Backgammon Strategy: The Complete Guide (2026)

A complete guide to backgammon strategy — opening moves, the doubling cube, racing vs. holding games, bearing off, match play, common mistakes, and how to actually improve. Reference-grade, 2026.

Backgammon Strategy: The Complete Guide

Backgammon is older than chess and rewards more dimensions of skill. A serious player must understand opening theory, mid-game position types, the doubling cube, race calculations, the bear-off, match-play scores, and how variance from the dice shapes every decision. This guide is the reference — long, organised by game phase, and linked through to the deep-dive posts for each topic. If you read it end to end, you will understand the strategic spine of the game; if you skim, the table of contents below points you to the right section.

It is also the foundation document for 6proclub's backgammon content. Every other backgammon piece on this site cross-links back here.

Table of contents

1. The four phases of a game
2. Opening moves — the 21 rolls that decide the first turn
3. Position types — race, prime, blitz, back game, holding game
4. The doubling cube — when to offer, take, drop
5. Pip counts and racing calculations
6. Bearing off — the endgame that wins more games than skill in the middle
7. Match play and the Crawford rule
8. Variance — why the better player still loses three games in ten
9. Common mistakes and how to fix them
10. ELO ratings and what your number actually means
11. Where to go next

1. The four phases of a game

Every backgammon game proceeds through four roughly recognisable phases, even if the boundaries blur:

Opening (turns 1–4). Both players are racing their back checkers home while trying to build a prime or take key inner-board points. The opening is the most theoretically studied portion of the game; the 21 possible first-turn rolls each have a known-best response.

Middle game (turns 5–15). Position type crystallises. The position is now a race, a prime, a blitz, a back game, or a holding game. Each calls for a different strategic approach. Most of the cube decisions happen here.

Race (turns 15–25). Contact is broken. The game becomes pure pip arithmetic; the player ahead in pips, with both sides off contact, is almost certainly winning unless wastage on the bear-off swings the calculation.

Bear-off (final 6–10 rolls). Both players are clearing checkers from the home board. The arithmetic is simple in principle but easy to misplay; the bear-off is where a winnable race is often thrown away.

A strong player thinks about which phase they're in and what the phase rewards. A weak player makes the same kind of move regardless.

2. Opening moves — the 21 rolls that decide the first turn

There are 36 possible dice combinations, but with order ignored (a 3-1 is the same opening problem as a 1-3) you get 21 distinct opening rolls. Modern theory — based on hundreds of millions of bot rollouts from XG and Gnu Backgammon — gives a clear best move for almost every one.

A few are textbook. 3-1 → 8/5 6/5 (make the 5-point, the most valuable point on the board). 6-1 → 13/7 8/7 (make the bar, a strong home-board anchor). 4-2 → 8/4 6/4 (make the 4-point, second only to the 5). 6-5 → 24/13 (run the back checker; this is the only roll that puts a single back checker safely on its midpoint).

A few are theoretically contested. 2-1, 4-1, 5-1 are the "splitting" rolls — modern bots prefer splitting the back checkers (24/23 + 13/11 type plays) over the older "slot the 5 and hope" approach.

For the complete table, see Backgammon Opening Moves: The 21 Best Plays for Every Roll. Memorising the table is worth ~50 rating points by itself — you stop making genuine opening mistakes, and the rest of your skill compounds from a sound start.

3. Position types — race, prime, blitz, back game, holding game

Backgammon positions cluster into recognisable patterns. Each pattern has a strategy that's known to win on average; the skill is recognising which pattern you're in and committing to the right plan instead of drifting.

Race. No contact between the two armies. The only variable is pip arithmetic. The player ahead in pips wins almost every time. Strategy: double aggressively when you're 8+ pips ahead, take cubes when you're within 5% of even.

Prime. You have four to six consecutive points blocking your opponent's escape. Strategy: keep the prime intact, extend it where possible, and let it grind your opponent into submission.

Blitz. You're hitting your opponent's blots aggressively while building a closed home board. Strategy: hit, cover, repeat — even at the cost of leaving secondary blots, because every check on the bar is a tempo lost for your opponent.

Back game. Your back checkers are trapped behind your opponent's prime, but you hold two or three anchor points in their home board. Strategy: wait. Time the position so when your anchor breaks, your opponent has already started to clear their inner board, leaving you a hit-and-cover late.

Holding game. You hold one anchor in your opponent's home board (usually 20- or 21-point) without backgame depth. Strategy: hold the anchor until your opponent races past, then accept the race or hit if a blot appears in front of the anchor.

A surprising amount of intermediate-level losing comes from playing the wrong type for the position — a backgame timing miss, a holding-game player who hangs on too long, a prime player who breaks the prime to chase a hit. Knowing which type you're in is half the work.

4. The doubling cube — when to offer, take, drop

The doubling cube is what makes backgammon a strategy game rather than a dice race. Without the cube, the player with a slight edge would win 51-49; with the cube, an aggressive player can leverage a winning position into a 2× or 4× swing while a passing player gets pressured into bad-equity decisions.

Offer (double) when: you're winning by enough that your opponent will fold ("drop"), but not so much that they fold and you give up future equity. The textbook trigger is around 65–70% winning chances — high enough that a sensible opponent drops, but low enough that the opponent who takes is making an equity mistake.

Take when: the cube is offered and your winning chances are above 25%. Below 25%, drop and lose the current cube value. Above 25%, take — even if you're a small underdog — because the doubled cube wins you 2× when you do win.

Beaver, raccoon, and the Jacoby rule — special cases for live cash play that 6proclub games typically don't use; standard online play turns off Beavers.

For the full treatment, see The Doubling Cube: When to Offer, When to Take, When to Drop. It's the single most rating-positive topic to study after opening theory.

5. Pip counts and racing calculations

In a race, the player ahead in pips wins. The only question is by how much, and how to estimate at the table without stopping the game.

A fast pip count: count both sides' checkers in each home position (1 through 24 from the player's perspective), multiply by the position number, sum. With practice, this takes about 12 seconds per side.

The doubling window in a race is well-defined: with N pips left for the leader, the cube goes from "no double" to "double, take" to "double, pass" in known ranges based on lead percentage. For pure races, see the bear-off tables in Bearing Off in Backgammon.

The other tool every player needs is the Match Equity calculator. It tells you, at any match score, what each side's win probability is for the match (not just the current game). The cube decision in a match depends on match equity, not just game equity — see Section 7 below.

6. Bearing off — the endgame that wins more games than skill in the middle

The bear-off is where good positions become wins and bad bear-offs become losses. Two principles:

Avoid wastage. A 5 rolled when your highest checker is on the 3-point bears off a checker but "wastes" 2 pips. Distribute checkers across the high points to minimise wasted pips, especially with one or two checkers left.

Watch for blot exposures during the bear-off. If your opponent still has a checker behind you, every roll that leaves a blot on a point your opponent can hit costs you the game on average. Bear off the safe way (highest point first) when contact is still possible; bear off optimally (no wastage) once contact is broken.

Full bear-off mechanics in Bearing Off in Backgammon: Endgame Strategy That Wins More Games.

7. Match play and the Crawford rule

Cash backgammon (single games for money) and match play (first to N points) play differently. The cube decisions are different, the opening is different, and certain positions that are clear takes in cash are clear drops in a tournament match.

Match equity is the probability of winning the match from any score. At 4-4 in a 7-point match, you're at ~50%. At 4-2 leader, ~66%. At 6-0 leader (Crawford rule applies — see below), ~94%.

The Crawford rule. In a match, the game immediately after the leader reaches match score - 1 (one short of victory) is played WITHOUT the doubling cube. This prevents the trailer from doubling on every game and freerolling for the win. After the Crawford game, the cube is back in play for the trailer to use aggressively.

Cube decisions in matches. When you're behind in the match, you want to double aggressively (high variance is your friend; you need to make up score). When you're ahead, you take more conservatively (low variance preserves your lead). The standard cube triggers — 70% to double, 25% to take — apply in cash, not in matches.

The complete match-play treatment is in Match Play and the Crawford Rule.

8. Variance — why the better player still loses three games in ten

A 200-point ELO difference in backgammon translates to roughly a 70-30 single-game edge. The stronger player wins 70% of single games — meaning the weaker player still wins 30% of them. Over a 7-point match, the same edge becomes around 80-20; over a 21-point match, around 90-10.

Three takeaways from this.

1. One game means very little. A loss against a weaker opponent is statistically expected sometimes. Don't tilt off the result of a single short game.
2. Match length amplifies skill. If you want to test skill, play longer matches. 21-point matches are dramatically more skill-sensitive than 3-pointers.
3. Variance is symmetric. It cuts your wins and your losses both ways. A bad streak of 5 losses against a weaker player happens, and so does a hot streak of 5 wins against a stronger one. The math says zero about either streak — only the long-run average matters.

A deeper treatment of these numbers, including the exact ELO-to-win-probability tables, is in Why the Better Player Still Loses Three Games in Ten — Variance in Backgammon.

9. Common mistakes and how to fix them

Five mistakes account for most of the gap between intermediate and advanced play.

1. Slotting too aggressively in the opening. Older opening theory loved slotting the 5-point with rolls like 3-1 actually-no, splitting plays. Modern bot rollouts say the safe play wins more equity over a thousand games. Bias toward splitting on small rolls.

2. Failing to recognise position type. A back-game player who plays the holding game's safe strategy loses. A prime player who breaks the prime to chase a hit loses. Stop and ask: what kind of position is this, and what's the canonical strategy?

3. Misplaying the cube. Many intermediate players never double, treating the cube as a punish-mechanism only. The cube is a value-extraction tool: every winning position that you don't double on is equity you left on the table.

4. Taking cubes that should be dropped. The 25% rule is hard to internalise. If your gut feels "I might win this", you're probably below 25% and should drop.

5. Bear-off wastage. Last-minute pip wastage is the single most expensive endgame mistake. With 2 or 3 checkers left, distribute across high points before bearing off, even if the move feels backwards.

Detailed treatment in Five Common Backgammon Mistakes (And How to Stop Making Them).

10. ELO ratings and what your number actually means

A backgammon ELO is a comparative skill rating. ~1200 is a beginner, ~1600 is an intermediate club player, ~1800 is a competent online player, 2000+ is a serious tournament player, 2200+ is approaching world-class.

ELO comparisons rely on a long sample — single results barely move a rating. A 1800 player losing one 7-point match to a 1900 player will lose maybe 8–12 rating points. The same 1800 player going 0-10 against the same opponent would lose around 80 — and would also indicate the original 1800 was overrated, not that they "got unlucky 10 times in a row."

If you want to improve your rating, play longer matches (more skill-sensitive), study opening theory (cheapest 50 rating points available), and run the position-type recognition drill (catch yourself once per game asking "what type is this?"). Three months of those three habits will move most players up 100–200 points.

The mechanics of how ELO updates after each game are in Backgammon ELO, Demystified.

11. Where to go next

Backgammon is the kind of game where reading is necessary but not sufficient. After this guide:

• Drill openings. Backgammon Opening Moves — memorise the 21 best plays.
• Drill the cube. The Doubling Cube, Demystified — internalise the 25% take rule and the 65% double trigger.
• Drill the bear-off. Bearing Off in Backgammon — practice safe vs optimal bear-off decisions.
• Then play. 6proclub backgammon — provably fair dice, real opponents, every roll cryptographically verifiable. The match length you play matters more than the platform; 11-point and 15-point matches are the sweet spot for serious play.

Two related pillar reads for the broader context:

• Provably Fair Gambling: The Complete Guide — why the dice on 6proclub are cryptographically verifiable.
• Backgammon vs Chess — why backgammon rewards more dimensions of skill than chess does.

In one paragraph

Backgammon strategy reduces to recognising your position type, playing the canonical opening for your roll, holding the cube as a tactical weapon rather than a punishment, and treating the bear-off as the high-skill phase it actually is. Variance from the dice is real and symmetric — better players still lose three games in ten — so match length matters far more than any single result. Pair the strategy with cryptographically verifiable dice (the provably fair guide covers the math) and you have everything you need to play the game seriously. 6proclub built the platform around exactly this premise.