The birthday paradox
Why just 23 people are enough for a roughly even chance of a shared birthday.
A captured spark. Unverified, unpolished, possibly wrong.
The birthday paradox sounds unlikely the first time you hear it: in a group of 23 people, the chance that at least two share a birthday is just over 50%. There are 365 possible birthdays in the usual model, so 23 seems like a very small group.
The part that’s easy to miss is which question we’re asking. Looking for someone who shares your birthday is different from looking for any two people who share one. In the second case, every person can be compared with everyone else.
With 23 people, there are 23 × 22 ÷ 2 = 253 pairs. That helps explain why the probability grows so quickly, although the pair comparisons aren’t all independent, so treating them as 253 separate coin flips wouldn’t give the right calculation.
Working it out
It’s easier to calculate the chance that everyone has a different birthday, then subtract that from 1.
The first person can have any birthday. The second has 364 days available if they are to avoid a match, the third has 363, and so on. For 23 people, the calculation is:
P(no shared birthday) = (365/365) × (364/365) × ... × (343/365)
P(at least one match) = 1 - P(no shared birthday)
≈ 0.5073
That’s about 50.7%. With 70 people, the chance of at least one match is about 99.9%.
These numbers assume birthdays are independent and equally likely across 365 days, leaving out leap days. Real birthdays don’t follow that model exactly, but it’s a useful way to understand the effect.
Why I like this example
The interesting part for me is how much the answer changes when I count the opportunities for something to happen properly. A coincidence can look remarkable if I only notice the eventual match and forget all the other possible comparisons.
That doesn’t make every coincidence uninteresting. It gives me a question to ask before deciding one needs an unusual explanation: how many chances were there for a match?
It’s a useful habit to carry into other probability questions I’m working through, where my first impression of the odds may leave out more than I realise.
Paths that lead here
No paths yet. This note is still off the beaten track.
Where this note points
- How much time should I give a bug hunt? · I'm wondering whether the Kelly criterion can help me think about research time, and where that comparison breaks down.
More from these beds
- Learning about quant trading · I want to learn the maths and testing behind trading, while keeping real money out of the experiment.
- Thinking about right and wrong · The Good Place got me thinking about outcomes, intentions, and how we decide what a good action looks like.
- Understanding trading edge and risk · Two ideas I'm trying to understand as a beginner: whether a strategy has an advantage, and how much it can lose.