We checked 2,126,892 Rollbit X-Crash rounds from 6 August 2024 to 4 October 2026. Every one matched Rollbit’s published maths, and all 20 proofs we sampled from 19 September 2024 on passed. Rounds before that date used an older starting value that Rollbit does not publish, so we could only check their maths. The game keeps 5% of what you bet.
📅 Checked 5 October 2026 · 2,126,892 rounds
Rounds tested
2,126,892
Formula mismatches
0
Proof samples passed
20 of 20
House edge
5%
In 30 seconds
All 2,126,892 Rollbit X-Crash rounds we tested matched Rollbit’s published maths. We worked out each result again and not one came out different.
We also sampled 20 rounds from 19 September 2024 on. All 20 proofs passed when tested with Rollbit’s public key.
Cash out at 2x every round and you get back 95.00% over time. Rollbit keeps the other 5%.
Six betting scripts ran over the real rounds. All six paid back less than they took in.
🔐 20 of 20 sampled Rollbit X-Crash proofs passed
All 20 Rollbit X-Crash rounds we sampled from 19 September 2024 to 5 October 2026 passed the proof check with Rollbit’s public key. That is a sample of 20 rounds, not every round.
31 sampled rounds
One of the 20 was round 4,817,482. That is the example round Rollbit shows on its own page.
Here is how it works. Rollbit uses a VRF, short for verifiable random function. That is a way of making random numbers where each number comes with a proof that anyone can test.
For each round, Rollbit makes a random value from the round number and a seed. The seed is a starting value that stays the same for a long run of rounds. Rollbit also makes a proof. Only Rollbit’s secret key should be able to make a proof that passes. Its public key lets anyone test it.
If a round’s value had been changed after the fact, its proof should fail. We checked 20 rounds this way, not all of them.
The seed today: the hash of Bitcoin block 861,970. That block was mined on 19 September 2024 at 12:42 UTC. Nobody knew that hash before then, Rollbit included.
The seed changed on 19 September 2024. Round 2,807,873 (01:13 UTC that day) fails with the current seed. Round 2,810,127 (21:23 UTC that day) passes. The switch happened between those two rounds.
That leaves about 2.01 million rounds that can be proof-checked today. We checked a sample of them, not all. The other 119,000 or so, from 6 August to 19 September 2024, used an earlier seed. Rollbit does not publish it. We sampled 11 of those older rounds and none pass with the current seed, as you would expect.
For those older rounds we could still check one thing: does the payout match the random value Rollbit lists? It does, every time.
📐 Rollbit X-Crash results match its 5% formula
Rollbit X-Crash reached every cashout target almost exactly as often as Rollbit’s formula predicts. Rollbit’s list had 2,126,968 rounds from 6 August 2024 to 4 October 2026. We left out 76 with no payout and counted the other 2,126,892.
The formula is simple. The chance a round reaches a target is 0.95 divided by that target. So 2x should come up 47.5% of the time. 10x should come up 9.5% of the time.
Share of rounds that reached each multiplier
The gaps are tiny. We also ran a standard statistical fit test, which checks if a spread of results fits a formula (score 6.25 on 10 degrees of freedom). The result is well inside normal chance, so the test passes.
We also worked out the payout of each of the 2,126,892 rounds from its random value, using Rollbit’s published formula. We got 0 mismatches.
The 76 rounds we left out: Rollbit’s round list shows a payout of 0 for these. They are spread from August 2024 to September 2026. They appear to be cancelled rounds. Their random values would give normal results, so we kept them out of the counts.
🔁 Low runs do not make 2x more likely on Rollbit X-Crash
On Rollbit X-Crash, after 5 rounds in a row under 2x, the next round reached 2x 47.35% of the time. Across all rounds the figure was 47.50%. Five low rounds in a row did not make a big round any more likely.
47.50% reached 2x
47.35% reached 2x
The line marks 50%. A losing run did not push the next round any closer to it.
We found 85,030 spots where 5 rounds in a row stayed under 2x. In 40,264 of them, the next round got to 2x. That is 47.35%.
The difference from 47.50% is 0.15 of a point. That small gap is normal chance. It is also lower, not higher, so a win was not due.
This fits how the game works. Each round’s value comes from its own round number and the seed. The last round’s result plays no part in it.
🤖 All six scripts paid back under 100% on Rollbit X-Crash
We ran six betting scripts over real Rollbit X-Crash rounds, and none paid back 100%. Each script played 425 sessions of 5,000 rounds. Each session started with 10,000 units. The base bet was 1 unit, except Grind, which bet 10.
The scripts:
- Flat 2x: bet 1 unit every round, cash out at 2x.
- Flat 10x: bet 1 unit every round, cash out at 10x.
- Martingale 2x: double the bet after each loss, cash out at 2x.
- Martingale 3x: raise the bet by half after a loss (rounded up), cash out at 3x.
- Wait 5: no bets until 5 rounds in a row finish under 2x, then Martingale 2x until a win.
- Grind 1.1x: bet 10 units every round, cash out at 1.1x.
Payback per script (scale starts at 90%)
Bar runs from 90% to 100%. The black line at the right is 100%, where you would break even. Every script stops short of it.
Flat 2x lost a little every session. Flat 2x lost 106,264 units across its 425 sessions. Not one session finished ahead. A typical session ended 252 units down.
Martingale 2x won most sessions but lost overall. A typical session finished 2,333 units up, and 57.2% of sessions ended ahead. But the 42.8% that went bust lost so much that the script was 721,478 units down overall.
Martingale 3x was worse. More than half its sessions (58.1%) went bust. It lost 1,005,157 units in total.
Grind lost the most. It bet 10 units a round, so far more money was bet against the 5% edge. It lost 1,063,900 units, and no session finished ahead.
Wait 5 won small most times and still lost overall. It ended ahead in 98.6% of sessions, and the middle session finished 94 units up. But the 1.4% of sessions that went bust lost enough to leave the script 9,239 units down in total. It paid back 98.62%, which is still under 100%.
- ✓ No crypto? You can buy it on Crashino with a card
- ✓ 30+ provably fair crash games: Aviator, JetX, Spaceman & more
- ✓ 10% weekly cashback for every player, paid automatically
- ✓ Extra value: 2x monthly bonus and up to 125 Friday Fun Spins
🏆 Rollbit X-Crash hit its 1,000,000x cap on 12 August 2024
The biggest Rollbit X-Crash round we found was 1,000,000x, on 12 August 2024 (round 2,705,299). That is Rollbit’s cap, so no round can pay more.
The longest run of rounds under 2x was 21 in a row. It ended on 31 March 2026, at round 4,312,320.
Twenty-one low rounds in a row sounds like a lot. Over 2.1 million rounds, long runs like this are bound to turn up now and then. They are not a sign that a big round is coming next.
💸 Rollbit X-Crash keeps 5 of every 100 bet, Bustabit keeps 1
Rollbit X-Crash has a 5% house edge, so over time you get back about 95 for every 100 you bet. At Bustabit, with a 1% edge, you get back about 99. At EtherCrash, with an edge of about 2%, you get back about 98.
Our own test shows the cost. Flat 2x bet 1 unit in each of 2,125,000 rounds. It lost 106,264 units. That is 5.0% of everything it bet.
Put it in session terms. Bet 1 unit a round for 5,000 rounds at a 5% edge and you should expect to lose about 250. At a 1% edge, the same play costs about 50. At 2%, about 100.
So the same night of play costs about five times more on Rollbit X-Crash than on Bustabit. You can see the full Bustabit numbers in our Bustabit test on 9.9 million rounds, and EtherCrash in our EtherCrash test on 2.26 million rounds.
🧾 Check a Rollbit X-Crash round yourself in six steps
Any Rollbit X-Crash round played since 19 September 2024 can be checked by you, with free tools. You need the round number, its random value and proof, the seed, and Rollbit’s public key. These are the steps we followed.
- Get the round. Find the round number, random value and proof in Rollbit’s public X-Crash round list.
- Get the seed. It is the hash of Bitcoin block 861,970. Rollbit also shows it on its own site.
- Build the message. Join these with a straight bar: 00000000-0000-0000-0000-000000000000, then rollbit:xcrash, then the round number, then the seed.
- Check the proof. Use a VRF tool (one that can test this kind of proof) that follows RFC 9381 with SECP256K1_SHA256_TAI. We used the free @roamin/ecvrf library. Give it the message, the proof and Rollbit’s public key. It should say the proof is good.
- Turn the value into a number. Take the first 8 bytes of the random value. Read them as one number, lowest byte first. Divide by (2 to the power of 64, minus 1). You get a number between 0 and 1.
- Work out the result. Divide 95 by that number. Round down to a whole number, then divide by 100. If it is under 1, the result is 1.00x. If it is over 1,000,000, the result is 1,000,000x.
Rollbit’s public key (as we used it):
03c41192ba26b7efafa9c69c941f759690d4fd46761c8424edbad1d5fc2c5c0947
For a round before 19 September 2024, step 4 will fail. That older seed is not public. Steps 5 and 6 still work, so you can confirm the payout fits the random value.
🚧 This test covers Rollbit X-Crash results only
Our test only shows that Rollbit X-Crash results match Rollbit’s published maths and proofs. It says nothing about the rest of Rollbit.
We did not test:
- Deposits, or how fast or fully withdrawals are paid.
- Account checks, limits or closures.
- Any other Rollbit game.
- The company behind the site, or how strong its licence is.
- The 119,000 or so rounds before 19 September 2024 for proofs, only for maths.
For the record, the footer on Rollbit’s site (checked 4 October 2026) says it is licensed by the Offshore Finance Authority of Anjouan, licence ALSI-042308006-FII. It names the operator as Acacia Tech Ltd, Nevis. We did not look into either.
❓ Rollbit X-Crash questions people ask
Is Rollbit X-Crash rigged?
We found no sign of it. Across 2,126,892 rounds from 6 August 2024 to 4 October 2026, every round with a payout fit Rollbit’s formula. All 20 proofs we sampled from 19 September 2024 on passed. The odds of each target matched the 5% formula.
Is Rollbit legit?
We can only speak for the X-Crash game. Its results match Rollbit’s published maths, and the 20 proofs we sampled from 19 September 2024 on all passed. We did not test deposits, withdrawals, the company or the licence.
What is the Rollbit X-Crash house edge?
5%. Rollbit’s formula sets it, and our numbers agree. Flat 2x betting paid back 95.00% over 2,125,000 bets.
Can a predictor guess the next Rollbit X-Crash round?
No. Each value should need Rollbit’s secret key to make. Past rounds give no clue either: after 5 rounds under 2x, the next hit 2x 47.35% of the time, against 47.50% overall. Our Predictor Scam Tracker lists apps and channels that claim otherwise.
Does Rollbit X-Crash use a hash chain?
No. It uses a VRF, a way of making random numbers that each come with a proof anyone can test. Each round has its own random value and proof, made from the round number and a seed. Since 19 September 2024 the seed has been the hash of Bitcoin block 861,970.
What is the highest multiplier on Rollbit X-Crash?
1,000,000x. That is Rollbit’s cap. We saw it on 12 August 2024, in round 2,705,299.
Using this data? You are free to use any figure on this page. Please link back to this Rollbit X-Crash test as the source.
👉 Keep going
Three things worth a tap next. Everything here uses free play money only.
Crash gambling and harm. 18+, play responsibly.