Auto Roulette Live Stats & Results: Reading 275,000 Spins of a Wheel With No Dealer
By LiveCasinoData editorial team · Updated 2026-08-28 · 14 min read
Our Auto Roulette tracker has logged 275,852 spins from Evolution's automated single-zero wheel since 28 December 2025, roughly one every 51 seconds. Over the last 90 days every number landed between 2.62% and 2.78% of the time against a 2.70% theory, red and black split 48.63% to 48.64%, and the longest zero drought was 396 spins. The widgets describe the past; none of them moves the odds of the next spin.
Key takeaways
- The tracker at /games/auto-roulette has stored 275,852 Auto Roulette spins since 2025-12-28, about 1,690 per day at one spin every 51 seconds.
- Across 103,221 spins in the last 90 days the most frequent number (10, at 2.78%) and the least frequent (21, at 2.62%) differ by only 167 hits, which is ordinary random variation around 2.70%.
- Red landed 48.63% and black 48.64% over 90 days; zero landed 2.73%, against theoretical shares of 48.65%, 48.65% and 2.70%.
- The longest 90-day zero drought was 396 spins, ending 2026-08-23, against an average zero gap of 36.6 spins; number 29 once went 469 spins without hitting.
- Auto Roulette has no multipliers and no bonus, so the multiplier field in every round record is 1.00 and the hot, cold, streak and sector widgets are the whole story.
Auto Roulette is the plainest game we track: a single-zero European wheel in Evolution's Riga studio, launched by compressed air, with no dealer, no multipliers and no bonus round. That simplicity makes its data unusually clean. Our Auto Roulette tracker has recorded 275,852 spins since 28 December 2025, about one every 51 seconds, and this page explains what each of those records holds, how to read every panel on the live page, and what 90 days of results say about the wheel. For the rules and bet types, start with the Auto Roulette guide.
How fast does Auto Roulette generate data?
About 1,690 spins a day, or one every 51 seconds, which is why the tracker has 50,658 spins from the last 30 days alone and 103,221 from the last 90.
Evolution's feed shows roughly 36 seconds from betting open to settlement, then a result and pause phase, giving a full cycle of around 50 to 60 seconds. There is no dealer to greet players, no card to read out and no bonus round to wait for, so the cadence barely changes hour to hour. Faster tables exist (Speed Auto Roulette runs a cycle near 25 seconds), but those are separate tables with separate feeds. Nothing on this page refers to them.
For a stats page the pace matters in two ways. First, sample sizes grow quickly: a week adds roughly 11,800 spins, enough for per-number shares to settle within a few tenths of a point. Second, and this is the point the strategy guide makes at length, the same pace is what makes the game expensive. Seventy spins an hour at a 2.70% house edge means the wheel takes its cut 70 times an hour, however you bet.
What does one Auto Roulette round record contain?
A winning number stored as text from "0" to "36", its color, a timestamp, and a multiplier field that is always 1.00.
That last field deserves a sentence, because it confuses people who compare Auto Roulette with Lightning Roulette. Our database uses one shape for every roulette table so that shared widgets work everywhere. Lightning tables fill the multiplier column with 50x, 100x and so on. Auto Roulette has nothing to put there, so the collector writes 1 on every row. If you ever see a page or a screenshot claiming an "average multiplier" for Auto Roulette, the number is meaningless. There are no multipliers to average.
The rest of the record is what you would expect:
| Field | Auto Roulette value | Notes |
|---|---|---|
| result | "0" to "36" | Winning number as text |
| result_details | winning number and color | red, black or green |
| multiplier | 1.00 on every row | No multipliers exist in this game |
| bonus fields | null or false | No bonus round exists |
| created_at | UTC timestamp | Basis for windows and the hourly chart |
Every panel on the live page is rebuilt from this table. With one meaningful value per spin, the whole analysis is about frequencies, gaps and runs of the winning number. That is less to look at than a money wheel offers, and harder to fool yourself with, provided you read the widgets for what they are.
Which numbers have landed most and least in 90 days?
Number 10 led with 2.78% of 103,221 spins (2,870 hits) and number 21 trailed at 2.62% (2,703 hits). Every one of the 37 pockets sat inside a 2.62% to 2.78% band around the 2.70% theory.
| Rank | Number | Observed share (90 days) | Hits | Difference from 2.70% |
|---|---|---|---|---|
| 1 | 10 | 2.78% | 2,870 | +0.08 |
| 2 | 25 | 2.77% | +0.07 | |
| 2 | 33 | 2.77% | +0.07 | |
| 2 | 1 | 2.77% | +0.07 | |
| 5 | 31 | 2.76% | +0.06 | |
| 5 | 8 | 2.76% | +0.06 | |
| ... | ||||
| 32 | 22, 29, 7, 12 | 2.64% | -0.06 | |
| 36 | 6 | 2.63% | -0.07 | |
| 37 | 21 | 2.62% | 2,703 | -0.08 |
The gap between the hottest and coldest number is 167 hits in 103,221 spins. Put differently, if you had bet one unit straight up on 10 for every spin of the last 90 days and a friend had bet on 21, you would be 167 units apart after more than a hundred thousand spins, and neither of you would know in advance which side of that gap you would land on. The hot and cold widget ranks the same numbers in whatever window you select, and the top of the list changes constantly as the window moves. It is a description of what already happened, nothing more.
The signature of a biased wheel is a persistent excess for one number or one arc that grows in absolute terms as the sample grows. A 0.08-point excess for number 10 is far inside the range a fair wheel produces by chance across 37 pockets. Our fairness index applies the same reasoning to every game we track.
What do the color, dozen and column panels show?
Red landed 48.63%, black 48.64% and zero 2.73% over 90 days, with every dozen and column within 0.15 points of its 32.43% theory.
| Bet group | Theoretical share | Observed (90 days) | Difference |
|---|---|---|---|
| Red | 48.65% | 48.63% | -0.02 |
| Black | 48.65% | 48.64% | -0.01 |
| Zero | 2.70% | 2.73% | +0.03 |
| 1-18 | 48.65% | 48.67% | +0.02 |
| Odd | 48.65% | 48.59% | -0.06 |
| 1st dozen | 32.43% | 32.49% | +0.06 |
| 2nd dozen | 32.43% | 32.29% | -0.14 |
| 3rd dozen | 32.43% | 32.49% | +0.06 |
| Column 1 | 32.43% | 32.55% | +0.12 |
| Column 2 | 32.43% | 32.41% | -0.02 |
| Column 3 | 32.43% | 32.31% | -0.12 |
These are the panels most players glance at, and they tell the same story from every angle: the ball is landing where a fair single-zero wheel says it should. Do not read the color and dozen shares as an observed RTP. Since every bet on this table pays true odds less 1/37, the shares and the return are the same fact seen twice, and a 48.63% red share against a 48.65% theory is sampling noise rather than a sign the edge has moved. The RTP page covers why the return is 97.30% on every spot and how our RTP monitor reports it.
The 2nd dozen at 32.29% is the largest gap in the table, 0.14 points short, or about 145 hits under expectation. Even "close to theory" leaves room for a swing a regular dozen bettor would have felt.
How is the sector heat panel different from hot and cold numbers?
It groups the wheel by physical position rather than by number, using the three racetrack sectors, and over 90 days all three sat within 0.4 points of theory: Voisins du Zero 45.62%, Tiers du Cylindre 32.61%, Orphelins 21.77%.
Hot and cold numbers are a list. Sector heat is a map. On a European wheel the numbers run 0, 32, 15, 19, 4, 21 and so on around the rim, and the racetrack bets cover contiguous arcs of that order: Voisins (17 numbers around zero), Tiers (12 numbers opposite it), Orphelins (the 8 numbers in the two leftover gaps). If a wheel had a mechanical lean, this is the panel where it would show, because a tilt favors an arc rather than a scattered set of numbers.
| Sector | Numbers covered | Theoretical share | Observed (90 days) | Difference |
|---|---|---|---|---|
| Voisins du Zero | 17 | 45.95% | 45.62% | -0.33 |
| Tiers du Cylindre | 12 | 32.43% | 32.61% | +0.18 |
| Orphelins | 8 | 21.62% | 21.77% | +0.15 |
Voisins ran slightly cold and the other two slightly warm, which is the pattern you get when a few numbers in the zero neighbourhood (21 and 6 among them, both in the bottom five) happen to be under-hit in the same window. Shift the window forward a month and the sectors reshuffle. The streaks, sectors and hot/cold page walks through the sector bets in more detail and why they are equivalent to any other bet on the table at 97.30%.
How does the streak detector treat colors and repeats?
It reports the current run of one color and the longest on record: over 90 days the longest red run was 16 spins, the longest black run 13, and the longest run of zeros was 3.
Runs are where intuition fails most reliably. The average run of a single color is about 1.95 spins, so most sequences flip every couple of spins and a run of five feels long. The probability that a run reaches k spins from a given start is (18/37) to the power of k minus 1. For 16 reds that is about 0.0021%, which sounds impossible until you multiply by the number of starting points in 103,221 spins: it comes to roughly two expected occurrences per 100,000 spins. Our 16 is exactly the sort of maximum a fair wheel should produce in this much data.
Repeats follow the same rule: the next spin matches the current number with probability 1/37 = 2.70%. Over 90 days the same number landed twice in a row 2,716 times, or 2.6% of spins, three in a row 85 times, and the longest identical run was 4. The three-zero run has a per-start probability near 0.002%, which is again about two per 100,000 spins; we saw one.
The point of this widget is not to tell you a run is ending. It cannot. A run of 16 reds says nothing about spin 17, which is red with probability 18/37 like every spin before it. Its value is the opposite: it shows how long a fair wheel can go one way, so that a system built on "black is due after six reds" is exposed before it costs money. The strategy page works through what a doubling sequence costs when a run like this arrives.
How should I read the zero drought widget?
It shows spins since the last zero against an average gap of 36.6 spins; the longest gap in 90 days was 396 spins, ending on 2026-08-23, and two more droughts of 277 and 270 spins both ended on 2026-07-25.
The median gap is 24 spins, well below the average, because most gaps are short and a few very long ones pull the mean up. This shape is what a geometric distribution looks like, and it is the reason a zero drought "feels" overdue long before it is statistically unusual. The table below gives the probability that zero, or any particular number, stays away for N consecutive spins.
| Spins without a given number | Probability | How often it should happen |
|---|---|---|
| 37 | 36.3% | Routine, several times a day |
| 100 | 6.5% | Several times a week per number |
| 200 | 0.42% | A few times a quarter per number |
| 396 (our longest zero drought) | about 0.002% | Rare for one number, expected somewhere across 37 |
| 469 (number 29, our longest of all) | 0.0003% | Same reasoning |
The last column is the part that matters. There are 37 numbers, each capable of a drought, and 103,221 spins in which to have one. A 400-plus gap for at least one number somewhere in that data is not a surprise; it is close to a certainty. The widget counts the wait for zero because zero is the number players watch, but the maths is identical for 29, which went 469 spins without landing.
What the drought widget cannot do is shorten the next wait. Zero's probability on the next spin is 1/37 whether it last hit 5 spins ago or 390. The gambler's fallacy is the belief that a long absence raises that probability, and on an automated wheel with a fresh launch every spin there is no mechanism by which it could. Our observatory tracks droughts across games for exactly this reason: they are the moments when players are most tempted to chase.
Does the time of day change anything?
Not the odds. Hourly spin counts over the last 30 days peaked at 15:00 UTC, with 10:00 and 13:00 to 16:00 the busiest hours, but the differences are marginal because the wheel spins at the same pace whether ten people or ten thousand are betting.
The hourly chart mostly confirms that the table never stops. Nothing in our data suggests any hour produces more zeros, longer streaks or a different color balance, and there is no reason to expect it: the launcher does not know what time it is.
If you want to match a session you played to the tracker, note that every timestamp is UTC. A session at 21:00 in Central European Summer Time appears at 19:00 in our data, and "today" on the 24-hour window starts at midnight UTC rather than your local midnight.
Where does the data have gaps?
The collector was offline from 2026-06-30 to 2026-07-09, a nine-day hole inside the current 90-day window.
That gap does not distort the shares: missing spins are missing at random with respect to outcome, so a number's percentage over the remaining spins is unbiased. It does distort anything measured per day. Spins per day for that stretch read as zero, and the 90-day total of 103,221 is below what an unbroken window would hold. If a chart looks like the wheel stopped for a week at the start of July, that is our collector, not Evolution.
Two other limits apply. First, results arrive from the feed within minutes of the spin, not instantly, so the last spins strip can trail the stream you are watching by a short margin. Second, the tracker follows the standard Auto Roulette table only. Evolution also runs Speed Auto Roulette, Auto Roulette La Partage and VIP variants; those are different wheels with different feeds and will not match what you see here. Some of those variants apply the la partage rule, which returns half of an even-money stake when zero lands and lifts even-money RTP to 98.65%. The standard table we track does not, so every bet stays at 97.30%.
Official payouts and the 97.30% figure come from Evolution's in-game rules; Evolution's roulette page is the reference. Beginners looking for the layout and the racetrack should read the how to play guide before using the tracker.
How do I use Auto Roulette stats without fooling myself?
Treat every widget as a description of the past and budget for the variance it reveals, rather than betting against it.
The honest use of this page is calibration. Knowing that a color can run 16 deep tells you how much a doubling system can cost before it hits the table limit. Knowing that zero can hide for 396 spins tells you what a "zero cover" strategy has to survive. Knowing that the hottest number in a quarter beat the coldest by 167 hits tells you how little a hot-number list is worth. The wheel has no memory, and every bet on it carries the same 2.70% edge, so the only variables under your control are how many spins you play and how much you stake on each.
At 70 spins an hour, that first variable is the one that matters. Set a stake you are comfortable losing in full, decide on a number of spins before you sit down, and let the tracker be a record rather than a signal. If a session starts to feel like chasing a drought or riding a streak, close the table; support for anyone who finds that hard is linked from our observatory.
Frequently asked questions
How many times has zero come up on Auto Roulette?
Over the last 90 days zero landed on 2.73% of 103,221 tracked spins, a fraction above its 2.70% theoretical share. The average gap between zeros was 36.6 spins with a median of 24, and the longest wait was 396 spins, ending on 23 August 2026. Three zeros in a row occurred once. The zero drought widget on /games/auto-roulette shows the current gap live.
What are the hot numbers on Auto Roulette right now?
The hot and cold panel on our tracker lists the numbers that landed most and least in the window you select. Over the last 90 days the top of the list was 10 at 2.78%, followed by 25, 33 and 1 at 2.77%; the bottom was 21 at 2.62%. Those ranks reshuffle constantly and carry no information about the next spin, which has a 1-in-37 chance for every pocket.
How do I see Auto Roulette results history?
Open /games/auto-roulette. The last spins strip shows recent winning numbers with their color, and the window selector lets you view 24 hours, 7, 30 or 90 days, or everything since 28 December 2025. The in-game panel only holds a short run of recent results, whereas the tracker keeps 275,852 spins and rebuilds every widget from the full record.
Is Auto Roulette rigged or biased?
Our 90-day data gives no sign of bias. Every one of the 37 numbers landed between 2.62% and 2.78% of the time against a 2.70% expectation, the red/black split was 48.63% to 48.64%, and the three racetrack sectors all sat within 0.4 points of theory. The spread between the most and least frequent number was 167 hits across 103,221 spins, which is what a fair wheel produces.
What is the longest streak of red on Auto Roulette?
The longest red run in the last 90 days was 16 spins in a row; the longest black run was 13. A run of 16 reds has a probability of about 0.0021% from any given starting spin, which works out to roughly two occurrences per 100,000 spins, so one in our 103,221-spin window is exactly on schedule. The streak detector on the tracker shows the current run and the record.