Coin flip
Flip one coin or a thousand, with running streak statistics.
Ready
How it works
Each flip is an independent draw from the browser's cryptographic random source. There is no memory: a coin that has come up heads five times running is still exactly 50% to come up heads next.
The gambler's fallacy
The belief that tails is “due” after a run of heads. It is not — the coin has no state. What is true is regression to the mean: as the number of flips grows, the proportion approaches 50% while the absolute difference between heads and tails tends to grow, not shrink. Both readouts are above so you can watch it happen.
Streaks are longer than intuition suggests
In 100 flips, a run of six or more of the same side is more likely than not. People asked to fake a random sequence almost always produce too few long streaks — which is exactly how fabricated data gets detected.
Real coins are not quite fair
Diaconis, Holmes and Montgomery showed in 2007 that a physically flipped coin lands on the same face it started on about 51% of the time, because it precesses rather than spinning cleanly. Spinning a coin on a table is worse still — the mass distribution around the rim can push it well away from even. This simulation has no such bias.