Random

Coin flip

A virtual coin with 3D animation, result history and best-of-N mode. Decide what you have to decide — fair and random.

Instant🔒In your browserNo signup
Live
HEADS
TAILS
Hit "flip"
History: — heads / — tails

The invisible bias of real coins

Even though "flip a coin" is shorthand for perfect randomness, studies show real coins have a bias: they land slightly more often (~51%) on the side that started face-up. Physicist Persi Diaconis published a 2007 analysis demonstrating the effect. A virtual coin doesn't have that bias: it's exactly 50/50.

How our generator decides

We generate a random number with crypto.getRandomValues. Even means heads, odd means tails. Distribution is statistically uniform, verifiable with any chi-square test: flip it 10,000 times and you'll see a very close to 5000-5000 split.

What flipping a coin is good for

  • Trivial binary decisions: pizza or sushi? Run today or skip?
  • Tie-breaking: two people want the same chair / dessert / task.
  • Game order: who starts a match, who serves first in tennis.
  • Education: demonstrate probability, binomial distribution, the law of large numbers.
  • Random role assignment: team A or team B, short or long presentation.

Best of N: when it helps

If a single flip feels too thin, try "best of 3" or "best of 5". Statistically each individual flip is still 50/50, but the set winner carries more confidence: a "best of 5" decision feels less random while remaining fair. In sports this is formal: 3-set tennis, 7-game NBA finals.

The gambler's fallacy

If the coin lands heads 5 times in a row, many people believe "tails is due now". It isn't: each flip is independent, the probability stays 50%. This fallacy is documented from 1796 and is why many lose money in casinos. Randomness has no memory.

The "rigged coin" trick

There are ways to make a physical coin land one way: spin instead of flip, catch in your hand vs let it bounce, etc. With a baseline ~51% bias plus a trained magician, a coin can hit 80% one way. The virtual version is immune to that: you can't influence crypto.getRandomValues.

When NOT to use a coin

  1. Important decisions with real consequences: jobs, moves, breakups. A coin doesn't add information; it only decides for you when you already know.
  2. Raffles with monetary value: you need a certified system and ideally a notary.
  3. When one option is objectively better: don't hand the decision to a coin if it isn't truly even.

FAQ

Is it 50/50?

Yes. crypto.getRandomValues with uniform distribution. Real coins bias ~51%; ours doesn't.

What is it for?

Quick binary decisions, breaking ties, game order, demonstrating probability in class.

Best of N?

Yes. Flips N times and the most frequent side wins. Doesn't change individual odds, but the set has more certainty.

Works offline?

Yes. Once the page loads everything runs in the browser.

Was this generator useful?