Coin Toss Streak Calculator: Probability of Consecutive Heads or Tails
Enter your streak length and total tosses to find the probability of seeing at least one run of consecutive identical outcomes. You also get the expected number of tosses until the first streak, the naive single-run probability, and a probability-vs-tosses chart that shows how the odds climb as you flip more coins. The calculator handles fair coins and biased ones, and lets you target heads, tails, or either symbol.
Formula
Worked example
Streak of 5 heads, 100 fair-coin tosses: single-block probability = (0.5)^5 = 1/32 = 3.125%. Expected wait = 2^6 - 2 = 62 flips. Using the DP recurrence for n=100, k=5 gives Q(100,5) approx 0.029, so P(at least one streak) = 1 - 0.029 = 97.1%.
What is a coin toss streak?
A streak (or run) is a sequence of consecutive identical outcomes in a series of coin flips. If you flip a coin 10 times and see H H H H H, that is a heads streak of length 5. Streaks are central to probability theory because they reveal the tension between two true but seemingly contradictory facts: any single long streak is extremely unlikely, yet long streaks are practically guaranteed to appear if you flip enough times. This calculator quantifies exactly that tension: given a streak length k and a total number of tosses n, it tells you the probability that at least one such streak will occur.
How the probability is calculated
The calculator uses a dynamic-programming recurrence to find Q(n, k), the probability of getting through n tosses with NO streak of k identical outcomes. The key recurrence is: Q(n, k) = sum over j from 0 to k-1 of: Q(n-j-1, k) * p^j * (1-p) where p is the per-toss probability of the target outcome and (1-p) is the "reset" probability. The sum captures every way the sequence can avoid a streak of k: the last run of matching tosses was of length 0, 1, 2, ... or k-1 before a different outcome broke it. The answer you see is P = 1 - Q(n, k). For the "either heads or tails" mode, a joint state-machine DP tracks the current same-symbol run length and accounts for either direction reaching k. The expected wait time for a fair coin uses the classical result: E[T] = 2 + 4 + 8 + ... + 2^k = 2^(k+1) - 2 for a target outcome (heads or tails). For "either", E[T] = 2^k - 1 because every toss contributes to some run.
The gambler's fallacy and why streaks feel surprising
People systematically underestimate how often streaks occur in independent random sequences. This is sometimes called the "clustering illusion": random data contains more runs than intuition predicts because the human brain pattern-matches aggressively. After three heads in a row, many people believe tails is "due", but each flip remains independent and the coin has no memory. Conversely, casino floor managers and sports commentators regularly attribute streaks to skill or momentum when the data is consistent with pure chance. The birthday paradox gives an analogy: the probability that two people in a room of 23 share a birthday is about 50%, far higher than most people guess, because there are many pairs to check. Streak probability works similarly: a streak of 5 has only a 3.1% single-run chance, but across 100 tosses there are roughly 96 overlapping windows of 5 consecutive flips, so the cumulative probability of at least one streak climbs to about 97%.
Biased coins and real-world applications
When the coin is not fair, the symmetry breaks. A coin with P(heads) = 0.7 will produce runs of heads far more often than a fair coin, and runs of tails less often. The calculator lets you enter any per-toss probability between 0.01 and 0.99 to model weighted coins, unfair dice converted to two-outcome experiments, or any binary process with a known bias. Real-world uses include:
- Sports analytics: testing whether a player's "hot streak" is statistically significant or explainable by chance alone.
- Quality control: detecting when a production line is producing too many consecutive defects, which may signal a systematic fault.
- Finance: testing whether a trader's consecutive winning days exceed what a random walk would predict.
- Games and gambling: understanding house edge, run probabilities, and setting bet-size rules for streak-based strategies.
Expected tosses to first streak by length (fair coin)
| Streak length (k) | Single-run probability | Expected tosses (heads only) | Expected tosses (either) |
|---|---|---|---|
| 2 | 1 in 4 (25%) | 6 | 3 |
| 3 | 1 in 8 (12.5%) | 14 | 7 |
| 4 | 1 in 16 (6.25%) | 30 | 15 |
| 5 | 1 in 32 (3.125%) | 62 | 31 |
| 6 | 1 in 64 (1.563%) | 126 | 63 |
| 7 | 1 in 128 (0.781%) | 254 | 127 |
| 8 | 1 in 256 (0.391%) | 510 | 255 |
| 10 | 1 in 1,024 (0.098%) | 2,046 | 1,023 |
How many coin flips you need on average before a streak of k consecutive same-outcome tosses appears for the first time.
Frequently asked questions
What is the probability of flipping 5 heads in a row?
The probability of any specific sequence of 5 heads is (0.5)^5 = 1/32, or about 3.13%. That is the chance that a particular block of 5 tosses is all heads. If you flip a coin 100 times, the chance of at least one run of 5 heads somewhere in the sequence rises to roughly 97%, because there are many overlapping windows across all 100 flips.
How many coin flips does it take to get 10 heads in a row?
On average you need 2^11 - 2 = 2,046 flips before a streak of 10 consecutive heads appears for the first time on a fair coin. The 63% probability point (the median wait time for a geometric-like process) is around 1,024 tosses. If you are looking for 10 of the same outcome in either direction, the expected wait drops to 2^10 - 1 = 1,023 tosses.
Does the coin "know" it is on a streak and become more likely to continue?
No. Each coin flip is an independent event. The probability of the next flip being heads is always p (0.5 for a fair coin) regardless of what came before. The gambler's fallacy is believing the coin is "due" for tails after a long heads streak. The coin has no memory. Long streaks feel surprising because we encounter them in hindsight, after many opportunities for one to occur.
What is the difference between single-run probability and cumulative probability?
Single-run probability is the chance that a specific block of k consecutive tosses all lands the same way: (0.5)^k for a fair coin. It applies to exactly one attempt. Cumulative probability is the chance that at least one such streak appears anywhere across n total tosses. Because there are approximately n - k + 1 overlapping windows to check, the cumulative probability grows much faster than the single-run value suggests, especially when n is much larger than k.
How does a biased coin change the streak probability?
A coin weighted toward heads (p > 0.5) makes heads streaks more likely and tails streaks less likely. For example, a coin with P(heads) = 0.7 has a single-5-heads-run probability of 0.7^5 = 16.8% instead of 3.1% for a fair coin. The expected wait time for a heads streak of 5 falls from 62 flips to about 18 flips. Enter your coin's bias in the "probability of heads" field to see the adjusted figures.
Why does the calculator cap calculations at 1,000 tosses?
The dynamic-programming algorithm runs an O(n * k) loop, which is fast for n up to 1,000. Beyond that, the probability of at least one streak is already extremely close to 100% for any streak length up to about 20, so the exact number matters very little. The expected-tosses output and the single-run probability are still exact regardless of the cap.
What streak length gives a 50-50 chance in 100 tosses?
For a fair coin with 100 tosses (heads-only), a streak of 7 gives about a 75% chance and a streak of 8 gives about 55%, so the crossover is near 8 or 9 consecutive heads. If you are looking for either heads or tails (same symbol), a streak of 8 gives roughly a 99% chance in 100 tosses, while a streak of 10 gives about 78%.