跳到主要内容
FunBox

聚会游戏

Probability of Consecutive Heads in Independent Coin Flips

Calculate the probability of consecutive heads for one to six specified flips. Follow a three-flip tree and distinguish a fixed block from a streak anywhere.

约 5 分钟阅读

本文正文为英文。

For a fair coin with independent flips, the probability of consecutive heads in a block of k flips chosen in advance is (1/2)^k. Three specified flips all landing heads has probability 1/8. Once those three heads have happened, the next flip still has a 1/2 chance of heads.

We checked OpenStax’s explanation of independent events and the current FunBox coin flip for this guide. The calculations below describe a fair, independent model. Choose a block before flipping. A separate four-flip example then shows how allowing more starting positions changes the answer.

Define the block of flips before calculating

Write down which flips you mean. You might choose the next three flips, or positions eight through ten in a planned series. For either specified block, success means every selected flip is heads. Results outside that block do not affect whether it succeeds.

That condition does not require the flips immediately before or after the block to be tails. HHHH still contains heads in its first three positions. Asking for a run of exactly three, bounded by tails, adds conditions and needs a different calculation.

Use H for heads and T for tails on your record sheet. With three specified flips, HHH is the only successful sequence. HTH contains two heads, but they do not fill the whole block. A question about the number of heads can therefore differ from a question about consecutive heads.

Multiply independent one half probabilities

Fairness assigns probability 1/2 to each face. Independence means that knowing earlier results leaves the next probability unchanged. Following the heads branch three times gives 1/2 × 1/2 × 1/2 = 1/8. This is the probability of three heads in a row in the specified three-flip block.

The tree below shows all eight sequences. Every branch has probability 1/2, so every complete three-flip path has probability 1/8. HHH is one path. The other seven fail the all-heads condition, even though several include heads.

Read the tree from Start to a leaf, keeping the flip order. The H branch followed by T and then H ends at HTH. Its probability is also 1/8. An alternating sequence has no special advantage over HHH when each complete sequence is specified beforehand.

Three specified flips with probability 1/2 on every branch
Start
├── H
│   ├── H
│   │   ├── H  HHH  1/8
│   │   └── T  HHT  1/8
│   └── T
│       ├── H  HTH  1/8
│       └── T  HTT  1/8
└── T
    ├── H
    │   ├── H  THH  1/8
    │   └── T  THT  1/8
    └── T
        ├── H  TTH  1/8
        └── T  TTT  1/8

Read a table for one to six specified flips

Each extra required head multiplies the probability by another 1/2. For k specified flips, there are 2^k equally likely complete sequences and only one all-heads sequence. The table gives exact fractions and exact percentages for blocks of one through six flips.

A 1/64 chance for six heads does not promise one success in every 64 blocks. It gives an expected count of one across 64 independent six-flip blocks, while the actual count can differ. Decide the block length before collecting results so that each trial tests the same event.

All heads in a block chosen before the flips
Specified flipsSuccessful sequenceExact probabilityExact percent
1H1/250%
2HH1/425%
3HHH1/812.5%
4HHHH1/166.25%
5HHHHH1/323.125%
6HHHHHH1/641.5625%

Separate a fixed block from a streak anywhere

Consider four planned flips. Heads on flips one through three occurs in HHHT and HHHH, giving 2/16 = 1/8. At least one HHH anywhere in those four flips occurs in HHHT, HHHH and THHH, giving 3/16. The second question allows a later starting position.

You cannot simply add 1/8 for each of the two possible starting positions. HHHH satisfies both windows, so that addition counts it twice. Removing the overlap gives 1/8 + 1/8 − 1/16 = 3/16. Longer sequences introduce more overlapping windows.

Is a streak anywhere in twenty flips the same event? No. It offers eighteen possible starting positions for HHH, and those windows overlap. The fixed-block value 1/8 cannot serve as the answer for the whole series. State the sequence length and the required run length before choosing a counting method.

Avoid assuming the next flip must be tails

After three heads, is tails more likely next? Under the same fair, independent model, heads and tails each remain at 1/2. Before four flips begin, HHHH has probability 1/16. After HHH is known, only the fourth result remains unknown. The first calculation includes four unknown results; the second starts with three results already known.

Independence also means the coin does not remove a face after using it. Our sampling guide explains why draws from a shrinking pool behave differently. The two dice probability guide uses independent outcomes to calculate sums, another distinct question.

Record fixed blocks without expecting alternation

To try the idea, mark eight separate three-flip blocks on paper before starting. Open Coin Flip, press its flip button and wait for each result. Write H or T in the next empty position. Complete all 24 flips, then count blocks containing HHH. This is a suggested experiment, not a reported test.

FunBox shows the latest result and running totals for heads, tails and all flips. Those totals do not preserve the order of the sequence, so keep your paper record. The current page resets its results and counts on refresh. A small sample cannot certify a coin or digital tool as fair.

If you also want to look for streaks crossing block boundaries, label that as a separate observation. A three-head run spanning positions three through five belongs to an anywhere-in-the-series question. Keep it out of the fixed-block success count so your observation matches the probability you calculated.

参考来源

这项活动用得上的工具

返回全部游戏指南