Liar's Dice strategy
Bluffing is the part people remember, but the players who win do the arithmetic first and act second. Every number on this page is computed at build time by the same code the game server runs — including the two thresholds the house bots themselves obey.
One number runs the whole game
With ones wild, a die helps a bid on two of its six faces: the face that was named, or a one. So every die you cannot see backs the standing bid one time in three. That single fraction is the spine of every decision at the table. Multiply the dice in play by a third and you have the count to expect before anyone has said a word.
At a full table of six that is 30 dice and an expected 10 behind any face you care to name. Learn the row for your table size and you already hold most of what separates a good player from a hopeful one.
| Players | Dice in play | Expected | Safe bid | Even money | Call at |
|---|---|---|---|---|---|
| 2 | 10 | 3.3 | 2 (90%) | 3 (70%) | 5 (21%) |
| 3 | 15 | 5.0 | 3 (92%) | 5 (60%) | 7 (20%) |
| 4 | 20 | 6.7 | 5 (85%) | 7 (52%) | 9 (19%) |
| 5 | 25 | 8.3 | 6 (89%) | 8 (63%) | 11 (18%) |
| 6 | 30 | 10.0 | 8 (83%) | 10 (57%) | 12 (28%) |
Safe bid is the highest quantity still true four times in five — you can say it and be caught only rarely. Even money is where a bid becomes a coin flip. Call at is where a bid's chance of being true falls under 28%, which is not a round number we chose: it is the constant the house bots call on. Bid below their line and they believe you. Cross it and they will take the count.
Your own dice are the only thing you know
The table's expectation is a starting point, not an answer, because you can see 5 of the dice and nobody else can. Those 5 are certainty; the rest is binomial. Every matching die in your cup shifts the whole distribution up by one and takes a chunk of risk out of a bid that would otherwise be reckless.
| You hold | Bid 8 | Bid 10 | Bid 11 | Bid 12 | Bid 13 | Bid 14 |
|---|---|---|---|---|---|---|
| 0 | 63% | 30% | 18% | 9% | 4% | 2% |
| 1 | 78% | 46% | 30% | 18% | 9% | 4% |
| 2 | 89% | 63% | 46% | 30% | 18% | 9% |
| 3 | 95% | 78% | 63% | 46% | 30% | 18% |
| 4 | 99% | 89% | 78% | 63% | 46% | 30% |
Read a single column and the lesson lands. “Eleven fives” at a full table is an 18% claim if you hold none — a bid you are throwing away. Hold two fives and the same words are 46%. Hold three and they are 63%, a bid you should be glad to make. Your hand is not a tiebreaker; it is worth about three steps of quantity, and playing as though it were not is the most common way to lose here.
The practical version: bid on the face you are longest in. Not the face that beats the standing bid most cheaply, and not the face someone else keeps naming — the one your own cup supports. It costs nothing and it moves every number on that row.
When to call the liar
Calling is not a mood. Take the standing quantity, subtract the matching dice in your own hand, and ask what is left of the bid for the dice you cannot see. If that remainder sits past the expectation for the unseen dice, the bid is in trouble; if it sits two past, it is usually dead.
Three rules of thumb that cost nothing to remember:
- Below the expected count, almost never call
- A bid under a third of the dice in play is true far more often than not. Calling it is a gift.
- One past the expectation, look at your hand
- This is the whole game. Holding two or more of the face, believe it; holding none, the bid is a coin flip at best and you have the better of it.
- Two past the expectation, call unless your cup argues otherwise
- The bidder needs the unseen dice to run hot. They usually do not.
One more thing the arithmetic will not tell you: who made the bid. A quantity that arrives after a long pause and a small raise is a different animal from the same quantity thrown out instantly. The number sets the odds; the person sets whether to trust them.
The exact call is a trap at every table but one
“Spot on” looks like the glamorous play — name the standing bid exactly right and every other player adds a round to their cylinder and pulls, while you watch. The payout is the biggest in the game. The trouble is how narrow the target is.
“At least eleven” is satisfied by eleven, twelve, thirteen and everything above. “Exactly eleven” is satisfied by eleven. Counting only the dice you cannot see, here is the best the exact call ever manages — its most likely count, and the chance of landing on it:
| Players | Dice you can't see | Likeliest count | Chance of exactly that | Worth it? |
|---|---|---|---|---|
| 2 | 5 | 1 | 33% | Heads-up only |
| 3 | 10 | 3 | 26% | No |
| 4 | 15 | 5 | 21% | No |
| 5 | 20 | 6 | 18% | No |
| 6 | 25 | 8 | 17% | No |
The bots reach for the exact call above 30%. Compare that with the column beside it: the exact call clears the bar heads-up and nowhere else. With one opponent and 5 dice out of sight it peaks at 33%; at a full table, with 25 hidden, its ceiling is 17% — and that is the ceiling, achieved only if you call on precisely the right count. Anywhere else on the distribution it is worse.
So: at a table of two, spot-on is a live weapon and you should be looking for it. At a table of five or six it is a way of donating a bullet with an audience. If the rule is on and the table is full, treat it as scenery.
What the wild rule is really worth
Hosts can turn wild ones off, and it is easy to read that as a small adjustment. It is not. It halves the chance any given die helps you, from two faces in six to one — and because bids compound that chance across every die on the table, the effect at the quantities people actually bid is enormous.
| Ones wild | Ones plain | |
|---|---|---|
| At least 8 | 83% | 11% |
| At least 10 | 57% | 2% |
| At least 11 | 42% | 1% |
| At least 12 | 28% | <1% |
The same words — “at least ten fives” at a full table — are a comfortable 57% bid with wilds and a 2% fantasy without them. If you sit down at a table with ones plain and keep bidding the numbers you learned with wilds on, you will be called every round and you will deserve it. Divide your instincts by three and start again.
Bluffing that survives contact
Everything above prices honest bids. The lie is what you do with the price.
The bluff that works is the one that is nearly true. A bid one past what your hand supports is cheap: it is still plausible, it still forces the next player to raise into thinner air, and if it is called you were only slightly wrong and the count may rescue you anyway. The bluff that fails is the jump — quantity three or four past the expectation, made because a big number feels intimidating. It is not intimidating. It is legible, and it is the easiest call anyone at the table will make all night.
Two habits worth building:
- Bid your second-longest face sometimes
- If you always name the face you hold most of, a watchful table learns to read your cup from your bids. Naming your second face costs you a little accuracy and buys back the whole read.
- Keep your raises the same size
- Players who raise by one when comfortable and by three when bluffing are announcing themselves. Pick a rhythm and keep it whether you are lying or not.
And remember the clock. A turn runs 60 seconds by default, and letting it expire is not a neutral outcome — the house loads every chamber you have left and fires for you. Deciding late is a decision; deciding never is a loss.
The revolver changes what a round is worth
In most versions of this game a lost round costs a die. Here it costs a bullet, and the bullet is permanent: the first loss loads the table's setting, every loss after adds one, and the cylinder is spun fresh before each pull. The chance is always the rounds you carry out of six.
That single rule should bend how you play as the evening goes on. On a clean cylinder you can afford a speculative call — a wrong one costs a 17% pull. Carrying three rounds, the same wrong call is a coin flip on your life, and marginal aggression that was correct an hour ago has stopped being correct. Tighten as you load.
The reverse is worth watching too. A player carrying four or five rounds cannot afford to be caught, which makes them predictable: they will bid small and true, and they will not call you without a hand behind it. Bluff the heavily loaded player and believe the clean one — the odds tables on the odds page price both sides of that in full.
The mistakes that cost the most
- Bidding a face you hold nothing of
- You are giving up the only information advantage you have, and every number on the leverage table above moves against you.
- Calling because a number sounded big
- At thirty dice, “ten” sounds enormous and is exactly the expectation. Do the third, every time.
- Reaching for spot-on at a full table
- Its ceiling there is 17%, and you will rarely be at the ceiling.
- Playing wild numbers at a plain table
- Check the house rules before the first bid, not after the first call.
- Playing the same aggression on a loaded cylinder
- The bid is priced the same; your loss is not.
None of this is a substitute for playing. Take a seat against the house bots — they run on exactly the constants quoted above, so a table against them is a way of testing every claim on this page — and when the arithmetic starts arriving before the panic does, you are ready for people. The rules take three minutes, the odds pages have the full tables, and the lobby is open.