Share this link via
Or copy link
Bet The Flip (β)
Your current points are 10.
Current win streak: 2
Win streak bonus: x1.5
Which one will you bet on?
How much would you like to bet?
You're betting 3 points on heads.
If you're ready, please tap the button below, and flip your coin!
Result
You Win! You've earned points!
Current points: 10
Real-time data from every coin flip performed on Flip a Coin.com since January 26, 2024. Snapshot refreshed daily.
An honest random coin flip should converge to 50% / 50% over many trials. Here is the live result of every flip on this site.
Result is essentially 50 / 50 (within 0.01 percentage points).
This chart runs fresh in your browser each visit, using the same cryptographic RNG (crypto.getRandomValues) that powers single flips on this site. Watch 1,000 simulated fair flips: the proportion of heads (cyan line) bounces wildly at first, then settles toward 0.5 (gold dashed line). Reloading the page generates a new random sequence — that's the demonstration. The math says any fair generator will show this convergence; refresh and see for yourself.
Learn more about the Law of Large Numbers in the classroom simulator article →
Same fresh simulation as the chart above (same cryptographic RNG), plotted differently. While the proportion of heads converges to 0.5, the count difference (heads minus tails) does not return to zero. It performs a random walk, drifting away from zero like the steps of a person without a destination. The gray band shows the ±2σ envelope, which contains the trajectory with about 95% probability for a fair coin. Reload to see a new walk.
After 100 fair flips, the expected gap between heads and tails is not zero but rather about 8. After 10,000 flips, the expected gap grows to about 80. The proportion shrinks because the denominator grows faster than the gap — not because the gap itself shrinks.
A 50 / 50 split is the headline, but how do we prove the coin is fair? With 347,595,127 flips on record, we can apply two standard tests from inferential statistics: the Z-test for a proportion and the chi-squared goodness-of-fit test. Both ask the same question — “could random chance alone produce the bias we observe?” — and arrive at the same answer.
| Total flips (n) | 347,595,127 |
|---|---|
| Heads observed | 173,791,769 (p̂ = 0.499983) |
| Expected under fair coin (n · p) ? | 173,797,564 |
| Standard deviation (σ = √(np(1−p))) ? | 9,321.95 |
| Z-score ((X − np) / σ) ? | -0.622 |
| Chi-squared statistic (df = 1) ? | 0.386 |
| Two-tailed p-value ? | 0.534207 |
| 95% confidence interval for p ? | [0.499931, 0.500036] contains 0.5 ✓ |
| Verdict ? | Fair — The observed bias is consistent with a 50/50 coin. There is no statistical evidence of unfairness. |
The Z-test and chi-squared test are asymptotically equivalent for two categories; Z² = χ²(df = 1). The p-value is computed from a normal approximation, accurate to about 7 decimal places at this sample size. With 347,595,127 flips, the test is so sensitive that any deviation greater than about 18,271 flips (just 0.0053% of all flips) would register as statistically significant — yet such a tiny deviation is meaningless in practice. Statistical significance does not mean practical bias.
See the interactive simulators behind these tests in our classroom article →
Most popular flip counts chosen by users — number of times each count was selected.
Flips performed via the public JSON API at /api/ — primarily by AI assistants and integrated apps. Counted separately from on-site and widget flips.
| Coin | Heads | Tails | Flips |
|---|---|---|---|
| fcc-heads-or-tails | 511,449 | 510,867 | 1,022,317 |
| fcc-yes-or-no | 61 | 54 | 115 |
| usd-25cent | 7 | 8 | 15 |
| ccc-bitcoin | 1 | 2 | 3 |
| fcc-go-or-stay | 2 | 0 | 2 |
| ccc-cardano | 0 | 1 | 1 |
| usq-texas | 1 | 0 | 1 |
| fcc-hot-or-cold | 0 | 1 | 1 |
| fcc-outdoor-or-indoor | 0 | 1 | 1 |
| fcc-play-or-work | 1 | 0 | 1 |
Integrate the API in your app or AI assistant: see /api/ (developer docs) or /ai-integration/ (Claude Desktop / ChatGPT setup).
Flips performed via the embeddable widget on third-party sites. Counted separately from on-site flips.
Get your own embed at /widget/.
Snapshot updated: Jun 30, 2026 03:00 UTC. Refreshed daily at 03:00 UTC.