Do lucky numbers work?

Short version first, because burying it would be rude: no. Lucky numbers don't work. There is no line the draw favours and nothing you write on a ticket that the balls can respond to. A birthday, a shirt number, a set you've played every week for twenty years — the machine has no idea, and no mechanism for having one.

That isn't the whole story, though, and the missing half is the interesting bit. Your choice of numbers cannot change your chance of winning. It can change how many other people are holding the same line as you — and therefore how thinly a jackpot gets sliced if it ever lands on it. That effect is real, and it's the only part of number selection worth any thought at all.

Every combination is equally likely

Start with the fact everyone half-accepts and nobody quite believes: 1, 2, 3, 4, 5, 6 is exactly as likely to come out of a Lotto machine as any wild-looking scatter of six numbers. Not nearly as likely. Not likely-ish. Identical.

The reason is dull and total. A 6-from-59 draw produces one of 45,057,474 possible sets, and a fair draw gives each the same chance — that's what "fair" means. Ordered-looking sets get no penalty, because the machine cannot see order. It isn't reading numbers, it's tipping out balls. The arithmetic behind that figure is in how lottery odds are calculated.

So why does 1-2-3-4-5-6 feel absurd? Because we don't judge a combination against the other 45 million; we judge it against a category. There is one perfectly ascending run of six, and there are millions upon millions of untidy-looking lines. Untidy lines win essentially every draw, so untidiness feels normal and tidiness feels rigged. But "an untidy line" isn't a bet — it's a description of a huge bucket of them. Bet on one specific untidy line and you're in exactly the same boat as the person who bet on 1-2-3-4-5-6.

Our brains treat visible pattern as evidence that something made it. That instinct is useful when you're working out whether a rustle in the grass is the wind. It's useless here, and it's the root of nearly every wrong idea about lottery numbers.

The number that's "due"

The next belief along: a number hasn't appeared for fifty draws, so it must be about to. It's owed.

Ask the obvious question — owed by what? For a number to be due, something would need to be keeping score: a record of what's come out lately, and a way of nudging the next draw to compensate. No such thing exists. There's a set of balls and a machine that mixes them. The balls don't know what they did last Saturday.

It's the coin fallacy in fancier clothes. Five heads in a row, and tails starts to feel overdue — but the coin has no memory either, and the next flip is 50/50 like every one before it. What happens over a long run isn't correction, it's dilution: the early imbalance stays exactly where it is and gets swamped by everything after. Nothing rebalances. The books just get bigger.

The fallacy runs both ways, and people cheerfully hold both versions at once: avoiding a number because it just came up, chasing another because it's hot.

Hot and cold numbers

Frequency tables are the respectable face of all this. They look like data because they are data — real counts of real draws. The trouble is what people do with them.

Here's what defuses the whole subject: uneven frequencies are exactly what you should expect. A pool of 59 numbers, and only a few thousand draws ever held. That's a small sample spread across a lot of slots, and at that scale pure chance produces visible lumps every time. Truly random data isn't smooth. Smooth would be the suspicious result.

And someone has to be top of the table. That isn't a discovery about the machine; it's a property of any list sorted by count. Wipe the records, run the draws again from scratch, and a different number would lead by an equally convincing margin — a table just as meaningful-looking, and just as empty.

Then the practical objection. Suppose a ball really were heavier, or a machine subtly skewed. To exploit it, the bias would have to be big enough to see through all that noise — and a bias that big is one the operator's own equipment testing is far better placed to catch first, with better instruments and a considerable incentive. So either the draw is fair, or it's broken and gets fixed. Neither leaves you an edge.

Nobody is good at being random

Here's where something genuinely does behave predictably — not the machine, us.

Asked to produce a "random" set of numbers, people reliably do the same things. They avoid adjacent pairs, because 23 and 24 together look like a mistake. They avoid repeating a decade, so having taken 41 they steer clear of 45. They drift to the middle of the range, and spread their picks evenly across the grid, because clustering feels like bad randomising.

Real draws don't behave like that. Genuine randomness clumps. Consecutive numbers turn up far more often than intuition allows, tight clusters are routine, and whole stretches of the range come up empty. A properly random line often looks slightly wrong — which is a decent sign that it's random.

None of which makes a hand-picked line less likely to win — every line is equally likely, hand-picked ones included. It means hand-picked lines are predictable to other humans. Which is where number choice finally bites.

What number choice actually changes: your share

Jackpots are divided between everyone holding the winning line. Two winners, half each. Ten winners, a tenth each. The prize doesn't scale up to meet demand — it gets cut into more pieces. So the popularity of a line matters: not to whether it comes up, but to what it's worth if it does.

Dates are the big one. A day of the month gives you 1 to 31 and nothing higher; a month gives you 1 to 12. Anyone building a line out of family birthdays is drawing from a truncated pool by definition, and the top of the range never gets a look in. Because so many people play meaningful dates, tickets as a whole lean low — and lean hardest of all on 1 to 12, which days and months both reach.

Nearly half of Lotto's pool sits above 31, invisible to birthdays. The top of the range is quieter territory — not luckier, just less crowded. (EuroMillions Lucky Stars run 1 to 12, entirely within reach of months, so there's nothing to dodge there.)

Then there are the deliberately clever choices, which are the trap dressed up as the solution. A straight line down the ticket grid. A neat diagonal. 1-2-3-4-5-6, chosen precisely because it's assumed to be underplayed. Multiples of seven, or any tidy arithmetic sequence. Each is picked by a great many people who have independently decided they're being original — and originality doesn't survive being obvious. A line that thousands reach by the same reasoning is a thousand-way split waiting to happen.

Now the distinction this whole article exists for, as plainly as I can put it. Avoiding crowded numbers does not improve your chance of winning. It cannot. Your odds are fixed by the format — 1 in 139,838,160 for the EuroMillions jackpot, whatever you write down. What it improves is your expected share, if you win at all: a slice of an almost impossible event, made slightly larger. That's the entire benefit, and anyone selling it as more is selling you something.

So should you play your lucky numbers?

Honestly? If you want to, yes.

The case for it isn't a consolation prize. The odds are long enough that what you're really buying is a few days of thinking about what you'd do with it, and if your kids' birthdays are the part that makes it feel like yours, then that is the product. UK games return roughly half of what they take as prizes, so you're paying for the experience either way. Trading a bit of that pleasure for a slightly better slice of a near-impossible event is not an obviously good deal.

The case for a random pick is just as simple: it sidesteps the clustering without you having to think about it, and stops you spreading six numbers too neatly and calling it chance. If no particular numbers mean anything to you, there's nothing to lose by letting something else choose.

What isn't true is that either is smarter. For winning they're identical. Pick whichever you'd rather have — then set a budget you won't mind losing, because that decision matters a great deal more than the six numbers do.

Read next

• How lottery odds are calculated →
• EuroMillions guide →
• Lotto guide →
• Responsible gambling →

Lucky Dip is for entertainment, and for over-18s. If gambling has stopped being fun, free and confidential help is available.