Coin Flip Simulator Calculator

A sequence of coin flips is the textbook example of independent Bernoulli trials, and the count of heads follows a binomial distribution with mean N times p and standard deviation the square root of N times p times one minus p. For 100 flips of a fair coin that gives an expectation of 50 heads and a standard deviation of 5, so anything from 40 to 60 heads sits inside two standard deviations and is entirely unremarkable. The z-score reported here is the observed head count minus the expectation divided by that standard deviation, and the chi-square statistic sums the squared deviation over the expectation across both outcomes. With one degree of freedom the conventional critical values are 3.84 at the 5 percent level and 6.63 at the 1 percent level. Streaks are where intuition fails most reliably. The longest run in N flips grows roughly as the base-2 logarithm of N, so about 6 to 7 in 100 flips and about 10 in 1,000, and sequences that look too clumpy to be random usually are not. Three caveats matter. The chi-square approximation degrades when an expected cell count falls below about 5, so it is unreliable for very small samples or extreme head probabilities. Repeatedly re-running until an unusual result appears is multiple testing and will manufacture apparent bias. And Math.random in a browser is a pseudo-random generator seeded per page, adequate for simulation but not for cryptographic or gaming use.

The mathematics behind this simulation predates any computer. Jacob Bernoulli proved the first law of large numbers for exactly this setting in Ars Conjectandi, published posthumously in 1713, establishing that the observed proportion of heads converges on the underlying probability as trials accumulate. Abraham de Moivre then derived the normal approximation to the binomial in The Doctrine of Chances in the 1730s, which is what makes the z-score interpretation possible. Karl Pearson introduced the chi-square goodness-of-fit test in 1900, giving a formal way to ask whether an observed split departs further from expectation than chance comfortably allows. Empirical verification has a long and slightly eccentric record. The Comte de Buffon is reported to have tossed a coin 4,040 times, and John Kerrich, a mathematician interned in Denmark during the Second World War, passed the time by flipping a coin 10,000 times and recording 5,067 heads, publishing the experiment in 1946. Persi Diaconis, Susan Holmes and Richard Montgomery analysed the physics of the toss in a 2007 paper and concluded that a flipped coin retains a slight bias, near 51 percent, toward the face it started on, a prediction later supported by a large collaborative study that recorded over 350,000 human flips. On the computational side, John von Neumann proposed the middle-square method in the late 1940s and Derrick Lehmer the linear congruential generator soon after; the Mersenne Twister of Matsumoto and Nishimura in 1997 and later xorshift-family generators are what browsers now use.