Skip to content
LearnMathora

Probability & statistics · 01 · Taming chance · 8 min

Thinking in probability

You can't predict one coin flip — and yet you can predict a million of them with eerie precision. Probability is the discovery that randomness, viewed from far enough back, has rules.

Build the intuition

Probability is long-run frequency

“The coin is 50% heads” means: flip it forever and the proportion of heads settles to one half. Single events stay wild; proportions calm down. That settling — the law of large numbers — is why casinos always profit eventually while any single gambler might win tonight.

P(heads)=12P(\text{heads}) = \tfrac{1}{2}

The two combining rules

Independent events both happen: multiply (two heads in a row: ½ × ½ = ¼). Mutually exclusive events, either happens: add (a die showing 1 or 2: ⅙ + ⅙ = ⅓). Most probability puzzles are these two rules plus careful reading of which applies.

Expected value: the long-run average

A lottery ticket costs $2 with a 1-in-a-million shot at $1M. Expected value: $1 per ticket — you pay 2 to receive (on average) 1. Expected value is how casinos price games, insurers price policies, and you can price decisions: it's what “on average” means, made exact.

EV=(outcome×probability)EV = \sum (\text{outcome} \times \text{probability})

See it move

InteractiveThe law of large numbers
40
After 40 flips: 23 heads = 57.5%. Single flips are pure chance — but the running average is drawn, inevitably, toward 50%. Randomness has long-run structure.

Early flips swing wildly — 70% heads happens. Add flips and watch the proportion get pulled, inexorably, toward 50%.

A worked example

The birthday surprise

  1. In a class of 23, what's the chance two people share a birthday? Intuition says tiny.

  2. Flip the question: chance all 23 differ. Multiply the dodges:

    365365×364365××3433650.49\tfrac{365}{365} \times \tfrac{364}{365} \times \cdots \times \tfrac{343}{365} \approx 0.49
  3. So a shared birthday has probability ≈ 51% — better than a coin flip. Pairs grow fast: 23 people make 253 pairs.

  4. Probability's job: replacing astonishment with arithmetic.

Out in the world

What “70% chance of rain” means

It means: of all days with conditions like today, about 70% saw rain. The forecast isn't hedging — it's reporting a long-run frequency. Reading probabilities as frequencies is the single most useful statistical habit.

Common confusion, cleared

After five heads, tails is due.

The coin has no memory — flip six is still 50/50. The law of large numbers works by swamping early streaks with volume, not by correcting them.

A 1-in-a-million event basically never happens.

To you, rarely. Across 300 million people, hundreds of times a day. Rare-at-scale is common — the insight behind both lotteries and safety engineering.

Recap

  • Probability = long-run frequency; singles are wild, proportions settle.
  • Multiply for independent “and”; add for exclusive “or”.
  • Expected value prices uncertain prospects: outcome × probability, summed.