ABC Tools

Streak Probability, Explained

After reading this you can compute the exact chance of hitting a winning or losing streak of a given length in a session of n bets, and judge how long a streak you should actually expect.

Flip a fair coin 200 times. What is the chance you get 8 tails in a row somewhere along the way? Most people guess "unlikely", maybe one in twenty. The real answer is about 54%. More likely than not. That gap between intuition and arithmetic is where a lot of gambling misjudgment lives.

This calculator gives you that number exactly, for any number of bets, any per-bet win probability, and any streak length. It also tells you the longest streak you should expect to see. This article explains where those numbers come from and how to read them without fooling yourself.

What a streak probability actually measures

The quantity here is the probability that at least one run of length k or longer appears somewhere in n trials. It is not the chance of a specific pattern at a specific position.

The distinction matters. The chance of losing your next 8 bets in a row, starting right now, is just 0.5^8 = 0.00391, about 1 in 256. That single position looks rare. But a 200-bet session offers 193 overlapping starting positions for an 8-long window, and those chances accumulate. The session-level probability climbs to roughly 54%.

A "streak of k or longer" is the standard reading. If you see 9 losses in a row, that session counts as containing a streak of length 8. The calculator reports the chance of reaching at least k.

When to use it, and when not to

Use this when you want to set expectations for a session of independent, identical bets: roulette on a single color, coin-flip style prop bets, or a slot session modeled as independent spins. It answers questions like "how bad can a normal night look?" and "is this losing run evidence of anything unusual?"

Do not use it when outcomes are not independent, or when the per-bet probability changes. Blackjack with counting, poker, and sports betting all violate independence or fixed probability. A streak model there will mislead you.

Also treat "win probability" as the probability of the event you are streaking on, not your net profit. A bet at even money on red in European roulette wins with probability 18/37 = 0.4865, not 0.5. Use the true figure.

The formula and the intuition behind it

The clean way to compute the chance of no run of length k in n trials uses a recurrence. Let p be the probability of the outcome you are streaking on (so for losing streaks, p is the loss probability). Let A_n be the probability that n trials contain no run of k or more. Then the streak probability you want is 1 - A_n.

A_n = \sum_{j=0}^{k-1} p^{j}(1-p)\, A_{n-1-j}

Read the sum this way. Condition on how long the current run lasts before it breaks. The run of the target outcome has length j (for j from 0 to k-1), which happens with probability p^{j}(1-p), and then the process restarts with n-1-j trials left, still needing no long run. The base cases are A_m = 1 for every m \lt k, because you cannot fit a run of length k in fewer than k trials.

A useful back-of-envelope approximation treats each of the n-k+1 starting windows as a near-independent chance to begin a fresh run:

P(\text{streak} \ge k) \approx 1 - \left(1 - (1-p)\,p^{\,k}\right)^{\,n-k+1}

Here (1-p)\,p^{k} is the chance a run of exactly k starts at a given position (preceded by a non-target outcome or the start of the sequence). This approximation is close for small p^{k} but the recurrence is exact, so the calculator uses the recurrence.

Worked example: 8 losses in 200 even-money bets

Reproducing the demo

Load the demo values: n = 200, prob = 50, kind = lose, k = 8. Because you are streaking on losses and the win probability is 50%, the loss probability is p = 0.5.

  1. Single-window chance of a fresh 8-run: (1-p)\,p^{8} = 0.5 \times 0.5^{8} = 0.001953.
  2. Number of starting windows: n - k + 1 = 200 - 8 + 1 = 193.
  3. Approximation: 1 - (1 - 0.001953)^{193} = 1 - 0.9980^{193} \approx 0.3130.
  4. The approximation undercounts here because it ignores runs longer than 8 and overlap effects. The exact recurrence gives 1 - A_{200} \approx 0.5453.

So the honest answer is about 54.5%. Note how far the crude window estimate (31%) sits from the exact value. Approximations are for feel, not for decisions.

Do not use the single-position number (1 in 256 for 8 in a row) as the session risk. It understates the real chance by more than a factor of a hundred here. The session is long, and long sessions manufacture streaks.

How the numbers move with session length

Hold k = 8 and p = 0.5 and vary n. The chance of an 8-long losing streak grows steadily as the session lengthens. At 50 bets it is about 15%. By 200 it is 54%. By 1000 it is 98%.

The probability rises fast and then flattens as it approaches 1. The vertical marker sits at the demo's 200 bets, near 54%.

The expected longest streak grows roughly like \log_{1/p}(n). For a fair coin over 200 flips, the expected longest run of either outcome is about \log_2(200) \approx 7.6. That is why 8 in a row feels shocking but happens routinely: your typical longest run is already close to 8.

Explore the tradeoff

With p = 0.5 and n = 200, the chance of a losing streak of length k is: k=5 gives 96%, k=6 gives 84%, k=7 gives 71%, k=8 gives 55%, k=9 gives 39%, k=10 gives 26%, k=11 gives 16%, k=12 gives 9.7%. Each extra required loss roughly halves the probability, because the per-window chance carries an extra factor of p.

Reading and interpreting the result

Treat the output as a description of a normal session, not a warning. If the calculator says a losing streak of 8 has a 54% chance in 200 bets, then seeing such a streak tells you nothing about the fairness of the game. It is the expected texture of random data.

This is the engine behind the gambler's fallacy. After 8 losses, the next bet still loses with probability p. The coin has no memory. The streak was likely to occur; its occurrence changes nothing about the next flip.

For risk decisions, pair the streak length with the bankroll it would drain. A streak of 8 flat-stake losses costs 8 units. A streak of 8 losses under a doubling system costs 2^{8} - 1 = 255 units. That is why streaks are ruinous for the Martingale system and gentle for flat staking.

Common mistakes

Using net-profit probability instead of event probability
Streak math needs the probability of the specific outcome you are counting, not your edge. For red in European roulette use 0.4865, not 0.5.
Confusing "next k" with "somewhere in n"
The next-k chance is p^{k}. The somewhere-in-n chance is far larger. Mixing them up understates risk by orders of magnitude.
Assuming a streak means the game is rigged
An 8-loss run in 200 fair bets is more likely than not. It is not evidence of tampering.
Ignoring dependence
If bets are correlated, this model does not apply. Streaks in correlated data cluster differently.

Related tools

Once you know your streak risk, size your response. The Risk of Ruin calculator turns per-bet variance and bankroll into a probability of going broke, which is the natural next question after "how long can the losing run be?" If you have a genuine edge, the Kelly Criterion calculator gives a stake that survives ordinary streaks. And if you are weighing a casino promotion, the Bonus Wagering EV calculator shows what the offer is worth once the rollover requirement is met.

Frequently asked questions

Why is a long losing streak so likely in a fair game?

Because a session offers many overlapping chances to start a run. In 200 fair bets there are 193 windows where an 8-loss run could begin. The individual chance is small (about 1 in 512 per window), but across 193 windows the total reaches roughly 54%.

Does a streak change the odds of the next bet?

No. In independent bets the next outcome has the same probability regardless of history. After 8 losses at even money, the next bet still loses with probability 0.5. Expecting a "correction" is the gambler's fallacy.

How do winning and losing streaks differ?

They use the same formula with a different p. For a losing streak, p is the loss probability; for a winning streak it is the win probability. In a game with negative edge, losing streaks are longer and more common than winning ones because the loss probability is higher.

What streak length should I expect over n bets?

Roughly \log_{1/p}(n). For 200 fair bets that is about 7.6, so a run near 8 is typical. Doubling the session adds only about one to the expected longest run, since the growth is logarithmic.

Is the calculator exact or an approximation?

Exact. It uses the recurrence for the probability of no run of length k, then reports one minus that. The multiplication approximation shown in this article is only for building intuition.

Try it yourself in the tool

Related tools

Kelly Criterion Calculator

Optimal stake size for a bet with an edge.

Martingale Simulator

Bust probability and true EV of doubling after losses.

Risk of Ruin Calculator

Chance of losing the bankroll before hitting a target.

Bonus Wagering EV

What a casino bonus is worth after the rollover.

Blackjack Strategy & EV Explorer

Exact basic strategy charts and EVs for your exact rules.

Card Counting Trainer

Learn Hi-Lo card counting: speed drills, true-count quizzes and a bankroll simulator with your bet spread.