The doubling cube is where most of the skill in backgammon lives, and where most casual players give away the most equity. The common error is not doubling too often. It is doubling far too late.
Doubling offers your opponent a choice: concede one point now, or play on for twice the stakes with the cube on their side. Both halves matter. The second half, handing them the cube, is a real cost, because they can later double you back.
There is a point where your position is so strong that doubling is a mistake, because your opponent simply passes and you collect one point when you were on your way to winning two with a gammon. Recognising 'too good' is an advanced skill, and it is the opposite problem from the one most players have.
The right moment sits in a window: strong enough that your opponent is close to being unable to accept, but not yet so strong that they will certainly refuse. Doubling early inside that window costs little; doubling after it has closed costs a great deal, because a position that was worth a double last turn may now be worth nothing if the dice turned.
Here is the whole take decision in four lines of arithmetic, for money play with the cube dead and gammons ignored. Pass and you lose exactly 1 point. Take and the stake is 2, so you win 2 with probability p and lose 2 with probability 1 minus p, which is an expected 4p minus 2. Taking beats passing when 4p minus 2 is greater than minus 1, which is p greater than 0.25. So the bare take point is 25 percent: if you win the game more than a quarter of the time, taking is correct even though you are losing.
That derivation assumes the cube is dead after the take, which it is not. Once you take, you own the cube, and the option to double back later is worth something. That recube vantage pushes the practical take point below 25 percent in most positions, often into the low twenties. Gammon risk pushes it the other way: if a meaningful share of your losses will be gammons, you are risking more than 2 points to take, so you need more winning chances to justify it. Those two adjustments are why published take points vary by position rather than sitting on one number, and why 25 percent is the anchor rather than the answer.
Turn the take point around and it tells you when to double. If your opponent should take at roughly a quarter, then the moment to double is when their chances have fallen near that line, not when they have collapsed to nothing. A double that gets taken is not a failure. A double that gets passed by a mile means you waited so long that you converted a two-point game into a one-point game.
Backgammon positions swing fast. The player who waits for certainty before doubling is systematically doubling in positions their opponent must pass, which means they win one point instead of two, over and over. The cost is invisible, which is why it persists.
Double when you are winning.
Double when you are winning enough that your opponent is nearly forced to give up, and not later. 'Winning' is far too vague to act on, and the instinct to wait until victory feels safe is exactly the error that costs the most. If your opponent takes every double you offer, you are almost certainly doubling too late rather than too early.
Play a rated game and every decision gets reviewed afterwards against the best move available. The dice are committed before you roll and checkable after, so the number you end up with is a measure of you, not of luck.
Play a rated game freeWhen your position is strong enough that your opponent is close to being unable to accept, but before it becomes so strong that they will certainly pass. Most players double far too late rather than too early.
A position so strong that doubling would make your opponent pass, when playing on would likely win a gammon worth two points. In that case doubling costs you a point.
© 2026 Backgammon Battles