How to set up a run in the craps simulator
Start with the strategy. The list holds eight systems, each played exactly as the name says, so you do not have to program anything. Then set a unit size, which is the base bet the strategy scales from, and a bankroll, which is the money you bring to one session. Those two numbers decide how often a system goes broke.
Next, choose shooters per session, anywhere from 10 to 200. Ten shooters is a short visit to a table; 200 is a long day. Then choose the number of sessions, from 100 to 10,000. Each session starts with a fresh bankroll, so 10,000 sessions means ten thousand separate trips to the casino, played back to back in your browser.
Press Run. The dice come from your browser’s cryptographic random number generator, seeded so the run can be repeated, and every bet is settled by the same math engine used across this site. A place bet pays 7 to 6 here because it pays 7 to 6 on a real table, and the edges match our strategy guides.
The eight strategies in the Strategy Lab
Each strategy below is a real system players use, from the plain line bet with odds to the progressions people hear about from friends. The last column is what the strategy costs as a share of the money it puts on the table, which is the single most useful number for comparing them before you look at any histogram.
| Strategy | What it bets | House edge on money wagered |
|---|---|---|
| Pass Line + 3-4-5x odds | Pass Line, full odds on every point | 0.37% |
| Don’t Pass + lay odds | Don’t Pass, 3-4-5x lay odds | 0.27% |
| Place 6 & 8 | Place 6 and Place 8 | 1.52% |
| Iron Cross | Field plus Place 5, 6 and 8 | Mixed: Field 5.56%, Place 5 4.00%, Place 6/8 1.52% |
| 3-Point Molly | Pass Line and two Come bets, all with odds | 0.37% with 3-4-5x odds |
| Martingale on the Pass Line | Pass Line, doubled after every loss | 1.41% |
| Field every roll | One Field bet each roll | 5.56% |
| Any 7 every roll | One Any Seven bet each roll | 16.67% |
The Iron Cross is the one people misread most. It wins on every total except the 7, which feels safe, but a common spread of $5 Field, $5 on the 5 and $6 each on the 6 and 8 costs about $0.39 per roll. That is almost 1.8% of the $22 at stake, rolled over again and again.
The 3-Point Molly looks busy but carries the same price as a single Pass Line bet with odds, because a Come bet is a Pass Line bet that starts later. The Martingale changes nothing about the edge at all. It only changes how the losses arrive, which is why it earns its own row in the craps simulator.
How to read the craps simulator results
Look at the histogram first. Each bar is a range of session results, and its height is how many sessions ended there. A low-edge strategy with odds makes a wide, lumpy spread either side of a small loss. The Martingale makes a tall spike of small wins with a thin tail of large losses to the left.
Under the chart sit the numbers. The share of sessions ahead is how often you would have walked out a winner. The share busted is how often the bankroll ran dry. The average result is what a typical session cost or won across the whole run, and it is usually the least exciting figure on the screen.
Total wagered is the sum of every bet placed, and it is larger than people expect. Money won gets bet again, so a $300 bankroll can put several thousand dollars through the table in a long session. The theoretical cost is the house edge applied to that total, and with enough sessions the average result lands close to it.
A quick check on any result
Compare the average result with the theoretical cost. If the two are close, the run was large enough to trust. If they sit far apart, add more sessions. A $10 Pass Line player with full odds who wagers $5,000 in total has a theoretical cost of about $18.70, which is 0.374% of the total wagered.
Why a winning simulation does not beat the house edge
Run the Pass Line with odds for 100 sessions and a good share of them will finish ahead. That is normal. Craps results scatter widely over a short stretch, and a single session may involve only a few dozen decisions. The edge is small compared with that scatter, so it hides well inside any one night.
Now run 10,000 sessions with the same settings. The share ahead barely moves, but the average result drifts toward the theoretical cost and settles there. That is what the house edge is: not a promise that you lose tonight, but a fixed price per dollar wagered that shows up reliably once enough dollars have gone through the table.
Progressions show it best. A Martingale on the Pass Line wins one unit per sequence most of the time. But seven straight Pass Line losses happen about 0.86% of the time, and at a $10 unit that streak asks for a $1,280 eighth bet. Over 100 sequences, it shows up at least once about 58% of the time.
The simulator does not change any of this. It simply lets you watch it happen over more rolls than you could play in years, which is what a craps simulation is for. Once you have seen the average land on the theoretical cost a few times, systems that promise to beat the edge become much easier to judge.
Three runs worth trying in the Strategy Lab
First, the baseline. Run Pass Line with 3-4-5x odds at a $10 unit, a $300 bankroll, 50 shooters and 10,000 sessions. Note the share ahead and the average result. Then switch to Don’t Pass with lay odds at the same settings. The average improves slightly, as the lower edge predicts, and the histogram shape hardly changes.
Second, the illusion. Run the Martingale on the Pass Line at a $10 unit with a $1,270 bankroll, which covers seven losses in a row. Most sessions finish ahead. Look at the busted share and the size of the losses in the left tail, then compare the average result with the baseline run. The spike of small wins costs more.
Third, the contrast. Run Field every roll and Any 7 every roll at the same $10 unit. Field costs $0.56 per roll on average, Any Seven $1.67. Over 100 rolls that is about $55.56 against $166.67. Both histograms slide left fast, and the Any Seven run shows why the center of the table pays for the casino’s lights.
Sharing a run and checking someone else’s
Every run creates a link that holds its settings and its seed. Open it and the Strategy Lab replays the exact same rolls, so the result is reproducible to the dollar. That makes it useful for settling arguments, and for checking a claim that a system “never loses”, which the link will either confirm or quietly take apart.
If a friend sends you a link, change one thing at a time. Raise the sessions, or change the seed, or drop the bankroll to what they really bring to a table. Our guide to bankroll management helps with that number. Most winning links shrink fast when the sessions go from 100 to 10,000. Does theirs?