Law of Large Numbers (Intuition) Examples in Math

Start with the recap, study the fully worked examples, then use the practice problems to check your understanding of Law of Large Numbers (Intuition).

This page combines explanation, solved examples, and follow-up practice so you can move from recognition to confident problem-solving in Math.

Concept Recap

The law of large numbers states that as the number of independent trials increases, the sample mean converges to the true population mean โ€” randomness averages out over many repetitions.

As the number of trials grows, the sample mean converges to the true expected value โ€” randomness averages out over many trials, making the average predictable.

Read the full concept explanation โ†’

How to Use These Examples

  • Read the first worked example with the solution open so the structure is clear.
  • Try the practice problems before revealing each solution.
  • Use the related concepts and background knowledge badges if you feel stuck.

What to Focus On

Core idea: As trials grow, the sample mean converges to the true population mean โ€” randomness averages out.

Common stuck point: The procedure for law of large numbers (intuition) is the easy part; the trap is using it to predict short runs. Asking "Is the average of many independent trials settling toward the true mean as nn grows?" first is what keeps a correct-looking calculation from being attached to the wrong concept.

Sense of Study hint: Ask: Is the average of many independent trials settling toward the true mean as nn grows?

Worked Examples

Example 1

easy
A fair coin is flipped. Show how the proportion of heads approaches 0.5 as nn increases, using simulation results: n=10n=10: 6 heads; n=100n=100: 53 heads; n=1000n=1000: 498 heads.

Answer

Proportions: 0.60, 0.53, 0.498 โ€” converging to 0.5 as nn grows.

First step

1
n=10n=10: proportion =6/10=0.60= 6/10 = 0.60 (60% โ€” far from 0.5)

Full solution

  1. 2
    n=100n=100: proportion =53/100=0.53= 53/100 = 0.53 (53% โ€” closer to 0.5)
  2. 3
    n=1000n=1000: proportion =498/1000=0.498= 498/1000 = 0.498 (49.8% โ€” very close to 0.5)
  3. 4
    Pattern: as nn increases, Xห‰โ†’0.5=ฮผ\bar{X} \to 0.5 = \mu, illustrating the Law of Large Numbers
The Law of Large Numbers states that as sample size nโ†’โˆžn \to \infty, the sample mean Xห‰\bar{X} converges to the true population mean ฮผ\mu. Small samples can show large deviations from the truth; large samples reliably estimate it.

Example 2

medium
A casino game has expected value โˆ’$0.05-\$0.05 per play (house edge). A player plays 10 games vs. 10,000 games. Explain how the LLN affects the likely outcome in each case.

Example 3

easy
A coin shows 77 heads in 1010 flips, 5151 heads in 100100 flips, and 5,0035{,}003 heads in 10,00010{,}000 flips. State each proportion and which is closest to the truth.

Example 4

medium
A simulation estimates ฯ€\pi by randomly throwing darts at a square containing an inscribed circle. After many darts, the ratio (inside circle)/(total) should approach what value, and why?

Example 5

medium
An insurance company writes 10,00010{,}000 identical home policies with expected payout $300\$300 each. By LLN, what total payout does it expect?

Example 6

hard
A roulette wheel has 1818 red, 1818 black, 22 green pockets. A player bets $1\$1 on red each spin (pays $1\$1 if red, โˆ’$1-\$1 otherwise). Find expected loss per spin, and the expected total loss over 10001000 spins.

Example 7

hard
Sketch (describe) what a graph of running mean vs number of die rolls would look like as nn goes from 11 to 10,00010{,}000.

Example 8

challenge
Explain why a casino with a 5%5\% house edge essentially cannot lose money over a year of 1010 million bets, even though it can lose any single bet.

Practice Problems

Try these problems on your own first, then open the solution to compare your method.

Example 1

easy
A die is rolled repeatedly and the running average is tracked. After 5 rolls, the average is 3.8. After 500 rolls, the average is 3.52. What does the LLN predict will happen to the average as rolls โ†’ โˆž?

Example 2

hard
A student argues: 'I've flipped heads 10 times in a row, so tails is overdue.' Using the LLN correctly, explain why this is wrong, and what LLN actually predicts.

Example 3

easy
As the number of independent trials increases, the observed proportion approaches the ___.

Example 4

easy
A fair coin is flipped 10,000 times. The proportion of heads will be close to what value?

Example 5

easy
A fair die's average roll over many throws approaches what number?

Example 6

easy
After 5 heads in a row with a fair coin, is the next flip more likely to be tails?

Example 7

easy
Does the law of large numbers guarantee that 100 flips give exactly 50 heads?

Example 8

easy
To estimate a coin's true bias, is it better to flip it 20 times or 5,000 times?

Example 9

easy
Does the LLN apply to a single trial?

Example 10

easy
Casinos rely on the LLN to ensure that over millions of bets their average profit per bet approaches the ___.

Example 11

medium
A die is rolled. After 12 rolls the average is 4.1; after 6000 rolls it is 3.51. Which estimate of the true mean (3.5) is better and why?

Example 12

medium
A gambler says: 'Red hasn't hit in 8 spins, so it's due.' Which two ideas is he confusing?

Example 13

medium
A coin shows 600 heads in 1000 flips (0.60). Over the next 1,000,000 flips, will the overall proportion move toward 0.50 by future flips averaging out the early surplus, or by the early flips becoming negligible?

Example 14

medium
Estimating P(rain)P(\text{rain}) from history: 30 rainy days out of 100 vs 300 out of 1000. Both give 0.30. Which estimate is more reliable and why?

Example 15

medium
A simulation estimates ฯ€\pi by random points; with more points the estimate stabilizes near 3.14159. Which law explains this convergence?

Example 16

medium
Insurance: one policyholder's loss is unpredictable, but an insurer with 1,000,000 policies can predict total payouts well. Why?

Example 17

medium
Why can a casino with a 1% edge still lose money on a single big bet but never over a year?

Example 18

medium
A weather app says '60% chance of rain' on many days. Over a long record, on what fraction of such days should it actually rain if well-calibrated?

Example 19

medium
Rolling two dice, the average of the sum over 10,000 rolls approaches what value?

Example 20

challenge
A fair coin shows 520 heads in 1000 flips (0.520). To dilute this 0.020 surplus so the running proportion is within 0.005 of 0.50, roughly how many additional fair flips are needed (assume they land near 0.50)?

Example 21

challenge
Estimating a proportion, standard error is 0.5/n0.5/\sqrt{n}. How many trials are needed so the estimate is within 0.010.01 of the truth (i.e. SE โ‰ค0.01\le 0.01)?

Example 22

challenge
To halve the standard error of a sample mean (which scales as 1/n1/\sqrt{n}), by what factor must nn increase?

Example 23

easy
A fair 6-sided die is rolled many times. What does the LLN predict for the average of the rolls?

Example 24

easy
After tossing a fair coin 5050 times you got 3030 heads. The LLN says the proportion of heads will trend toward ___ as more flips are made.

Example 25

easy
To estimate the long-run average of a slot machine, would you rather play 2020 pulls or 20,00020{,}000 pulls?

Example 26

medium
A casino game has expected value โˆ’$0.04-\$0.04 per dollar bet. Over 11 million $1\$1 bets, what is the casino's expected gross profit?

Example 27

medium
A spinner has P(red)=0.3P(\text{red}) = 0.3. In 10001000 spins, roughly how many reds do you expect?

Example 28

medium
After 100100 fair coin flips you observed 6060 heads (a 1010-head surplus). True or false: by LLN, you should expect 1010 extra tails in the next 100100 flips to 'balance' the imbalance.

Example 29

medium
A weather model is 'well-calibrated' if on days it says '70% chance of rain', it rains about 70%70\% of the time long run. Which law underpins this?

Example 30

medium
A pollster surveys 2525 voters in a state of 55 million. Why does this small sample produce a noisy estimate of voter preference?

Example 31

medium
Two children play a fair game 1010 times. If one child has won 77 rounds, can we predict they have a long-run advantage?

Example 32

hard
For a fair coin, what is the standard deviation of the proportion of heads in n=10,000n = 10{,}000 flips, and how close to 0.50.5 is the proportion likely to be?

Example 33

hard
How many fair-coin flips nn are needed so the proportion of heads is within 0.0010.001 of 0.50.5 with 95%95\% confidence (use 2ฯƒ2\sigma rule, ฯƒ=0.5/n\sigma = 0.5/\sqrt n)?

Example 34

hard
A student rolls a die 3636 times and gets a sum of 150150. Does this give strong evidence the die is biased?

Example 35

hard
Strong vs Weak LLN: state the difference in one sentence each.

Example 36

challenge
You bet $1\$1 at fair odds on a 0.510.51-probability event for 10,00010{,}000 rounds (pays $1\$1 on win, โˆ’$1-\$1 on loss). Find expected total winnings and the (CLT-based) standard deviation of total winnings.
โ† Back to Law of Large Numbers (Intuition) Formula Explained Practice Mode

Background Knowledge

These ideas may be useful before you work through the harder examples.

probabilitymean