Interactive law-of-large-numbers lab
Coin Flip Probability Experiment
Run 10, 100, 1,000, or a custom number of independent flips. Compare counts, deviation, streaks, and the cumulative proportion across repeated runs.
Watch a proportion, not a promise
The expected probability of Heads is 50% on each trial. A particular run does not have to end at 50%, and the cumulative line can move in either direction along the way. Larger runs usually show a smaller proportional deviation, but exceptions remain possible.
Streaks are normal
Independent does not mean alternating. The longest-streak value helps make clusters visible; a long run of one side is not, by itself, evidence that the source is biased.
Compare independent runs
Repeat the same trial count several times. Each row is generated from new browser Web Crypto values, so the table shows how samples of the same size can still differ.
Export what you generated
CSV export is created on your device and includes the trial number, outcome, cumulative Heads total and proportion, plus the method identifier and version. It contains no account or label data.
A simple classroom workflow
- Predict how variable a ten-flip run might be, then run it several times.
- Repeat with 100 and 1,000 flips and compare absolute percentage-point deviation.
- Inspect the cumulative graph rather than looking only at the final total.
- Download one run and calculate another statistic from the local CSV.
The experiment uses the same shared module as every decision tool. See the full method, source links, test fixtures, and limitations.