1The fairest decision-maker ever invented

Two friends, one slice of pizza left, zero willingness to share. The solution is older than writing: flip a coin. Heads, you get it; tails, I do. No arguments, no favoritism, no lengthy debate — just physics and chance settling it in two seconds.

A coin flip is the purest example of a 50/50 random event. Two outcomes, each exactly as likely as the other, with no memory of what happened before. That's what makes it the universal tiebreaker: it's fair in a way humans struggle to be. A judge can't be bribed by a coin. A coin doesn't play favorites. It doesn't remember that you won last time.

Our free coin flip tool recreates that fairness digitally. Each flip draws from your browser's cryptographic random number generator — the same source of randomness your device uses for security keys — so heads and tails are each exactly 50% on every flip, and no flip remembers or influences the next. Flip once for a quick decision, or hit the Flip 100× button to watch probability play out at scale.

This guide covers everything around that simple act: why coins are fair, the famous mental trap they set, the math of streaks, how to run proper best-of tiebreakers, and the 2,000-year history of humans letting metal discs make their decisions.

2Why a coin flip is fair (and what "fair" means)

"Fair" has a precise meaning here: each outcome has exactly a 50% probability, and every flip is independent — the coin has no memory. Flip heads five times in a row, and the sixth flip is still 50/50. The coin doesn't "owe" you tails. It doesn't know the past happened.

Physical coins are fair because the physics is chaotic: the spin speed, the bounce, the air currents — tiny variations in the toss produce effectively unpredictable results. (Researchers have found real coins land very slightly off 50/50 due to manufacturing, and the side facing up at the toss has a tiny edge — but for any human purpose, it's 50/50.)

Our digital coin is fair for a different reason: it doesn't simulate physics at all. It takes a single unbiased bit from crypto.getRandomValues() — your browser's cryptographic randomness source — and maps it to heads or tails. No physics to be slightly off, no wear patterns, no "lucky" flipping technique. Just a clean 50/50 from the same well of randomness that protects your passwords. If that API is ever unavailable, it falls back to Math.random(), which is plenty fair for casual decisions.

Three flips, eight futures — each path is 1/8 go H T H T T H HHH HHT HTH HTT THH THT + TTH, TTT HHH is exactly as likely as HHT, HTT, or any other sequence: (1/2)³ = 1/8.
Every path through the tree is equally likely — HHH surprises us, but the math doesn't care.

Study that tree for a moment, because it kills the most common coin misconception: HHH feels special, but it's exactly as likely as HHT or HTT — each specific sequence of three flips has probability (1/2)³ = 1/8. Our brains flag the "patterned" outcome as remarkable and the "messy" ones as normal, but the coin makes no such distinction. Every path is 1/8.

3The gambler's fallacy: the trap every brain falls into

You flip heads five times in a row. Tails is "due," right? The universe owes you a tails. Wrong — and this mistake has a name: the gambler's fallacy. It's the belief that past random results change future odds. After five heads, the next flip is still exactly 50/50. The coin has no memory, no sense of fairness, no ledger to balance.

The fallacy gets its name from a famous night in 1913 at the Monte Carlo Casino, when the roulette ball landed on black 26 times in a row. Gamblers lost millions betting on red, convinced it was "due" — each time more certain as the streak grew. But the wheel, like the coin, remembered nothing. Every spin was independent.

Why do smart people keep falling for it? Because humans are pattern detectors in a world where patterns usually mean something. In nature, five heads in a row would be suspicious — it'd suggest a trick coin. Our instincts correctly flag streaks as noteworthy; they just misfire on truly random processes, where streaks are not only possible but guaranteed given enough flips. (Flip 100 times and you'll see several 5-streaks — randomness demands it.)

The one-sentence antidote: the coin doesn't know what happened before. Each flip starts the universe fresh at 50/50. If you feel tails is "due," that's your pattern-detector misfiring — smile at it and flip anyway.

4Streak math: how unlikely is "unlikely"?

Let's put numbers on streaks, because the results surprise almost everyone. Each additional flip in a streak halves the probability:

StreakProbabilityAs oddsRoughly…
2 heads in a row(1/2)² = 1/425%Every 4th pair of flips
3 heads in a row(1/2)³ = 1/812.5%Common — happens all the time
5 heads in a row(1/2)⁵ = 1/323.1%Once in ~32 tries
7 heads in a row(1/2)⁷ = 1/1280.78%Once in ~128 tries
10 heads in a row(1/2)¹⁰ = 1/10240.098%About 0.1% — rare but real
20 heads in a row(1/2)²⁰ ≈ 1/1,048,5760.000095%Lottery territory

Two things to notice. First, 5 in a row is only 1 in 32 — flip a few dozen times and you'll probably see one. It feels miraculous; it's routine. Second, the probabilities collapse fast: each extra flip halves the chance, so 10 in a row (1/1024) is genuinely rare in casual flipping but will happen to someone, somewhere, every day.

Here's the subtle part most people miss: those probabilities are for a streak starting now. After you've already flipped 4 heads, the chance the next flip makes it 5 is just 1/2 — the earlier flips already happened and don't count anymore. "1 in 32 for five in a row" and "50/50 for the next flip" are both true; they answer different questions.

Each extra flip halves the streak odds 2 in a row25% 3 in a row12.5% 5 in a row3.1% 7 in a row0.78% 10 in a row0.1% Halving, halving, halving — exponential decay in one picture.
Streak odds collapse exponentially — but short streaks are far more common than they feel.

5The 100-flip experiment (try it on the tool)

Here's where intuition really breaks. Hit the Flip 100× button on our tool and look at the result. You'd expect 50 heads — and you'd be wrong, in an interesting way.

The expected number of heads is 50, yes. But the standard deviation is 5, meaning results typically land between 45 and 55 — and the chance of hitting exactly 50 is only about 8%. "Exactly half" is one specific outcome out of many near-misses. You'll far more often see 47, 52, or 49 than a clean 50.

Now look at the sequence itself. You'll see several streaks of 5 or more — and that's proof of fairness, not evidence against it. A sequence of 100 flips with no 5-streak would actually be suspicious: humans faking "random" data avoid long streaks (it feels wrong), which is how statisticians catch fabricated data. Real randomness clusters; fake randomness spreads out too evenly.

Try this: flip 100× three times and write down the heads counts — say 52, 47, 55. Average them: (52+47+55) ÷ 3 = 51.3. Already drifting toward 50. This is the law of large numbers in action: more flips → closer to the true 50%. Ten flips can land 8-2 by luck; ten thousand flips basically can't.

6Best-of-three and other fair tiebreakers

One flip settles most debates, but sometimes you want more ceremony — or the stakes feel bigger than a single toss. Enter best-of-N: first to win the majority takes it.

MethodHow it worksEach side's chanceBest for
Single flipOne toss decides50/50Quick everyday calls
Best of 3First to 2 toss-wins50/50Feels more decisive
Best of 5First to 3 toss-wins50/50Ceremonial showdowns
Two flips, 4 peopleHH / HT / TH / TT assigns each person25% eachSmall groups, pick-one-of-four
Random number 1–NGenerator picks the winner1/N eachBigger groups, raffles

A fair warning about "fairness theater": best-of-3 doesn't make the outcome more fair than a single flip — 50/50 is 50/50. What it does is make the process feel more legitimate, which matters when humans need to accept the result. Sports use best-of-7 series for the same psychological reason (plus revenue).

For group decisions — picking who goes first among four people — flip twice per person (HH, HT, TH, TT gives four outcomes) or just use our random number generator to pick 1–4 directly. Same fairness, less coin juggling.

7A 2,000-year history of letting coins decide

The coin flip is ancient. The Romans called it "navia aut caput" — "ship or head" — because their coins showed a ship on one side and the emperor's head on the other. Same game, different pictures: ship or head, heads or tails.

Since then, coins have settled some gloriously consequential things. The 1968 NFL used a coin toss to decide draft picks. Wright brothers lore says a coin toss decided who flew first at Kitty Hawk in 1903 (Wilbur won the toss, Orville won history — his flight was the successful one). Cricket, soccer and American football still open every match with a coin toss, broadcast to millions. Portland, Oregon got its name from an 1845 coin flip — "Portland" beat "Boston" two out of three.

The through-line across two millennia: when humans can't agree and won't fight, chance is the neutral arbiter everyone accepts. The coin doesn't care who you are. That's not a bug in decision-making — it's the feature that kept the peace.

Two thousand years of "ship or head" Rome"navia aut caput" 1845Portland named 1903Wright brothers TodayCrypto-grade flips The technology changed from bronze to browsers. The fairness didn't.
From Roman bronze to browser cryptography — the coin flip keeps its job.

8What NOT to use a coin flip for

Let's be straight about the limits. Our coin flip — and this guide — is for fun and casual decision-making: settling debates, picking teams, choosing who goes first, breaking lunch deadlocks.

The coin's job description: break ties between options you've already decided are both fine. If the flip's result makes you feel disappointed, pay attention — that feeling just told you which option you actually wanted. (Psychologists call this the "flip test," and it's the coin's secret second use.)

Flip a crypto-fair coin — free

Heads or tails from your browser's cryptographic randomness. Flip once, run best-of-three, or fire the 100× experiment.

🪙 Flip the Coin

100% free · No account · Private by design

9Frequently asked questions

Is this online coin flip fair?

Yes. This coin flip draws each result from cryptographically secure randomness (crypto.getRandomValues) with an unbiased bit, so heads and tails are each exactly 50% on every flip — and no flip remembers or influences the next.

What is the gambler's fallacy?

The gambler's fallacy is the mistaken belief that past results change future odds — for example, thinking tails is "due" after five heads. A fair coin has no memory: after five heads in a row, the next flip is still exactly 50/50.

Can I use this coin flip for gambling or betting?

No. This tool is for fun and casual decision-making only — settling debates, picking teams, or choosing who goes first. Never use it for gambling, money decisions, or anything with real stakes.

What are the odds of flipping 5 heads in a row?

Each flip halves the chance, so 5 heads in a row has probability (1/2)^5 = 1/32, about 3.1%. For 10 in a row it's 1/1024, roughly 0.1%. Short streaks are far more common than they feel.

If I flip 100 times, will I get exactly 50 heads?

Probably not exactly 50 — the expected value is 50 heads with a standard deviation of 5, and the chance of landing on exactly 50 is only about 8%. Try the Flip 100× button and watch the result land near, but rarely on, 50.

How does the randomness work?

Each flip takes a single unbiased bit from your browser's cryptographic random number generator (crypto.getRandomValues), the same source used for secure keys — far more unpredictable than basic pseudo-random functions. If that API is unavailable, it falls back to Math.random.

Keep exploring

Chance is a big family. These tools cover its closest relatives: