Even betting the Pass Line with full 3-4-5x Odds — about the best deal in the casino — only 48.27% of ~3 hours sessions finish ahead, and a typical $300 buy-in busts in 65.73% of full-day sessions.
We played 500,000 craps sessions roll by roll through Felt Trainer's actual craps engine — the same code that runs the trainer — betting a flat $10 Pass Line and taking full 3-4-5x Odds on every point. That is the lowest house-edge way to play craps (a blended edge well under the flat Pass Line's 1.41%). The run uses a fixed random seed, so anyone can reproduce these exact numbers. Across the studies below it resolved roughly 600 million dice rolls.
Variance, not the house edge, decides any single night — but the edge decides the trend. The longer you stay at the table, the more your odds of being ahead erode:
| Session length | Finished ahead | Finished behind | Average result | Middle-90% range |
|---|---|---|---|---|
| ~1 hour (100 rolls) | 48.36% | 50.14% | -$3 | -$440 to $440 |
| ~3 hours (300 rolls) | 48.27% | 50.87% | -$12 | -$770 to $760 |
| an all-day session (600 rolls) | 48.02% | 51.37% | -$25 | -$1,100 to $1,060 |
Even on the best bet in the building, the table slowly wins. After an all-day session, only 48.02% of sessions were ahead, and the average player was down $25.
How often does a realistic $300 buy-in (30× the flat bet) run dry before the session ends?
| Session length | Busted before the end |
|---|---|
| ~1 hour (100 rolls) | 21.16% |
| ~3 hours (300 rolls) | 51.23% |
| an all-day session (600 rolls) | 65.73% |
A simulation is only as trustworthy as its engine. As a control, we measured the realized house edge of several bets over the run and compared it to the published figure. They match, which is the same guarantee our methodology describes:
| Bet | Simulated edge | Published edge | Match |
|---|---|---|---|
| field | 2.7778% | 2.78% | ✓ match |
| anySeven | 16.6667% | 16.67% | ✓ match |
| yo | 11.1111% | 11.11% | ✓ match |
No signup, no download. Learn the lowest-edge way to play.
In this simulation of 500,000 sessions of Pass Line + 3-4-5x Odds (the lowest-edge craps play), about 48.27% of three-hour sessions finished ahead and 50.87% finished behind. The longer you play, the lower your chance of being ahead, because the small house edge compounds over more bets.
Odds bets pay true odds (zero house edge), so they lower the overall edge and increase your swings, but they do not flip the game in your favor. Over a long session the house edge on the flat Pass Line still grinds your bankroll down on average.
Every session was played roll-by-roll through Felt Trainer's actual craps engine — the same code that runs the trainer — using a fixed random seed so the run is reproducible. As a self-check, the engine's realized house edge matched the documented edge for each bet tested. See the methodology page for how we compute odds.