How roulette works
Roulette has a wheel divided into numbered pockets and a ball that eventually lands in one of them. Each time the wheel and ball are set in motion and produce a result, one spin is completed.
A European roulette wheel has 37 pockets: the numbers 1 to 36 and zero. Of the numbers 1 to 36, eighteen are red and eighteen are black. Zero is green.
The table layout allows a bet to be associated with a particular result or a group of results. For example, a bet on one number depends on that number appearing; a bet on a colour depends on any of its eighteen numbers appearing.
If the bet wins, the payout depends on the type of bet. Otherwise, the amount staked is lost. The relationship between the chance of winning and the payout for doing so is what explains the casino’s advantage.
In the examples below, we assume a balanced wheel on which every pocket has the same chance of receiving the ball.
Why red and black are not a 50% chance
It is easy to think that, because there are two main colours, each has half the chances. But that leaves out zero.
| Result | Pockets on a European wheel | Approximate probability |
|---|---|---|
| Red | 18 out of 37 | 48.65% |
| Black | 18 out of 37 | 48.65% |
| Zero | 1 out of 37 | 2.70% |
Red and black have equal chances relative to each other, but neither reaches 50%. Some of the possible outcomes belong to zero, which is neither red nor black.
These percentages are not a timetable. They do not mean that a particular number of spins must contain exact counts of reds, blacks and zeros. They describe the chances of each result before a spin.
The usual American roulette wheel adds a double zero, marked “00”. It has 38 pockets: eighteen red, eighteen black and two green. This means that figures for one version cannot simply be applied to the other.
Where the casino’s advantage comes from
Consider a fictional example of one unit staked on red. A unit is simply an amount used for the calculation, not a suggested sum of money.
Under the rules of this example, a red result returns the unit staked and pays one additional unit. The profit is one unit. If black or zero appears, the unit staked is lost.
On a European wheel, this gives us:
- 18 pockets that produce a profit of one unit.
- 19 pockets that produce a loss of one unit.
To see the imbalance, we can make an imaginary comparison: count each of the 37 possible results once. There would be 18 units won and 19 lost. The balance would be a loss of one unit.
This does not mean that each pocket will appear once in 37 spins. It is a way of comparing all the possible results, which have equal chances in our example.
That difference of one unit, spread over 37 units staked, is approximately 2.70%. This is the house edge in this case.
With zero and double zero, the same example has 18 winning outcomes and 20 losing ones. The difference is two units out of 38: approximately 5.26%.
These figures apply to the payouts and treatment of zero just described. Special rules can change what happens to certain bets when zero appears, and other bets may have different payouts.
What losing 2.7% on average actually means
The casino’s advantage does not work like a fee deducted in an exact amount from every session.
Someone may finish ahead, lose a little or lose a lot. The 2.70% describes a mathematical average for the previous example over many repetitions, not the result that any particular person must get.
The amount it applies to also matters: it is the total amount staked, not just the starting money.
Imagine the same unit being used ten times because winnings and returned stakes make it available again. Even though ten new units have not been added, that unit has been involved in ten bets. The total amount staked counts each of those occasions.
For a total of 100 units staked under the conditions of the example, the mathematical average loss would be approximately 2.70 units. A real session does not have to end with that loss.
This distinction explains why winning for a while does not contradict the house edge. Individual results can vary widely while the rules continue to favour the casino.
It also does not mean that continuing to play will make up for losses. Every new bet is subject to the same imbalance.
Why results are not perfectly balanced
Now consider a fictional sample of twenty spins:
1, 2, 3, 4, 5, 6, 7, 8, 10, 0, 1, 3, 5, 7, 9, 12, 14, 16, 2, 4.
This sequence was constructed to explain the example; it is not a real session. Counting the colours gives us:
| Result | Number of appearances | Percentage of the sample |
|---|---|---|
| Red | 12 | 60% |
| Black | 7 | 35% |
| Zero | 1 | 5% |
Red appears in 60% of these spins, even though we previously calculated its probability at approximately 48.65%. There is no contradiction.
Probability describes the chances before a result happens. The observed percentage counts what has happened in a particular set of spins. This count is also called frequency.
Twenty spins can have an uneven distribution. That difference alone does not show that the wheel has a problem or that red has become more likely.
On a balanced wheel, percentages tend to approach their probabilities as very large numbers of spins accumulate. But this approach need not be steady, and it does not mean that the next results must correct the previous ones.
After several reds, is black due?
After several reds in a row, it can feel as though black is due. But the wheel does not keep track of what has appeared or have a distribution to complete.
If spins are independent, as our example assumes, previous results do not change the chances on the next spin. There are still eighteen red pockets, eighteen black pockets and one zero.
The opposite idea does not work either: a run of reds does not show that red will keep appearing.
The same applies to a number that has not appeared for many spins. That absence describes its past, but does not by itself increase its chances. On a balanced European wheel, it still has one chance out of 37 on the next spin.
Thinking that a result is “about to happen” because it has been absent for a while is a mistake known as the gambler’s fallacy. The name matters less than the idea: chance does not have a debt to repay on the next spin.
What statistics can and cannot tell us
Statistics help organise a record that would otherwise be difficult to interpret. We can count colours, observe repetitions or check how many spins have passed since a number last appeared.
A streak is a sequence of results that share a characteristic: several reds in a row, for example.
So-called hot numbers are those that appear most frequently in the sample being examined. Cold numbers, under the definition used by RouletteLab, reflect how long they have been absent within that sample. These names describe the record; they do not indicate which result to expect next.
The picture also changes depending on which part of the record is examined. In our fictional sequence:
| Part of the record | Red | Black | Zero |
|---|---|---|---|
| All 20 spins | 12 — 60% | 7 — 35% | 1 — 5% |
| Only the last 10 | 8 — 80% | 2 — 20% | 0 — 0% |
Both counts are correct. The second row focuses on a stretch with more reds and leaves out other spins.
Looking only at that stretch can make a trend seem more important. That is why it matters to know how many results a statistic contains and which part of the record it describes.
Neither row changes the chances of the next spin. Nor is this sample enough to demonstrate a pattern that can be exploited.
Why a strategy does not remove the house edge
If past results cannot tell us what comes next, another question may arise: does changing the amounts make a difference?
Systems such as Martingale and Paroli vary the amount staked after particular results. The classic Martingale increases it after a loss; Paroli proposes increasing it after a win, within a sequence.
This changes how wins and losses accumulate and how much money is exposed. It does not change the pockets on the wheel or the payout associated with each type of bet.
A larger amount can produce a larger win, but also a larger loss. A sequence that seemed to work can end with an unfavourable result that wipes out earlier gains.
The money available and the table limits are also finite. A system that depends on continuing to increase amounts can reach a point where it is impossible to carry on.
Changing bet sizes does not, by itself, turn the unfavourable rules of the example into favourable ones. Understanding these systems requires looking at what happens when a sequence is interrupted, not just at successful cases.
What analysing a session is useful for
Analysing a session helps check what happened more accurately than memory alone.
A striking streak can make us think that one colour dominated the entire session. Counting the results may reveal that this impression came from one particular stretch. We can also distinguish a real repetition from an exaggerated memory.
RouletteLab allows spins to be recorded and explored through statistics and visual displays. Its purpose is to organise the past and make it easier to understand, not to identify the next number or turn a streak into a recommendation.
Understanding roulette statistics also helps put stories about systems that “always work” into perspective. A favourable session describes what happened on one occasion; it does not show that the conditions giving the casino an advantage have disappeared.
A record can answer the question “What happened in these spins?” That answer is not the same as knowing what will happen on the next one.