On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

Math · Statistics & Probability · Grade 9-12

Law of Large Numbers (Intuition)

The law of large numbers says that as you run more independent trials, the running average Xˉn settles toward the true mean μ.

Xˉn→μ as n→∞

Quick Answer

The law of large numbers says that as you run more independent trials, the running average Xˉn settles toward the true mean μ. Use it to explain why casinos, insurers and pollsters can rely on long-run averages even though single outcomes are unpredictable. It works by dilution, not correction: an early lucky run is never cancelled out, it just becomes a smaller and smaller part of a growing total. The proportion converges, while the raw counts of heads and tails can still drift apart.

Why This Matters

This law is why probability works in practice. A casino cannot predict one roulette spin, yet over millions of spins it earns very close to its 2.7% edge, and an insurer can price policies because the average claim per customer becomes predictable across thousands of customers. It also tells you what not to expect: over 10 or 20 trials, an average can sit far from the true mean.

Intuitive Explanation

Flip a fair coin and track the share of heads. After 10 flips you might have 7 heads, a share of 0.70. Keep going. After 100 flips the share is very likely to be between about 0.40 and 0.60; after 10,000 flips, between about 0.49 and 0.51. The running proportion jumps around at first and then flattens toward 0.5. The law of large numbers says this flattening happens for any sequence of independent trials: Xˉn→μ as n grows.

The mechanism is dilution. Suppose those first 10 flips gave 7 heads and 3 tails, 4 more heads than tails. The next 990 flips are fair and give about 495 heads, so the total is about 7+495=502 heads in 1,000 flips, a share of 0.502. The early surplus did not go away; heads still lead by about 4. It just matters far less once it is spread over 1,000 flips. Nothing pushes the coin toward tails to restore the balance.

Two ideas get confused with this law. The gambler's fallacy claims that after five heads, tails is due; the law says no such thing, because each flip stays at 12. And the law is about averages, not counts. The typical gap between the number of heads and tails grows from about 2 or 3 after 10 flips to about 80 after 10,000, even though the proportion ends up far closer to 0.5. The average settles; the count gap does not have to.

Think of it like this. Picture a few drops of food colouring in a glass, with clear water poured in and in. The colour fades as the volume grows, even though the drops are never taken out. An early lucky run in a long series of trials behaves the same way: it is diluted, not undone. The picture leaves out one thing: each new trial adds a little fresh randomness of its own.

Core idea. As trials grow, the sample mean converges to the true population mean — randomness averages out.


When to Use

Use the law of large numbers when a question involves a long run of independent trials: why a casino or insurer can rely on averages, why a large poll is steadier than a small one, or what happens to a running proportion as the number of trials grows. It is also the tool for answering claims that an outcome is due. For a short run of 10 or 20 trials, expect wide swings instead; to describe the shape of the spread of sample means, you need the Normal Distribution.

Tip. Ask: Is the average of many independent trials settling toward the true mean as n grows?

How to Recognize It

Before you invoke the law of large numbers, check that the trials are independent repeats of one process and that the question is about an average or proportion, not a raw count or the next single result.

  1. Is the question about what the next flip, spin or roll will be?

    If yes, the law says nothing about it: each trial is as unpredictable as the first. Claiming a result is due after a run is the gambler's fallacy, which this law does not support.

  2. Is the quantity a proportion or an average, rather than a total count?

    If yes, it settles as n grows. A count gap like heads minus tails typically grows like n, so a gap of 50 in 10,000 flips is ordinary, yet it moves the share of heads only to 0.5025.

  3. Does the question ask for the value the average settles at, such as 5.5 for a spinner numbered 1 to 10?

    Then it wants the expected value μ, the target. The law of large numbers says the running average heads toward that target; it is not the target itself.

  4. Does the question ask how spread out averages of n trials are, or what shape their histogram has?

    Then it is the central limit theorem's job: sample means pile up in a roughly normal shape whose width shrinks like σn. The law of large numbers only promises that they close in on μ.

Law of Large Numbers (Intuition) compared with Gambler's fallacy, Expected value, and Central limit theorem

All four rows describe what happens over many repeated trials, so a long-run fact is easily turned into a promise about the next trial. Ask whether you need the target value, the fact that averages approach it, the shape of their spread, or a prediction the law never makes.

Law of Large Numbers (Intuition)

Meaning
Use this when you explain why an average of many independent trials becomes stable and predictable.
Key test
Is the average of many independent trials settling toward the true mean as n grows?
Formula
Xˉn→μ as n→∞
Example
After 10 flips you have 7 heads (70%). What does the law of large numbers predict as you keep flipping?

Gambler's fallacy

Meaning
Wrongly claims past outcomes must be balanced soon, the opposite of this law.
Key test
Never use it; it's a fallacy to recognize and reject.
Example
Expecting heads after five tails

Expected value

Meaning
Is the target mean μ that the average converges to, not the convergence itself.
Key test
Use when computing the long-run average value of outcomes.
Formula
E[X]=∑xiP(xi)
Example
Average die roll of 3.5

Central limit theorem

Meaning
Describes the shape (normal) of sample-mean variation, not just its convergence.
Key test
Use when you need the distribution of the sample mean.
Example
Sample means forming a bell curve

Formula & Notation

Xˉn→μ as n→∞
Xˉn→Pμ as n→∞; i.e., for any ε>0, P(∣Xˉn−μ∣>ε)→0

How to read it: Xˉn is the sample mean after n trials; μ is the true population mean

Worked Examples

Example 1 — Running coin proportion

Easy

After 10 flips you have 7 heads (70%). What does the law of large numbers predict as you keep flipping?
Show the steps and the answer
  1. Identify the true mean: for a fair coin, μ=0.5 heads per flip.

    The law says the running proportion heads toward this value, so it is the target.

  2. Project 990 more fair flips: about 990×0.5=495 heads.

    Future flips are independent of the first 10, so on their own they average 0.5.

  3. Combine: about 7+495=502 heads in 1,000 flips, a share of 502÷1000=0.502.

    The early 70% has been diluted to a share just 0.002 above 0.5.

  4. Push further: after 100,000 flips, the 2 extra heads from the first 10 flips shift the share by only 2÷100000=0.00002.

    The early surplus never vanishes; it just becomes a negligible part of the total.

Answer: The proportion converges toward 0.5

Takeaway: The proportion converges toward 0.5 by dilution: the early surplus stays, but it is spread over more and more flips.

Example 2 — Gambler's fallacy

Standard

After 5 heads, a friend says tails is 'due' to even things out. Is that the law of large numbers?
Show the steps and the answer
  1. State the friend's claim: the next flip is more likely than 12 to be tails.

    Saying tails is due is a claim about one particular future flip.

  2. Check independence: the coin keeps no record of the last 5 flips, so P(tails)=12.

    Independent trials cannot change their probability in response to past results.

  3. Compare with what the law does say: after 5 heads, 1,000 more fair flips give about 500 heads, a share of about 505÷1005≈0.502.

    The law gets close to 0.5 by swamping the run of heads with new flips, not by producing extra tails.

Answer: No — that's the gambler's fallacy

Takeaway: The law of large numbers works by dilution. It never makes any particular outcome due.

Example 3 — Find the mistake: the growing gap

Application

A student flips a coin 10,000 times and gets 5,080 heads and 4,920 tails. After her first 100 flips she had 54 heads and 46 tails. Jae writes: "The law of large numbers isn't working: the gap between heads and tails grew from 8 to 160." Find the mistake and correct it.
Show the steps and the answer
  1. Compute the proportions: 54÷100=0.54 and 5080÷10000=0.508.

    The law makes a claim about the proportion, so that is the number to check.

  2. Compare distances from 0.5: 0.04 after 100 flips, 0.008 after 10,000.

    The proportion moved five times closer to 0.5, as the law predicts.

  3. Explain the count gap: the typical gap between heads and tails grows roughly like n, from about 8 at n=100 to about 80 at n=10000.

    Counts are not expected to balance, and a gap of 160 is only 160÷10000=1.6% of the flips.

  4. Correct the conclusion: the law is working, since the proportion is converging even though the count gap grew.

    Jae checked the raw counts, a quantity the law makes no promise about.

Answer: The law is working: the share of heads moved from 0.54 to 0.508, closer to 0.5. The gap in counts is expected to grow.

Takeaway: The law of large numbers is about averages and proportions. Raw counts of heads and tails can drift further apart as n grows.

Where You'll See It

  • Why the casino always comes out ahead

    In European roulette, a $1 bet on red wins with probability 1837 and loses with probability 1937, an average loss of 137≈0.027 dollars, about 2.7 cents, per dollar bet. Over a million $1 bets, the casino's take is very close to $27,000, give or take about $2,000, even though no single spin can be predicted.

  • Pricing home insurance

    If 1 home in 1,000 has a $200,000 fire claim each year, the expected cost is $200 per home. With 100,000 customers the insurer expects about 100 claims, and the actual number usually stays within about 20 of that, so a premium a little above $200 reliably covers the claims.

  • Why polls ask a thousand people

    A random poll of 10 people can easily miss the true share by 20 percentage points; a random poll of 1,000 is usually within about 3 points. The law of large numbers is what makes the bigger poll steadier, though it only helps if the sample is random rather than biased.

Common Mistakes

  1. The slip: Using it to predict short runs

    Instead: convergence is a long-run effect; 10 flips can stray far from 50%.

  2. The slip: Slipping into the gambler's fallacy

    Instead: the average approaches μ, but no single outcome is 'due'.

  3. The slip: Expecting the count, not the average, to balance

    Instead: it's the proportion that converges, while raw counts can drift apart.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. A fair die is rolled 6,000 times. About how many sixes should come up, and about what proportion is that?

    Hint: Multiply the number of rolls by 16.

  2. After 100 flips there are 55 heads. After 10,000 flips there are 5,080 heads. What happened to the proportion of heads, and to the gap between heads and tails?

    Hint: Compute the percent of heads and the heads-minus-tails gap at both stages.

  3. A $1 casino bet has an expected value of −0.05 dollars for the player. About how much should the casino expect to win over 100,000 such bets?

    Hint: Over many bets, the average result per bet settles near the expected value.

  4. A small hospital has about 15 births a day and a large one about 45. Over a year, which has more days when over 60% of babies born are boys?

    Hint: Which sample size lets the daily proportion swing further from 50%?

  5. A question asks what shape the averages of 400 dice rolls would form if you repeated the experiment many times. Is this the law of large numbers or the central limit theorem?

    Hint: One idea says where the average settles; the other describes the shape of many averages.

  6. After 20 flips there are 14 heads and 6 tails. Rio says: "By the law of large numbers, the next flips must have extra tails to even things out." Correct him.

    Hint: Does the coin know about the earlier flips?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Which hospital records more days when over 60% of the babies born are boys, a small one or a large one?

The small one, though most people guess the two are about equal. The puzzle comes from the psychologists Daniel Kahneman and Amos Tversky. With about 15 births a day, more than 60% boys happens on roughly 15% of days; with about 45 births a day, on roughly 7%. Averages of fewer trials swing further from one half, which is the law of large numbers seen from the other side.

Why do insurers worry about risks that hit many customers at once?

Because the law needs independent trials. An insurer with 100,000 homes spread across a country can predict its yearly claims closely, since one family's kitchen fire says nothing about another's. Homes along one coast are different: a single hurricane can damage thousands together, so their claims do not average out. Insurers handle such shared risks by spreading them across regions and by buying insurance of their own, called reinsurance.

Does every long-run average settle down eventually?

Not every one. The law also needs repeats of the same process with a finite mean. If the process changes, say a basketball player improves over a season, the average chases a moving target. In heavy-tailed situations, where rare giant values keep arriving, like a few viral videos among millions, the average can lurch whenever a new giant shows up, and for some extreme processes it never settles at all.

Learning Path

← Before

ProbabilityMean
Law of Large Numbers (Intuition)

You are here

You need Probability and the Mean first: the chance of an outcome and the average of a set of values. The law of large numbers connects them, saying that the average of many independent trials closes in on the true mean. Next, the Normal Distribution describes the bell-shaped spread of averages around that mean, which tells you how close to expect them to be for a given n.

See Also