Sic bo strategy: which bets are least bad and how to size them

Sic bo is three dice and a large menu of bets. The game overview, table layout and online versions are covered in sic bo online. Strategy here means one thing: knowing what each bet costs, so you can pick the cheapest ones.

Three dice produce 6 × 6 × 6 = 216 equally likely outcomes. Every bet on the table wins on some number of them. If a bet wins on w outcomes and pays p to 1, its return per unit is (p + 1) × w / 216, and its house edge is one minus that.

Example: Small

Small wins when the total is 4 to 10, unless the roll is a triple. Totals 3 to 10 account for 108 outcomes. Three of those are triples (1-1-1, 2-2-2 and 3-3-3), and a total of 3 is only possible as 1-1-1. So Small wins on 108 − 3 = 105 outcomes. At 1 to 1, the return is 2 × 105 / 216 = 210/216, a 2.78% edge. Big is the mirror image.

That triple clause is the whole house edge on Big and Small. Without it, the bets would be exactly fair.

Why it matters

Paytables vary by casino, by country and by software provider. If you learn the formula, you can price any bet in a few seconds, including ones this guide does not list. Dice combinatorics in more depth are in dice roll probability.

The table below uses one common paytable. Your table may differ; recompute with the formula if it does.

Where paytables get worse

Lower payouts on the same bets are common and expensive:

• Specific triple at 150 to 1: 151/216 returned, a 30.09% edge.
• Any triple at 24 to 1: 150/216 returned, a 30.56% edge.
• Specific double at 8 to 1: 144/216 returned, a 33.33% edge.

At those rates, a third of every bet is gone in expectation. The attraction is the headline number. A 150 to 1 payout sounds generous until you notice the true odds are 215 to 1.

A single-number bet picks one face, say 4. You win 1 to 1 if one die shows 4, 2 to 1 if two do, and 3 to 1 if all three do.

Counting the outcomes:

• No 4s: 5 × 5 × 5 = 125
• Exactly one 4: 3 × 25 = 75
• Exactly two 4s: 3 × 5 = 15
• Three 4s: 1

Expected result per unit = (75 × 1 + 15 × 2 + 1 × 3 − 125) / 216 = −17/216 ≈ −7.87%.

The intuition that trips people up: with three dice and six faces, "half the time" one of them should be a 4. In fact at least one 4 appears 91 times in 216, about 42%. The multiple payouts on doubles and triples do not make up the gap.

This is the same bet as the carnival and pub game chuck-a-luck, which also runs 7.87%. Some Asian tables pay 12 to 1 (or more) when all three dice match your number. Then the numerator becomes 75 + 30 + 12 − 125 = −8, an edge of about 3.70%. If you prefer single numbers, that paytable is worth finding.

Betting Big and Small together

Some players cover both to "stay in the game". On any non-triple roll one bet wins and one loses, so you break even. On a triple both lose. That happens 6 times in 216, so you lose 2 units about 2.78% of the time. You pay the same edge on double the action and win nothing. There is no hedge in sic bo that removes the edge; every combination is a blend of the edges of its parts.

Stacking long shots onto a safe bet

Adding an any-triple bet to your Small bet to "cover" the triple loss is a common idea. It does cover it: the triple pays 30 to 1. But you now pay 13.89% on the triple money in addition to 2.78% on the Small money. The combination costs more per round than Small alone.

Result boards

Electronic sic bo tables display recent rolls and label numbers as "hot" or "cold". Dice have no memory, and the shaker resets every round. A total that has not appeared for 50 rolls is exactly as likely next roll as it always was. Waiting for it is the gambler's fallacy.

Multiplier variants

Some live studios add random multipliers to certain outcomes and reduce the base payouts to fund them. Their RTP can be higher or lower than a classic table. Read the RTP in the help screen rather than assuming the standard numbers apply.

Big and Small behave like even-money bets, so the swings are about one unit per roll. For a $10 bet on Small:

• Expected loss per roll: $10 × 2.78% ≈ $0.28.
• Over 100 rolls: about −$28 expected, with a standard deviation of roughly $100.

That spread means a 100-roll session commonly finishes anywhere from about −$225 to +$170 at two standard deviations. The average is the cheap part; the variance is what you need a bankroll for.

A simple structure

1. Set a session budget, for example $300.
2. Use a unit of 1/30 of it: $10.
3. Bet Big or Small (or Odd/Even) flat.
4. If you want the occasional long shot, cap it at one small chip per few rounds, and count it as entertainment spending at its real price.
5. Stop at a written loss limit or time limit.

The bankroll calculator can model other sizes, and variance in gambling explains why short sessions stray far from the expected loss.

Compared with other dice games

Big and Small at 2.78% are more expensive than the craps pass line at 1.41% and much more expensive than the craps odds bet at 0%. If low cost is the priority, craps strategy is worth a look. The casino games hub ranks the rest of the table games.

Put three players at the same sic bo table for 100 rolls, each spending $10 a roll, and the edges above turn into dollars.

The maths behind each row:

• A: 100 × $10 × 2.78% ≈ $27.80.
• B: 100 × $5 × 2.78% ≈ $13.90 on Small plus 100 × $5 × 18.98% ≈ $94.90 on total 9, about $108.80 in total. Half the money is on a cheap bet and half on an expensive one, so the blended edge is about 10.9%, nearly four times player A's cost for the same $10 a roll. Split bets inherit the worse edge in proportion to the money placed on it.
• C: 100 × $10 × 16.20% ≈ $162. A specific triple hits 1 time in 216, so in 100 rolls the chance of seeing at least one is 1 − (215/216)^100 ≈ 37%. Most sessions for player C end at −$1,000. The ones that hit once end near +$810.

Player C's results are the clearest example of how a large payout hides a high edge. The average is −$162, but almost nobody experiences the average; they either lose everything they staked or win big once.

Reading an unfamiliar paytable

When you meet a table or app with different numbers:

1. Count the winning outcomes out of 216 (or look them up in the ranking above).
2. Add 1 to the payout to get the total return on a win.
3. Multiply, divide by 216, and subtract from 1.

Example: a specific double paying 11 to 1 wins on 16 outcomes. 12 × 16 = 192, and 192/216 ≈ 88.9%, an 11.1% edge. Better than 10 to 1, still four times the cost of Small.

The sic bo formula, return equals payout times probability, works on every game. Apply it to PVPspinArena Roulette: Purple wins on 7 of 15 slots and pays 2x, so 2 × 7/15 = 14/15. Green wins on 1 of 15 and pays 14x, so 14 × 1/15 = 14/15. Every bet returns about 93.33%, a 6.67% edge, whichever colour you pick. That makes the colour choice a question of variance, not cost, much like choosing between Small and a specific total with a similar edge.

Round results come from committed seeds and can be checked on the fairness page. Gambling is 18+ (or the local legal age). If you find yourself raising stakes to recover losses, the responsible gambling page has limits and support.

In the same cluster, see also teen patti and pai gow tiles.