Sic Bo Strategy: Which Bets the Dice Math Favors

Sic bo looks like a game of hunches, but every sic bo strategy comes down to one decision: which boxes on the layout get your chips. Pick the bets with the smallest house edge, size them flat, and you have done everything the math allows. No progression, pattern or streak chart changes the edge on a single roll.

That is the finding the video tutorials skip. Three dice settle every wager at once, the paytable is fixed by state regulators, and each bet's edge follows from how often the dice land against what the table pays. The rest is variance: how hard and how often a bet swings.

The game behind every plan

New Jersey's rules spell out the hardware: three dice sealed in a shaker, and no shaker dice stay in play past 24 hours. A gambler cannot read wear on the dice or pick a lucky set. Every roll starts fresh.

The bets fall into two families. The even-money pair, small and big, pays 1 to 1. Under the New Jersey code, small wins on totals of 4 through 10, and it loses when the dice show three of a kind. Big is its mirror on the high totals. Everything else on the layout trades a lower chance of winning for a longer payout.

One reading tip before any numbers: New Jersey requires payouts to be posted with "to" or "win," never "for". Read every posted price that way: the first number is what you win, the second what you staked.

Sic bo strategy cheat sheet

The table below ranks the main wagers from the steadiest to the wildest. Payouts are the regulated minimums as of October 2026; the house edge column shows only figures the Wizard of Odds publishes.

Read it top to bottom. Small and big win close to half the time: the Wizard of Odds puts small's win probability at 48.61% and its edge at 2.78%. The triple is what tips it to the house. When all three dice match, small and big both lose.

The single-number wager, which pays more as more dice show your number, costs more. The same source puts the usual one-number edge at 7.87%. That is nearly three times the price of small or big for a bet that pays modestly.

Below that line sit the long shots. A total of 4 pays 50 to 1 under New Jersey's minimum table, and a specific triple pays 150 to 1. These bets lose most rolls and pay rarely, so a bankroll spent on them swings far harder than one spent on small or big.

Why the state you play in moves the edge

Here is the detail a strategy chart rarely mentions: the same box can be cheaper in one state than another. New Jersey's code sets any three of a kind at 24 to 1 as a minimum. Pennsylvania's base paytable sets the same bet at 30 to 1, and pays a total of 5 at 20 to 1 against New Jersey's 18 to 1. The dice do not care which state they are in, so a higher payout on the same event means a smaller house edge.

Pennsylvania's Super Sic Bo adds random multipliers to some boxes, but the regulation excludes small, big, odd and even from multipliers. The low-edge, even-money bets stay plain. A multiplier can make a long shot pay more on a given roll; it does not turn a long shot into a steady bet.

The single-number bet also differs. Pennsylvania's base table pays it 1 to 1, 2 to 1 or 3 to 1 by how many dice match, which is the usual paytable behind the 7.87% figure above.

Progressions do not change the math

Betting systems promise to fix a negative game by changing stake sizes. They cannot. The Wizard of Odds ran a long simulation of Martingale, doubling after each loss, on the craps pass line, an even-money bet much like small or big. In 19.65% of sessions the bettor reached a point where the next doubled stake could not be covered. The average session loss rose from $1.12 to $4.20 for the Martingale player against a flat bettor.

The conclusion is general: whatever the system, money lost over money wagered approaches the house edge. A progression on small still pays 2.78% of every dollar wagered over time. It only bunches the losses into rarer, larger hits.

That is the practical sic bo strategy: choose small or big for the lowest published edge, keep stakes flat, and treat any long-shot box as entertainment priced well above the even-money bets. For the full paytable and how the game runs on the Strip, see our Las Vegas sic bo payouts and odds story.

Letters: your questions

Is Martingale a good strategy?

No. In a simulation of Martingale on an even-money craps bet, 19.65% of sessions ended with the bettor unable to cover the next wager, and the average session loss rose from $1.12 for a flat bettor to $4.20. The same arithmetic applies to small or big at a sic bo table.

How should a beginner bet?

Start with small or big at flat stakes. Those bets carry the lowest published edge on the layout, 2.78%, and a flat stake keeps losses in line with that edge instead of letting a progression multiply them.

Does the state change the odds on a sic bo bet?

It can. Regulators set minimum payouts, and they differ: any three of a kind must pay at least 24 to 1 in New Jersey and 30 to 1 on Pennsylvania's base table. The dice are the same, so the higher payout means a smaller house edge on the identical bet.

Sources for this story

• New Jersey Administrative Code via Cornell LII, N.J. Admin. Code 13:69F-9.1, Dice; number of dice; sic bo shaker, accessed 2026-10-23
• New Jersey Administrative Code via Cornell LII, N.J. Admin. Code 13:69F-9.2, Permissible wagers (sic bo), accessed 2026-10-23
• New Jersey Administrative Code via Cornell LII, N.J. Admin. Code 13:69F-9.4, Payout odds (sic bo), accessed 2026-10-23
• Pennsylvania Gaming Control Board via Cornell LII, 58 Pa. Code 625b.1, Super Sic Bo, 2023-08-05
• Wizard of Odds, The Truth about Betting Systems, 2026-08-03
• Wizard of Odds, Sic Bo, 2026-10-08