Craps Odds and House Edge: Which Bets Cost Less?

→

Updated September 2026

Licensed

Available in AU

Fast payouts

18+ Only

Craps bets vary sharply in mathematical cost. The useful way to compare them is to separate the low-edge line decisions from place, field and proposition wagers, then examine the probability and payout attached to each. The odds bet deserves special context: when paid at true odds it has no house edge on that portion, but it is attached to a line bet and increases total money at risk. Wizard of Odds approaches craps as probability analysis, not as a profitable betting system, and the same distinction matters throughout this guide.

The craps bets worth separating first

Craps presents many wagers on the same layout, but they do not carry the same mathematical cost. The first useful distinction is between the main line bets – Pass Line and Don’t Pass – and proposition or one-roll bets. Line bets are built around a sequence: a come-out roll may resolve immediately or establish a point, after which the result depends on whether that point or seven appears first. Proposition bets are usually resolved more quickly and often pay at prices that are less favourable relative to their true probability.

House edge is a comparison measure, not a forecast for one session. It expresses the average mathematical cost of a wager under a defined set of rules and payouts. Short runs can finish far above or below that expectation because variance is substantial. The useful question is therefore not whether a bet can win on the next trial, but what price its rules impose over repeated play.

Pass Line: why the sequence matters

A Pass Line wager cannot be evaluated from a single dice total alone because its state changes after the come-out roll. On the come-out, 7 or 11 wins, 2, 3 or 12 loses, and 4, 5, 6, 8, 9 or 10 establishes a point. Once a point exists, the bet wins if that point repeats before a 7 and loses if 7 arrives first. The familiar low house edge comes from combining all of those branches, not from treating every roll as an even contest.

This structure is a good example of why probability trees matter. The chance of rolling a total depends on how many combinations of two dice produce it: 7 has six combinations, 6 and 8 have five each, 5 and 9 have four each, and so on. A correct expected-value calculation weights each branch by those combination counts and then applies the actual payout. The result can be compared with other craps decisions on a common per-unit basis.

Don’t Pass and the role of the 12

Don’t Pass largely reverses the Pass Line logic, but it is not a perfect mirror because a come-out 12 is normally a push rather than a win. That small rule detail matters. It changes the probability model enough that the mathematical cost is slightly different from Pass Line. The lesson is broader than craps: when two wagers look like opposites, check for pushes, commissions and asymmetric settlement rules before assuming their expected values are exact negatives of one another.

After a point is established, Don’t Pass benefits from the fact that 7 is more likely than any individual point total. For example, if the point is 4, there are three combinations that make 4 and six that make 7. The bet is still not a guaranteed advantage because the come-out rules and payouts are part of the complete calculation. Ranking the wager requires the whole sequence rather than one favourable stage.

Free odds: zero house edge needs context

The odds wager behind a Pass Line or Don’t Pass bet is unusual because it is paid at true odds for the point. In isolation, that means the odds portion has zero house edge: its payout matches the probability relationship between making the point and seven appearing first. That does not make the combined position a positive-expectation system, because the underlying line bet remains in place and still carries its own mathematical cost.

This is why percentage comparisons can be misleading if the denominator is not stated. Adding odds can reduce the blended house edge when measured against the total amount placed on the table, but it also increases the amount of money exposed. A lower percentage of a larger amount is not automatically a smaller expected dollar loss. For decision-making, compare both the rate and the total action required by the strategy.

Place 6 and Place 8

Place bets on 6 and 8 are commonly among the lower-cost alternatives outside the line-bet family. Each number has five dice combinations, while 7 has six. A standard place payout of 7 to 6 is close to, but below, the fair price implied by that race. The shortfall between fair odds and the actual payout creates the house edge.

The calculation also shows why a familiar table label is not enough. A place bet, buy bet and lay bet can refer to the same underlying number but use different payout or commission rules. The correct comparison therefore begins with the exact wager type and fee treatment. A percentage from a place-bet table should not be transferred to a buy-bet version without checking those terms.

Field and proposition bets

The Field resolves on the next roll and pays according to a defined set of totals, with 2 and 12 sometimes receiving enhanced payouts. Because casino rules vary, the exact edge depends on whether those extreme totals pay double, triple or another amount. That makes the Field a useful case study in rule sensitivity: the winning numbers alone do not determine value; the payout schedule does.

Hardways and other centre-table propositions similarly combine distinctive probabilities with quoted payouts that are below fair odds. They can produce larger headline wins than a line bet, but payout size should not be confused with value. A bet can offer a high multiple precisely because the event is rare, while still retaining a larger expected cost per unit than a quieter wager.

A practical cost-ranking method

Instead of memorising every percentage, rank craps wagers with a repeatable process. Define the exact bet, list the outcomes that resolve it, count the dice combinations for those outcomes, note any pushes or commissions, and apply the net payout. Expected value is the probability-weighted average of those net results. Converting a negative expected value into a percentage of the initial stake gives a house-edge measure that can be compared across wagers.

This approach also protects against category errors. The odds portion of a line bet, for example, should not be described as if it were available independently under ordinary rules. A buy bet with a commission collected only on wins should not be treated the same as one with a commission charged up front. The maths follows the rule wording, so small procedural details can move the ranking.

Volatility is separate from mathematical cost

Two craps bets can have different hit frequencies even when their house edges are relatively close. A wager that resolves often may generate a smoother stream of small outcomes, while a rare proposition can produce long losing sequences interrupted by a larger payout. That difference is volatility. It affects the experience and bankroll swings, but it does not replace expected value as the measure of long-run cost.

For a finite session, variance can dominate the expected loss. A player can win on a high-edge proposition or lose quickly on a lower-edge line bet. Neither result invalidates the underlying probabilities. The purpose of a house-edge table is comparative: over repeated exposure, lower-edge choices surrender less value per unit on average than otherwise comparable higher-edge choices.

What odds multiples do and do not change

Tables may allow different maximum odds multiples behind a line bet. Increasing the odds component changes the total amount at risk and lowers the blended edge as a percentage of total money wagered because more of the position is placed at true odds. It does not erase the expected loss attached to the original line bet, and it does not make the dice more predictable.

A sensible comparison therefore reports the convention being used. If one source quotes edge per initial line bet and another quotes edge per total amount including odds, the percentages can look inconsistent even when both calculations are internally correct. Before comparing numbers, identify the denominator and the assumed odds multiple.

Reading Wizard of Odds craps material

Wizard of Odds is most useful here as a reference for structured probability analysis rather than as a place to gamble. A good craps table should make the bet name, payout and resulting edge clear enough that a reader can distinguish low-cost line decisions from expensive proposition wagers. Where a rule varies by casino or table, the variation should be treated as an input to the calculation rather than ignored.

Wizard of Odds is a gambling information and casino-analysis website rather than an online casino operator. For Australian readers, the material is best treated as mathematical education: identify the rules, compare expected values, understand volatility and avoid turning a lower-cost choice into a claim that gambling has become profitable.

Decision takeaway

The main practical lesson is that the craps layout is not one product with one house edge. It is a menu of mathematically different contracts. Pass Line, Don’t Pass, odds, place bets and proposition bets resolve under different rules, so they should be compared separately. Lower-edge choices reduce expected cost relative to higher-edge choices, but they do not create a guaranteed winning system.

For an educational comparison, start with the line bets, understand why true-odds additions are different, then scrutinise one-roll and proposition payouts. If a claimed system relies only on changing stake size after wins or losses, ask whether it changes any dice probability or payout. If it does not, it changes the distribution of session outcomes rather than the underlying expectation.

Related guides

• Wizard of Odds Australia
• Wizard of Odds game strategy
• roulette odds Wizard of Odds
• Wizard of Odds baccarat
• casino house edge explained

Written by the editors at Wizard Of Odds.