Five mistakes in thinking about lotteries

The most common mistakes in thinking about lotteries boil down to one assumption: that drawings are connected to each other. They are not. The balls don't know what came out a week ago, the machine keeps no records, and an "overdue" number has no mechanism that would force it to return. Below are five variants of the same misunderstanding — each with a number that settles it.

The gambler's fallacy: a number that "must come up already"

This is the oldest of these mistakes and it even has its own date. In August 1913, at the Monte Carlo casino, the roulette ball landed on black twenty-six times in a row. Players bet on red with increasingly larger sums, convinced that after so many blacks, red was "due". They lost millions of francs. The wheel had no memory of the previous twenty-six spins — it only had the next spin.

In Lotto this works identically. A number that hasn't come up in fifty drawings has, in the next drawing, exactly the same probability of appearing as a number that came up yesterday: six chances out of forty-nine, or about 12.2 percent. Being overdue is a description of what hasn't happened. It is not a forecast.

The word "must" is misleading here. Over a large number of drawings, each number does indeed approach the average frequency — that's the law of large numbers, and it does work. But it works through dilution, not through equalization. A historical deviation is not compensated by a series of hits; it stays where it was, it just stops being visible in the growing denominator. Statistics, including those in our Lotto statistics overview, describe the past and nothing beyond it.

Does the drawing machine remember anything?

No. A drawing machine is a drum with balls of similar mass and size, set in motion by an air stream or a mixer. After each drawing, the balls return to the drum as a full set. The starting state of the next drawing is the same as the previous one — there is no element of the system in which information about the previous result could be stored.

Formally, drawings are said to be independent: the probability of outcome B after outcome A equals the probability of B alone. This is not a simplifying assumption adopted for the convenience of calculation. It is a description of the device's construction.

From this follows a conclusion that many players consider controversial: the set 1, 2, 3, 4, 5, 6 has exactly the same probability as any other specific set of six numbers. It looks "unrandom", but that's an aesthetic judgment, not a statistical one. The only practical difference is different: such a set is crossed off by many people at once, so if it were drawn, the prize would be split among many parts.

Hot and cold numbers — how much does the statistical difference matter

A "hot" number is one that has come up more often than average in a given period. The question is whether the difference is large enough to indicate anything beyond chance. Let's calculate this for Lotto.

In Lotto, six numbers out of forty-nine are drawn in each drawing. With a thousand drawings, six thousand balls appear in the pool, so the expected number of hits for a single number is 6000 divided by 49, or about 122. That's the average. Now the spread: for a binomial distribution with n = 1000 and p = 6/49, the standard deviation equals the square root of n·p·(1−p), or about 10.4.

Number of drawings	Expected hits	Standard deviation	Typical range (±2σ)
100	12.2	3.3	6 to 19
500	61.2	7.3	47 to 76
1,000	122.4	10.4	102 to 143
5,000	612.2	23.2	566 to 659

Binomial distribution for Lotto, p = 6/49 ≈ 0.1224. The ±2σ range covers about 95 percent of cases in a fully random drawing.

Let's read the row with a thousand drawings. A number that came up 143 times with an average of 122 looks clearly "hot" in a ranking — it's a fifth above average. Yet it falls within the typical spread we expect from pure chance. To suspect anything other than chance, the difference would have to be much larger and persist as the number of drawings grows — and as n grows, the ±2σ range grows more slowly than the average itself, so percentage deviations shrink.

The practical consequence: a ranking of numbers from most to least frequent always exists, even for data generated by a perfect random generator. Someone has to be first. The fact that a number sits at the top of the list doesn't, by itself, carry information about the future. It's worth seeing the distribution — that's what number heatmaps are for — but as a picture of history, not a forecast.

Does a reduced system increase the probability of hitting the six?

No. A system guarantees a hit structure, not the hit itself. This distinction decides everything and is often blurred in advertising descriptions.

A full system on ten numbers is all possible six-element bets from that ten: two hundred ten tickets. If all six drawn numbers fall within your ten, the six is certain, because one of those two hundred ten bets is exactly that set. A reduced system chooses a subset from those two hundred ten — say twenty tickets — selected so that if six numbers are hit within the ten, at least a four is guaranteed. For a twenty-fifth of the price, you get a guarantee of a lower tier.

The key thing is what remains in the condition: "if all six fall within your ten". The probability that this happens results solely from the fact that you bet on ten numbers out of forty-nine, not from how you arranged them into tickets. It equals the ratio of the number of six-element combinations from ten to the number of six-element combinations from forty-nine, i.e. 210 to 13,983,816 — roughly one in sixty-six thousand. The system doesn't move this number by a hair.

A system is therefore a tool for managing the cost and distribution of lower-tier wins, not a tool for raising the odds. A description of the construction and examples can be found in our overview of reduced and full systems. If anywhere you read that any way of crossing off numbers raises the probability of hitting the six, you're dealing with advertising material.

Why does someone always win, if the odds are so small

This is the fifth mistake and it works in the opposite direction to the previous ones: here we don't overestimate our own odds, we confuse them with the odds of the whole group. The probability of the six from a single bet in Lotto is one in 13,983,816 — that's the number of six-element combinations out of forty-nine. This number doesn't change because someone else is playing.

At the same time, when several million bets take part in one drawing, the probability that at least one of them hits all six numbers becomes quite decent. These are two different questions about two different events: "will I win" and "will anyone win". News of a win answers the second, and intuition transfers it to the first.

On top of this comes what isn't visible in the media. A headline is produced after every six, and never after any of the millions of tickets that were lost. This creates a picture in which wins seem more frequent than they are — not because anyone is lying, but because losses have no news genre of their own.

The scale of this number is hard to imagine and it's worth translating it into something concrete. If all 13,983,816 combinations of Lotto were printed, one per line, the resulting stack of paper would be several hundred meters tall. Your ticket is one line in that stack.

What this means for the player

None of the points above lead to the conclusion that you shouldn't play. They lead to the conclusion that playing is entertainment with a known price up front, not a financial strategy. Everything that promises an edge — analyzing overdue numbers, choosing hot numbers, buying "proven" sets — rests on one of the five mistakes described above.

In practice, two things remain that you have control over. The first is the amount you set aside — fixed in advance and not increased after a series of losses, because that's exactly the gambler's fallacy in action. The second is avoiding popular combinations, if you care about not splitting the prize should you win: birth dates limited to 31, arithmetic sequences, patterns from the ticket. This won't change the probability of hitting, but it will change the size of any eventual win.

If you want a set free of your own habits, the random shot-in-the-dark generator will do. It won't increase your odds — nothing increases them except the number of bets — but it takes off your hands a decision that has no statistical significance anyway.