Are the results truly random?

For practical purposes, yes. JavaScript Math.random() uses a pseudorandom number generator (PRNG) that produces statistically indistinguishable results from true randomness for non-cryptographic uses. Each flip independent with 50/50 probability. For higher randomness rigor (cryptography, scientific research), use sources like Random.org (atmospheric noise) or quantum random number generators. For decision-making and probability demonstrations, modern PRNGs work fine.

Why don't I always get exactly 50/50?

Expected statistical variation. For 10 flips, exactly 5 heads happens only ~25% of the time. Most common results are 4-6 heads (combined probability ~66%). The expected average IS 50/50, but individual sample results vary. As sample size grows, ratios converge toward 50/50 (Law of Large Numbers), but absolute count variation grows with √N. For 100 flips: typical 45-55 of each. For 1000: typical 480-520. Pattern of 50/50 emerges over time, not within individual short sequences.

What is the "gambler's fallacy"?

The mistaken belief that past outcomes affect future probability of independent events. Example: "I've flipped 5 tails in a row, so heads must be due." WRONG. Each flip is independent — past 5 tails don't affect 6th flip's 50/50 odds. The misconception: thinking randomness "evens out" within short sequences. Reality: randomness produces clusters and streaks; long-term balance emerges over very large samples, not within recent observations. Gambler's fallacy contributes to lottery-buying, casino losses, and various poor decisions based on misunderstanding probability.

How likely is a streak of N same results?

Probability of exactly N consecutive same results = (1/2)^N. Examples: 2 in row 25%, 3 in row 12.5%, 4 in row 6.25%, 5 in row 3.13%, 6 in row 1.56%, 10 in row 0.098%, 20 in row 0.000095%. But in long sequences, streaks accumulate: in 100 flips, 5-streak appears ~80% of the time. In 1000 flips, multiple 7+ streaks expected. Streaks aren't bias — they're mathematically inevitable consequences of large random samples. Counter-intuitive: streaks look unusual to humans but are statistically expected.

Is a coin flip truly fair?

Theoretically yes (50/50). Real coins: very slight bias from wear, manufacturing, or flip technique. Persi Diaconis research showed flipped coins land same orientation as start ~51% of time. Spinning coins (not flipping) shows more bias. Digital simulators: use pseudorandom generators, mathematically indistinguishable from fair flip. For practical decisions: any modern coin flip is fair enough. For high-stakes scenarios (cryptography, sampling for research): use certified random sources. For most everyday uses: physical coin or digital simulator both work.

Coin Flip Simulator

Simulate flipping one or many coins at once. See heads/tails distribution, percentages, and streak information. Great for probability demonstrations and quick decisions.

Coin flipping is the simplest random binary decision tool — pure 50/50 chance between heads and tails (assuming a fair coin and unbiased flip). Used for centuries to settle disputes, make decisions, start sports games (the "coin toss" in football, cricket), and demonstrate probability concepts. While the math is simple (P(heads) = P(tails) = 0.5), real-world coin flipping has subtle complexities: a worn coin can have slight bias, the flipper's technique affects outcomes (some flippers consistently produce more of one result), and statistical sequences can show patterns that feel "non-random" but are mathematically expected.

The digital coin flipper here uses pseudorandom number generation, which behaves identically to a fair physical coin for practical purposes. With 10 flips, expect ~5 heads and ~5 tails — but actual results vary 3-7 each ~70% of the time, even with perfect randomness. Streaks of 4-5 consecutive heads or tails happen regularly in larger samples. These "patterns" are normal random behavior, not bugs in the system. The "gambler's fallacy" — believing past flips affect future flips — is psychological, not mathematical. Each flip is independent; 10 heads in a row doesn't change the next flip's 50/50 odds.

This calculator simulates 1-1000 coin flips and reports distribution statistics. Use it for: quick binary decisions (yes/no, A/B), probability demonstrations and teaching, statistical learning, settling friendly disputes, or just exploring randomness. Important context: results from any specific run won't be exactly 50/50. For 10 flips, normal results range 3-7 of each. For 100 flips, normal results range 40-60 of each. The "law of large numbers" guarantees long-term convergence to 50/50, but short-term variation is expected and normal. For sample size large enough to show near-50/50: flip 1000 times, expect ~480-520 of each (within 2% of theoretical 50/50). This is what makes coin flipping fair — predictable in aggregate while unpredictable in individual outcomes.

Inputs

Number of Flips

Reflip (change to reflip)

Change this value to flip again

Results

Heads

3 (30.0%)

Tails

7 (70.0%)

Longest Heads Streak

Longest Tails Streak

Heads vs Tails

Last updated: May 27, 2026

Formula

Coin flip probability: P(Heads) = 0.5 (50%) P(Tails) = 0.5 (50%) Each flip is independent. Past flips don't affect future flips. For N independent flips: Expected heads = N × 0.5 = N/2 Expected tails = N × 0.5 = N/2 Standard deviation (binomial distribution): σ = √(N × 0.5 × 0.5) = √(N/4) = 0.5√N For 10 flips: σ = 0.5 × √10 ≈ 1.58 Expected: 5 ± 1.58 ~68% of results: 3.4 to 6.6 heads ~95% of results: 1.8 to 8.2 heads For 100 flips: σ = 0.5 × √100 = 5 Expected: 50 ± 5 ~68%: 45-55 heads ~95%: 40-60 heads For 1000 flips: σ = 0.5 × √1000 ≈ 15.8 Expected: 500 ± 15.8 ~68%: 484-516 ~95%: 469-531 As N grows, ratio approaches 50/50 (Law of Large Numbers), but absolute count variation grows with √N. Streak probability: Probability of N consecutive same results: (1/2)^N 2 in a row: 0.25 (25%) 3 in a row: 0.125 (12.5%) 4 in a row: 0.0625 (6.25%) 5 in a row: 0.03125 (3.125%) 6 in a row: 0.0156 (1.56%) 10 in a row: 0.00098 (0.098%) 20 in a row: 0.00000095 (one in a million) But in long sequences, even rare streaks become common. For 100 flips: expected to see at least one 5-streak ~80% of the time. Probability of NO 5-streak in 100 flips: only ~20%. For 1000 flips: expect to see multiple 7-streaks; almost certainly at least one 10-streak. These streaks aren't bias — they're mathematical expectation. Probability of specific outcomes (binomial distribution): For N flips, probability of exactly k heads: P(k heads) = C(N,k) × 0.5^N Where C(N,k) = N! / (k! × (N-k)!) — combinations Example: 10 flips, exactly 5 heads: P = C(10,5) × 0.5^10 = 252 × 0.000977 = 0.246 (24.6%) Only ~25% chance of EXACTLY 5 heads in 10 flips. More common: 4-6 heads ranges. P(4 heads) = 0.205 P(5 heads) = 0.246 P(6 heads) = 0.205 P(at least 4 and at most 6) = 0.656 (65.6%) 35% chance of getting outside 4-6 heads range (very uneven results). Real coin flips vs. theoretical: Real coins: Slight bias possible from wear, manufacturing Persi Diaconis (Stanford mathematician) showed initial coin orientation slightly biases result (~51/49 toward same orientation as start) Spinning a coin (not flipping) shows more bias due to coin balance Digital simulators: Use pseudorandom number generators Mathematically truly 50/50 Sequences indistinguishable from physical fair coin For practical decisions, any method works. For scientific purposes, certified random sources (atmospheric noise, quantum decay) are most rigorous. Common psychology issues: Gambler's fallacy: believing past outcomes affect future probabilities. WRONG. Each flip independent. "After 5 tails, heads is due" — false. Still 50/50. Hot hand fallacy: believing streaks indicate biased system. Random sequences contain expected streaks; not evidence of bias. Confirmation bias: noticing/remembering streaks while ignoring expected mixed results. Memory selectively recalls "lucky runs." Apophenia: seeing patterns in random data. Brain naturally finds patterns where none exist. Decision-making applications: Binary choices: coin flip = perfectly fair decision when both options seem equal Reducing decision fatigue: for low-stakes choices, flip a coin to bypass extended deliberation Conflict resolution: agreed-upon coin flip prevents ongoing dispute Game starting: cricket toss for batting first, football toss for kickoff/receive When NOT to use coin flip: - Decisions with non-equal expected outcomes (one option better — use it) - Decisions where wrong choice has serious consequences (use deliberation) - Important choices where the "right" answer matters more than fairness

How to use this calculator

Enter number of flips (1-1000).
Change the "reflip" value if you want to redo the simulation.
Review the heads/tails distribution.
For quick decisions: use a single flip.
For probability demonstrations: try 10, 100, and 1000 flip simulations to see how distributions converge toward 50/50.
For statistics learning: observe how variability decreases (proportionally) as sample size increases.
For teaching: explain the gambler's fallacy — past flips don't affect future flips.
For settling disputes: pre-agree on which side represents which outcome before flipping.
For honest randomness: any modern digital flip is mathematically equivalent to a fair physical flip.
For comparison: try 10 flips of "heads" only by clicking 10 times — note that consecutive same results happen even with truly random flips (especially in short sequences).
For probability calculation: P(N same results in row) = (1/2)^N. 5 in a row: ~3% chance per attempt.

Worked examples

Single flip decision

Heads I go to the gym, tails I take a rest day. Flip result: 7 (or any random value) — heads Decision: gym Coin flipping for decisions reduces decision fatigue. Works best for low-stakes choices between equally appealing options. Forces a decision when paralysis would otherwise prevail. Important: respect the result. Pre-commit to following coin flip outcome. Flipping repeatedly until desired result obviates the value. Useful for: meal choice when undecided, which task to start, where to eat, simple yes/no questions, breaking ties between equivalent options.

100 flip distribution

Simulate 100 coin flips: Result example: 53 heads, 47 tails Expected: 50 each. Actual variation: 3 from expected. Standard deviation for 100 flips: 5. So 53-47 split is well within normal variation (1 standard deviation from expected). Repeat 10 times, typical results: 47-53 48-52 51-49 46-54 53-47 50-50 52-48 48-52 54-46 49-51 Note: 50-50 exact split is just one possible outcome among many. Distributions cluster around but rarely exactly hit 50-50. For 100 flips: 95% of all sample runs produce 40-60 of each. Outside this range possible but rare.

Streak demonstration

Simulate 50 flips. Count longest streak. Typical results: longest streak 4-7 consecutive same results. In 50 flips, expect: Multiple 3-streaks (common) Several 4-streaks (regular) 1-2 5-streaks (typical) Possible 6-streak (~50% chance) Possible 7-streak (~25% chance) These streaks look "non-random" but are mathematically expected. Don't treat them as evidence of bias. Practical insight for sports/business: "hot streaks" are usually random clustering, not predictive of continued success. Don't overweight recent positive (or negative) outcomes when their durability is uncertain. Reverse application: if you observe a 20-flip streak of the same result, THAT would indicate something's biased. 7-streaks: normal. 20-streaks: pause and verify.

When to use this calculator

Use this calculator for binary decision-making, probability demonstrations, statistical learning, settling friendly disputes, demonstrating random behavior, or exploring expected vs. actual outcomes.

Pair with dice-roller (multi-outcome random) and random-number (general randomness).

Important coin flip considerations:

1. **Each flip is independent.** Past results don't affect future probability. Gambler's fallacy is psychologically tempting but mathematically wrong.

2. **Streaks are expected.** Even truly random sequences contain runs of 5+ same results. Don't treat streaks as evidence of bias in short sequences.

3. **Convergence to 50/50 takes large samples.** 10 flips often shows uneven results. 1000 flips closer to 50/50. Law of Large Numbers.

4. **Small variations are normal.** 100 flips with 45-55 of each is typical. 100 flips with 30-70 split is unusual but happens occasionally.

5. **Digital flips are as fair as physical.** Modern pseudorandom generators indistinguishable from physical fair coins for practical purposes.

6. **Real coin slight bias.** Persi Diaconis research showed coins land same orientation as start ~51% of time. Spin (not flip) coins for even slighter bias.

7. **Decision-making is psychological as much as mathematical.** Coin flip often reveals your preference (you hope for one side). Useful for clarifying preferences, even when not following result.

8. **Pre-commit to outcomes.** Reflipping until desired result obviates the value. Either follow the coin or don't bother flipping.

9. **Variance decreases proportionally with N.** Standard deviation grows as √N, but proportion (sigma/N) decreases. 100 flips: variation ±5/50 = 10%. 10,000 flips: ±50/5000 = 1%.

10. **Patterns are illusory in random data.** Brain finds patterns where none exist. Apophenia is real cognitive bias.

11. **Useful for ethical randomization.** When fairness matters more than optimal choice (room assignments, task distribution, random sampling).

12. **Don't use for critical decisions.** Important choices deserve deliberation, not random selection.

Common mistakes to avoid

Believing in "due" outcomes after streak. Gambler's fallacy — each flip independent.
Treating short-sequence variation as bias. 8 heads in 10 flips is unusual but expected variance.
Reflipping until desired result. Defeats the purpose of randomization.
Using coin flip for important decisions. Major choices deserve deliberation.
Seeing patterns in truly random sequences. Brain pattern-matches even when no pattern exists.
Forgetting that "almost always" can fail. 99% probability still fails 1% of time.

Frequently Asked Questions

Sources & further reading

Probability and Statistics Resources — U.S. Census Bureau
Random Number Generation Standards — U.S. National Institute of Standards and Technology
Mathematics Education Resources — National Council of Teachers of Mathematics

SponsoredShop Top Deals on AmazonSupport CalcMountain — browse top-rated products at no extra cost to you.