Start with twenty boring rounds
Place the smallest bet the demo allows, let the round run, and keep your hands off the cash-out button. Do that twenty times. It feels like a waste of a session and it earns its keep, because it shows you the one thing no written description gets across. You see how fast some rounds end, and how long others keep climbing while you sit there with your thumb hovering. A few bust under a second. Occasionally one runs for tens of seconds and the number on screen gets silly.
Then set an auto cash-out at 2.00× and leave it alone for another twenty. By now you know the controls, the pace and the reset, which is most of what free play is good for. The rest of this page is about the part people get wrong immediately afterwards.
What the free rounds can't tell you
- Nothing about the next bust point. Each round's outcome is settled before the plane takes off, so the rounds you just watched don't shift the odds of the one coming. Watch for an hour and you have exactly as much predictive information as when you started. The Aviator predictor breakdown goes through why the "pattern" claims fall apart, and is Aviator rigged? covers the seed-and-hash machinery underneath.
- Nothing you can carry across to the paid mode of a branded game. Whether a demo shares the odds of real play depends on the algorithm, the parameters, the payout and the rounding rules behind each, and only the game's published rules can answer that. Where all of those match, the only difference is whose balance is on the line.
- Nothing about the edge. Practice doesn't make a hidden number easier to guess. Operators configure platforms differently, and unless a game's rules state that demo and paid play use the same distribution, a winning free streak tells you nothing about any particular branded game. From the outside you can't verify the link.
Any casino demo also has a practical limit. You can't replay it. The rounds arrive, you react, they're gone, and if you want to know whether a different cash-out target would have done better over the same stretch you're left guessing. That's the gap the crash simulator here fills. It stays clear of Aviator, JetX, Spaceman and anything else carrying a brand, and it makes no claim to reproduce any provider's algorithm or RTP. It implements the general crash model the whole genre sits on, a multiplier climbing from 1.00× with a bust point drawn from P(reach x) = (1 − e) / x, on seeds you choose. Fix the seeds and the same hundred rounds come back in the same order, which means you can change one setting at a time and watch what it actually does.
Three experiments you can run
All three use one fixed seed pair, server seed cashoutodds-guide-demo with client seed g1. Same seeds, same rows, every time you load them. Type them in or use the link under each table.
Custom seeds
Each link opens the simulator pre-filled from the page URL, using a fragment shaped like #serverSeed=..&clientSeed=..&target=..&rounds=..&mode=bulk. Nothing in that fragment is sent anywhere; it configures the page in your browser. Swap in any seeds, target or round count you like. (Or ignore the links and type the two seeds into the seed fields by hand.)
Experiment A, 2× target over 100 rounds
Stake 1.00 a round, auto cash-out at 2.00×, 3% house edge, over 100 rounds.
| This session | Theoretical (100 rounds) | |
|---|---|---|
| Wins / losses | 58 / 42 | — |
| Net result | 16.00 | -3.00 (expected loss 3.00) |
| Longest losing streak | 5 rounds | — |
| Max drawdown | 8.00 | — |
| Chance of ending in profit | — | 34.43% |
| Chance of losing everything staked | — | 1.5×10⁻²⁷% |
Every figure in that table comes from the model described above, with simulated payouts rounded to the cent. The session finished at 16.00 against an expected net of -3.00 for the same stake and the same hundred rounds, which tells you how loosely a short run tracks the theory. Run experiment A on these seeds →
Experiment B, 10× target on the same hundred rounds
Keep the seeds, keep the stake, and move the target to 10.00×.
| This session | Theoretical (100 rounds) | |
|---|---|---|
| Wins / losses | 8 / 92 | — |
| Net result | -20.00 | -3.00 (expected loss 3.00) |
| Longest losing streak | 19 rounds | — |
| Max drawdown | 29.00 | — |
| Chance of ending in profit | — | 37.74% |
| Chance of losing everything staked | — | 0.0037% |
The simulated column is one particular sequence of rounds, so it lands above or below the theoretical figure next to it by chance. (At a 1.00 stake and these targets every payout is a whole number of cents, so rounding plays no part here.)
Start with the expected loss, 3.00, which matches experiment A exactly on the same stake and round count. With un-rounded payouts the target you pick leaves the expected return per unit staked where it is. What does move is everything to do with waiting: the longest drought here runs 19 rounds against 5 at 2×, because a 10× target wins so much more rarely. Same edge, far rougher ride. Open the same seeds at a 10× target →
Experiment C, running a bankroll down
Say you give yourself 20.00, stake 1.00 a round, cash out at 2×, up to 100 rounds. The risk of ruin maths answers this before any simulation does, with a 8.11% chance the bankroll falls below one stake inside those rounds, an expected final balance of 17.06 and a median of 16.00 in our model.
The simulation then walks one specific path through that spread, and because the seed pair is fixed it walks the same path every time you open it. Open this run pre-filled with a 20.00 bankroll →
For a different path at identical odds, hit "New seeds" and run it again with a fresh seed pair. Some runs survive and some go bust, and ten runs won't necessarily show you both. None of them forecasts your own session.
"I won in the demo, so I'll win for real"
This is the one worth taking apart, because it quietly treats the free session as a rehearsal and the paid one as the performance. Two independent hundred-round sessions at identical odds can land anywhere between comfortable profit and near wipe-out, and the probability columns above describe precisely that spread. The mathematics behind that goes back to Jacob Bernoulli, whose Ars Conjectandi appeared in 1713, eight years after he died. He proved an early form of what's now called the law of large numbers. Run enough trials and the share of wins you observe closes in on the true probability, as tightly as you care to specify, provided you are willing to run enough of them. The promise is about the long run and it is silent about everything shorter. Simulated runs carry their own small wrinkle, since payouts land on whole cents rather than exact fractions; the methodology page sets out how that is handled.
A hundred demo rounds is firmly in the shorter category. Even a thousand leaves plenty of room for a 2× target to come back ahead of theory, or well behind it, which is why the session column in experiment A sits well away from the expected line. So a free session can teach you the timing, the interface, the feel of a drought. It cannot certify your result as the one the game will keep giving you, because nothing about a short run is designed to do that job.
Learn the mechanics in the demo and leave it there. If you want to think about the trade-off between a low target that wins often and a high one that pays rarely, that's the best cash-out multiplier guide.
One last thing. If you catch yourself staking real money to match a demo result, or to climb back to the balance a simulation said you'd have, shut the tab for the day. The responsible gambling page lists places to get help.
