: you can lose a match and post the better PR, and over enough games it is the most honest number in backgammon.
The PR scale at a glance
| PR range | Tier | Description |
|---|---|---|
| 0.0 – 2.0 | World-class | Top ~50 worldwide. Indistinguishable from neural-net play in most positions. |
| 2.0 – 5.0 | Expert | Tournament champions and top regional players. Consistent winning play. |
| 5.0 – 9.0 | Advanced | Strong club players. Confidently apply all five game plans and cube theory. |
| 9.0 – 15.0 | Intermediate | Solid hobbyists. Know the openings, basic cube play, and pip-counting basics. |
| 15.0 – 25.0 | Casual / improver | Learned with friends; understand the rules and beginning strategy. |
| > 25.0 | Beginner | Just learned the rules. Plays move by move, without a plan yet. |
Beginner (PR above 25)
Picked the game up from a relative, a café, or a holiday where the game is everywhere. Beginners race everything home, leave avoidable blots, stack the 6-point “for safety,” and touch the cube only when it's far too late.
Fastest gains: memorize the 15 opening plays; learn why the 5-point and 20-point matter; stop leaving blots within direct-shot range; start doubling at all, even imperfectly.
Intermediate / casual (PR 9–25)
Hundreds or thousands of games played. Knows the openings and the five game plans, counts pips with the cluster method, and knows the 25% take point. The leaks: cube timing (too late on big swings, too loose on takes), doomed back games, and a careless bear-off.
Fastest gains: build race-cube intuition (the formulas), study timing in holding games, and start reviewing every game's PR — the review habit alone is worth several PR points.
Advanced (PR 5–9)
Hundreds of study hours. Knows the reference positions, holds match-equity numbers for common scores, counts a position in under ten seconds, and switches plans as the dice demand. Still misplays the genuinely hard stuff: prime-vs-prime tangles and deep back-game timing.
Fastest gains: disciplined daily analysis of your own games; targeted study of specific match scores (2-away/4-away is its own subject); tournament reps under a clock.
Expert (PR 2–5)
Tournament regulars — the names on final-table brackets. Almost no outright blunders; near-optimal cube judgment; opponent-aware adjustments (cube earlier against loose takers, steer complex positions against weaker players); calm under time pressure.
World-class (PR 0–2)
The top 30–50 players on earth — a group in which Japan, Denmark, the USA, Greece, Cyprus, Iran and Turkey are all conspicuously represented. Play is indistinguishable from the engines in 95%+ of decisions, sustained through nine-hour matches. At this level the game resembles professional poker: a thin, relentlessly compounded edge over hundreds of decisions per match.
One position, five minds
The opening roll is 3-1. Watch the tiers think:
| Tier | Move | Reasoning |
|---|---|---|
| Beginner | 13/10, 6/5 | “Move forward, keep things safe.” (Loses 0.10+ equity.) |
| Casual | 13/10, 24/23 | “A builder and a split.” (Loses ~0.05.) |
| Intermediate | 8/5, 6/5 | “Make the golden point!” (Best play.) |
| Advanced | 8/5, 6/5 | “Strongest opening in the game — about +0.07 over the alternative.” |
| Expert+ | 8/5, 6/5 | “Rolled out long ago. Move without thinking.” |
At the top, the move is often the same as an intermediate's — what differs is certainty, speed, and making decisions of this quality in every position, not just the famous ones.
Track your own PR
Every game on Boardgammon — online, versus the AI, even pass-and-play — can be analyzed move by move: PR, blunder list, equity lost per play, cube errors, and a luck score that settles the “I only lost because of the dice” debate with numbers.