Craps RTP

Craps RTP is best understood as a translation of house edge, not as a slot-machine promise. If a wager is quoted with a 1.41% house edge on a clearly defined basis, the corresponding return-to-player figure is about 98.59% on that same basis. The difficulty is that craps contains one-roll bets, multi-roll contract bets, working number bets, and zero-edge odds wagers, so the denominator has to be stated before an RTP percentage is meaningful.

RTP is a price label, not a forecast of your session

“Return to player” sounds as though the casino will eventually hand back a fixed percentage of your buy-in. That is not what it means.

RTP is an expected-value label over a very large sample of comparable wagers. If a $10 wager has 98.59% RTP on the standard house-edge basis, the long-run mathematical return is $9.859 per $10 wagered, with $0.141 expected loss. A real player can lose the full $10, win $10, push, or continue through several rolls depending on the bet.

Craps makes this especially important because a $500 buy-in is not the same thing as $500 of action. The same chips can be wagered, returned, pressed, and wagered again many times. RTP applies to wagered action under a defined model, not to the amount you originally carried to the table.

For the underlying casino-price concept, see Craps House Edge. For the dollar translation, use Craps Expected Value and Expected Loss Per Hour.

The denominator problem matters more in craps than the formula suggests

The simple relationship is:

RTP = 100% - House Edge

That equation is only useful when the house-edge figure and RTP use the same denominator.

A one-roll Any Seven wager resolves immediately, so there is little ambiguity about the action being priced. A Place 6 may remain on the layout through many neutral rolls. A Pass Line wager may begin on the come-out and later become a point contract. An odds bet does not exist until the flat line bet has established a point.

Analysts therefore report craps cost in more than one way: per initial bet, per resolved decision, or per roll. Converting a number from one basis into “RTP” without naming the basis creates false precision.

The safest site convention is to use the familiar standard house-edge basis for common headline comparisons and then explain per-roll figures where they matter operationally.

A clean RTP map for common wagers

Using the standard headline house-edge figures for familiar bets, the approximate RTP translation looks like this:

This table is useful for price comparison, but it should not be read as a ranking of hit frequency or session safety. A bet can have high RTP and still lose often. A bet can hit frequently and still have poor RTP if the payout is too short.

The distinction between probability and payout is why craps odds and RTP are separate subjects.

The 100% RTP odds wager is a special case, not a free game

Once a point is established, the odds portion behind Pass/Come or laid behind Don’t Pass/Don’t Come is paid at true odds. On the right side:

Because the payoff matches the decisive probability, the odds portion has zero expected house advantage: 0% edge, 100% RTP.

That statement applies only to the added odds money. The flat Pass, Come, Don’t Pass, or Don’t Come wager that gives you access to odds retains its own casino price. This is covered in detail on Odds Bet House Edge.

A 100% RTP wager can also lose. If the point is 4, the odds bet loses to seven twice as often as it wins to 4; the 2:1 payoff is what makes the expectation fair.

Blended line-plus-odds RTP needs careful wording

Players often say, “Taking more odds increases my RTP.” That is directionally correct when they mean the percentage price of the combined action becomes smaller because a larger share of the money is on a zero-edge component.

But a careless calculation can be misleading. Odds are not placed on every Pass Line come-out result; they appear only after a point is established. Table limits may use single, double, 3-4-5x, 5x, 10x, or another schedule. The amount of odds allowed can also depend on the point.

For that reason, this page does not use a simplistic 1.41% ÷ (1 + odds multiple) formula as though every dollar were present for the same number of rolls. The dedicated combined house edge with odds page is the better place to compare complete structures.

The robust principle is enough here: adding zero-edge odds reduces the percentage casino price of the combined money committed, while increasing the dollar amount exposed when a point decision occurs.

High RTP can coexist with severe short-session losses

Consider two players with the same $300 bankroll.

Player A makes $10 Pass Line bets and takes no odds. Player B makes $10 Pass Line bets and routinely adds $50 odds when permitted. Player B has a lower blended percentage edge on the combined action, but a point-seven sequence can cost $60 instead of $10.

RTP improved. Bankroll volatility increased.

This is not a contradiction. RTP measures expected price; variance describes dispersion around that expectation. In a short session, dispersion dominates what the player feels.

The same distinction appears with Place 6/8. Their headline edge is low, but covering both numbers with larger bets creates more dollars at risk. “Good RTP” does not impose a stop-loss, prevent a seven-out, or guarantee enough time for averages to stabilize.

For the volatility side, read craps variance and the broader explanation in why low house edge does not mean safe.

Casino operators think in theoretical win, hold, mix, and volume

RTP is useful mathematically, but a casino floor does not manage a live craps game by waiting for each player to realize a published RTP percentage.

Management looks at average bet, decisions or rolls per hour, wager mix, rated action, table occupancy, actual win/loss, and theoretical win. A table dominated by low-edge line bets with odds may have a lower theoretical margin than a table with heavy proposition action, yet it can still be economically valuable because of volume and time.

Actual hold can also differ sharply from theoretical expectation over one shift. A hot shooter can produce a losing table even when every wager was booked correctly. A sequence of early seven-outs can produce the opposite. RTP does not disappear; it is simply a long-run expectation operating underneath noisy results.

This operational distinction is why Craps Roll Speed and Total Action matters. Price per dollar and number of dollars cycled through the game are separate drivers.

Translate RTP into dollars before deciding whether a difference matters

Percentages become more useful when connected to action.

If two wager structures are compared on the same legitimate house-edge basis:

Expected loss = relevant amount wagered × house edge
RTP = 100% - house edge

For $1,000 of comparable action:

1.41% edge -> about $14.10 expected loss -> about 98.59% RTP
4.00% edge -> about $40.00 expected loss -> 96.00% RTP
16.67% edge -> about $166.70 expected loss -> about 83.33% RTP

The word comparable matters. Do not multiply a per-resolved-decision edge by raw roll count without converting the exposure correctly. Do not treat a multi-roll contract bet as though it is re-bet from scratch after every neutral roll.

When you want a dollar estimate, the expected loss calculator and house edge calculator are more useful than staring at RTP alone.

Use RTP for comparison, then switch to the metric that answers your real question

RTP answers, “How expensive is this wager under its stated mathematical basis?” It does not directly answer:

• How often will I win a decision?
• How large will my swings be?
• How much will I lose tonight?
• How many decisions will I make per hour?
• How much of my rating will the casino credit as theoretical action?
• How long will my bankroll last?

Those questions need probability, variance, pace, rating, and bankroll analysis.

A disciplined way to use the site is therefore: compare the price on Craps House Edge, understand the probability on Craps Odds, convert the result into dollars on Expected Loss Per Hour, and use Craps Variance for the part RTP cannot describe.