Pile ou face

How a flip is decided

Each flip generates one bit of randomness via crypto.getRandomValues() and maps it to heads or tails with rejection sampling applied so neither side is favored by how the raw random byte happens to be structured. Ask for multiple flips and the tool repeats this process independently for each one, then tallies the running total of heads versus tails so you can see the split at a glance instead of counting results yourself.

Single flips vs. batches and tallies

A single flip is useful for on-the-spot decisions — who goes first, which side of a rule applies. Running a batch of, say, 50 or 100 flips is more useful for demonstrations, classroom probability lessons, or settling a "best of N" contest in one action instead of flipping repeatedly. The tally counts heads and tails separately and shows the sequence in order, which matters if you care not just about the final count but about the pattern — for instance, checking whether a streak looks unusual.

Worked example: streaks are normal, not suspicious

Flip a fair coin 100 times and look for the longest run of identical results. Many people expect heads and tails to alternate fairly evenly, so a run of 5 heads in a row feels like it must mean something's off. It doesn't: with 100 independent flips, a run of 5 or more identical results is a common occurrence, not a rare one — the math of streak lengths in random sequences means you should expect at least one run of that length in a batch this size. If you flipped 100 times and got exactly 50 heads and 50 tails with no streak longer than 2, that would actually be closer to statistically unusual than a natural-looking batch with a streak of 5 or 6.

When to use this vs. other RandoKit tools

The coin flip is the right choice for binary decisions and best-of-N tie-breaks. If your decision has more than two options — say, choosing among three restaurants — the yes or no wheel only handles two outcomes as well, so for several named options you'd want the wheel of names instead. If you need a numeric outcome, like rolling for damage or picking a random seat number, use the random number generator or dice roller. For best-of-N tie-breaks specifically, run a batch flip of the needed size (best of 3, 5, 7) in one go so the whole series is decided transparently rather than one flip at a time.

Common mistakes

• The gambler's fallacy: believing that after several heads in a row, tails is "due." Each flip is independent — the coin has no memory, and the probability of heads on the next flip is still exactly 1/2 regardless of what came before.
• Re-flipping because a single result "doesn't feel random" — a truly fair process will sometimes produce streaks, lopsided short runs, or other patterns that look meaningful but aren't.
• Using a small number of flips (like 3 or 4) to judge fairness. Short sequences are far too noisy to reveal bias one way or the other; you'd need hundreds or thousands of flips to detect a subtle skew statistically.

Fairness note: digital vs. physical coins

A digital flip here gives heads and tails each exactly a 50% probability, drawn from a cryptographic random source rather than a simple pseudo-random generator. This is actually more evenly fair than a physical coin: research on real coin tosses has found that physical coins are very slightly biased toward landing on the same face they started on, due to precession during the flip, and a coin's mass distribution can introduce a small skew of its own. Those effects don't exist in a digital flip, where the outcome is a direct 50/50 random choice with no physical asymmetry to correct for. For more on how this randomness source works generally, see how randomness works.

Privacy

Flips happen entirely in your browser — no flip results, counts, or session data are sent to a server. Tallies persist in local storage on your device only, and you can export a flip sequence as .txt or .csv, or share it via an encoded URL, without uploading anything.