A craps Buy bet pays true odds and then adds a casino commission, so the house edge cannot be stated correctly until the commission rule is known. The number matters, but the vig timing matters just as much. A Buy 4 or Buy 10 with commission collected only after a win can be a very different wager from the same number with commission charged in advance.
Start with the commission rule, not the number
Players often ask, “What is the house edge on a Buy bet?” as if Buy were one wager. It is better to ask three questions first:
- Which number are you buying?
- Is the commission charged on every bet or only after a win?
- How does the table calculate and round the commission at your stake size?
The underlying payoff before commission is fair. The 4 and 10 pay 2:1, the 5 and 9 pay 3:2, and the 6 and 8 pay 6:5. The vig is what moves the expected value toward the casino.
That makes Buy-bet analysis very different from Place Bet House Edge, where the casino edge is embedded directly in the short payout. For booking mechanics rather than pricing, use Buy Bets Explained.
Exact 5% vig produces two very different price tables
The standard mathematical comparison is easiest when the commission is treated exactly rather than rounded. Wizard of Odds publishes both “commission always paid” and “commission on win only” versions. On a per bet made basis, the approximate house edges are:
| Number | 5% vig always charged | 5% vig on wins only |
|---|---|---|
| Buy 4 or 10 | 1.19% | 0.42% |
| Buy 5 or 9 | 1.32% | 0.56% |
| Buy 6 or 8 | 1.46% | 0.69% |
Those percentages use the bet-made basis, which includes neutral rolls while the Buy bet remains working. If you instead measure only resolved decisions—number versus seven—the effective edge is higher because the same expected loss is divided by a smaller slice of action. Under exact 5% commission, the resolved-decision edge is about 4.76% when the commission is always paid, versus roughly 1.67%, 2.00%, and 2.27% for win-only Buy 4/10, 5/9, and 6/8 respectively.
The denominator matters. A page that quotes “1.67%” and another that quotes “0.42%” can both be correct if one is pricing a resolved decision and the other is pricing the bet per roll/bet-made basis.
Why 4 and 10 are the natural Buy candidates
Place 4 and Place 10 commonly pay 9:5 even though true odds are 2:1. That payout gap is large enough that a commission-charged true-odds Buy can still beat the Place version.
Suppose you compare a $20 Place 4 with a $20 Buy 4 under a favorable win-only rule.
- Place 4 pays $36 profit at 9:5.
- True odds on $20 are $40 profit.
- If the casino collects $1 vig only after a win, the Buy produces $39 profit.
The exact effective edge depends on the commission convention and rounding, but the economic intuition is clear: the Place payout gives up $4 relative to true odds, while the Buy gives up the commission instead.
That is why “buy the 4 and 10” became common craps shorthand. It is not a magic strategy. It is a price comparison.
For the separate practical article on those two numbers, see Place 4 and Place 10.
Buying 5 or 9 becomes interesting only under favorable vig handling
The Place 5 and Place 9 payout of 7:5 is closer to true odds of 3:2 than Place 4/10 is to 2:1. There is less pricing room for a commission before the Buy loses its advantage.
Under an exact 5% commission charged every time, the Buy 5/9 is not attractive relative to the standard Place version. When the vig is collected only after a win, however, the comparison can reverse. Wizard of Odds’ bet-made figures put a win-only Buy 5/9 around 0.56%, while the standard Place 5/9 is about 1.11% per roll/bet made and 4.00% per resolved decision.
That does not mean every table offering “Buy 5” gives you that edge. Small-stake rounding and minimum commission rules can change the real price. A nominal 5% charge that rounds to $1 on a small bet is not economically identical to an exact 5% calculation.
Place 6 and 8 usually beat buying them
Six and eight are different. Their standard Place payout of 7:6 is already close to true odds of 6:5. A commission can cost more than the payout improvement is worth.
Using the common exact-commission model, Place 6/8 carries roughly 0.46% expected loss per roll (about 1.52% per resolved decision), while a win-only Buy 6/8 is about 0.69% per bet made. If the vig is charged regardless of outcome, the Buy is worse again.
This is why a blanket rule such as “true odds are always better” fails here. True odds before a fee are not the same thing as a fee-free true-odds bet. The commission has to be included in the comparison.
The useful shortcut is:
- 4/10: Buy is usually the first comparison to make.
- 5/9: check win-only commission and rounding.
- 6/8: standard Place pricing is normally better unless the Buy rule is unusually generous.
Rounding can matter more than the textbook percentage at small stakes
Casino commission handling is not always a clean spreadsheet calculation. A 5% vig on $25 is $1.25. One table may collect $1, another may require a different bet size, another may use a minimum commission, and another may calculate the fee on a specified unit.
That means two casinos can both advertise “5% vig” and still create different effective returns.
Consider a $25 Buy 4 with vig collected only on a win:
True-odds win = $50
If vig is exactly $1.25, net win = $48.75
If vig is rounded down to $1, net win = $49Conditional on a decision, 4 wins with 3 combinations and 7 wins with 6. With exact $1.25 vig:
EV = (3/9 × $48.75) - (6/9 × $25)
EV = $16.25 - $16.67
EV ≈ -$0.42
Resolved-decision edge ≈ $0.42 / $25 = 1.67%With a $1 rounded vig, the expected loss falls to about $0.33, or roughly 1.33% on the same resolved-decision basis. The rule did not change; the effective commission did.
Buy-bet math is a race against seven plus a fee
The underlying dice race is straightforward. Once a Buy number is working, only the chosen number and seven settle the wager. Other totals are neutral.
| Number | Winning combinations | Seven combinations | Fair payoff before vig |
|---|---|---|---|
| 4 or 10 | 3 | 6 | 2:1 |
| 5 or 9 | 4 | 6 | 3:2 |
| 6 or 8 | 5 | 6 | 6:5 |
If there were no commission, those payouts balance the decisive probabilities and the wager would have zero house edge. The casino’s price is therefore visible: it is the commission and any unfavorable rounding convention.
That makes Buy bets unusually transparent compared with proposition wagers. You can identify the fair price first and then see exactly where the casino margin enters.
At the table, commission control is part of game protection
A Buy wager is not only a probability problem. It is also a procedure problem. The dealer must distinguish Buy action from Place action, associate the wager with the correct player position, apply the correct payout, and collect or account for vig according to house rules.
Those details create opportunities for repeated small errors. A missed commission, a commission charged twice, an incorrect lammer, or a payout made using Place odds rather than Buy odds may look minor on one transaction but become material over hundreds of decisions.
A strong floor therefore cares about three things at once: correct rule, correct player ownership, and correct payout/commission sequence. Surveillance reviews the same sequence if a dispute or suspicious pattern appears.
For the broader operational side of number bets, Place Bets Explained gives useful context on booking and player-position control.
Use a comparison rule instead of memorizing one “best” Buy bet
Before making a Buy wager, compare the actual table offer with the alternative available on that number.
- Confirm the Buy payout is true odds.
- Ask whether vig is paid in advance or only after a win.
- Ask how the commission is rounded at your intended stake.
- Compare Buy 4/10 with Place 4/10.
- Compare a win-only Buy 5/9 with Place 5/9 if the table permits it.
- Do not assume Buy 6/8 beats the standard Place payout.
- Separate percentage price from dollar exposure: a cheaper edge on a larger wager can still create a larger expected dollar loss.
For a site-wide comparison, use Craps House Edge and the house edge calculator. If your real question is how much the percentage costs over repeated action, move from edge to dollars with Expected Loss Per Hour.