Skip to main content
Category: Laws
Type: Statistical Law
Origin: Probability Theory, 16th-17th Century, Jacob Bernoulli
Also known as: Bernoulli’s Law, LLN
Quick Answer — The Law of Large Numbers is a fundamental principle in probability theory stating that as the number of trials or observations increases, the average result gets closer to the expected value. First rigorously proven by Jacob Bernoulli in 1713, this law explains why larger samples yield more reliable estimates and why casinos always win in the long run.

​
What is the Law of Large Numbers?

The Law of Large Numbers establishes a fundamental relationship between probability and frequency: as you repeat an experiment more times, the observed frequency of an outcome converges to its theoretical probability. In simpler terms, luck averages out over time.
“Even the most stupid of men, by some instinct of nature, is persuaded to believe that the more observations have been made, the less danger there is of wandering from one’s goal.”
This principle is counterintuitive because humans tend to over-interpret small samples. We see a “hot streak” at a casino and believe it’s due to change, or we draw conclusions from a handful of experiences. The Law of Large Numbers reminds us that patterns emerge only with sufficient data—and that short-term variance is not evidence against underlying probabilities.

​
The Law of Large Numbers in 3 Depths

  • Beginner: If you flip a coin 10 times, you might get 7 heads. But if you flip it 10,000 times, you’ll get close to 50% heads. More data = results closer to expectation.
  • Practitioner: In business, customer acquisition costs and conversion rates stabilize over larger samples. Don’t panic over small-sample fluctuations—wait for sufficient data before making decisions.
  • Advanced: The law has two forms: weak (convergence in probability) and strong (almost sure convergence). Understanding the distinction matters for financial modeling and risk assessment.

​
Origin

The Law of Large Numbers was first conceived by Jacob Bernoulli (1654-1705), a Swiss mathematician who devoted twenty years to developing a rigorous mathematical theory of probability. His work was published posthumously in 1713 in “Ars Conjectandi” (The Art of Conjecturing). Bernoulli’s insight was revolutionary: he proved that the probability of an event could be understood not just as a theoretical construct, but as something that becomes observable through repeated trials. His theorem showed mathematically what gamblers and insurers had long suspected—that random events become predictable in aggregate. Later mathematicians, including Chebyshev, Markov, and Kolmogorov, refined and extended the law, making it a cornerstone of modern statistics, insurance mathematics, and quantum mechanics.

​
Key Points

1

Large samples reduce variance

The more observations you collect, the less your results will deviate from the expected average. This is why opinion polls with larger samples are more accurate.
2

Short-term doesn't predict long-term

A streak of successes doesn’t increase your odds of continued success—the underlying probability remains constant. Each trial is independent.
3

Convergence is gradual, not instant

The law describes a tendency, not a guarantee. Even after many trials, you might still observe deviations—just smaller ones.
4

Sample quality matters as much as quantity

A large sample that’s biased will converge to the wrong value. The law assumes each trial is independent and identically distributed.

​
Applications

Insurance and Actuarial Science

Insurers can predict losses with remarkable accuracy because they have massive datasets. The Law of Large Numbers is why insurance is mathematically sound.

Quality Control

Manufacturing defects are predictable across large production runs. Quality engineers use statistical sampling to estimate defect rates.

A/B Testing

In digital marketing, A/B tests require sufficient sample sizes before you can trust the results. Small tests lead to false conclusions.

Investment Returns

Individual stock prices are highly volatile, but index funds that track thousands of companies deliver stable returns over decades—the law in action.

​
Case Study

​
The Birth of Actuarial Science

In the 17th century, the insurance industry operated largely on intuition and guesswork. Lloyd’s of London had opened in 1686, but insurers had no mathematical basis for setting premiums—they simply guessed at risk and hoped for profitability. The breakthrough came when mathematicians applied the Law of Large Numbers to mortality data. By analyzing birth and death records across entire populations, they could predict with astonishing accuracy how many people in a given age group would die in a given year. This insight transformed insurance from gambling into a science. Today, life insurance companies hold trillions of dollars in assets, confident that they can predict mortality rates within fractions of a percentage point. A life insurer knows that out of 100,000 healthy 30-year-old men, approximately 761 will die in any given year—not through crystal-ball gazing, but through the Law of Large Numbers applied to actuarial tables. The case demonstrates a broader principle: when you have enough data, the random becomes deterministic. Individual deaths are unpredictable, but population mortality is highly predictable—which is why we can have life insurance at all.

​
Boundaries and Failure Modes

The Law of Large Numbers has important limitations:
  1. Requires independent trials: If events are correlated or dependent (like in financial crises), more observations won’t help—they might make things worse.
  2. Doesn’t apply to one-time events: The law describes repeatable processes. There’s no “long run” for unique events like market crashes or natural disasters.
  3. Sample size needs can be massive: To get close to the expected value, you may need far more trials than intuition suggests. Getting within 1% might require thousands of observations.
  4. Bias doesn’t disappear with size: A biased coin will converge to its true (biased) probability, not to fairness. The law doesn’t correct for systematic errors.

​
Common Misconceptions

The Law of Large Numbers doesn’t mean you’ll see exactly 50/50 outcomes. It means the ratio will approach 50/50, but deviations can persist for a very long time.
In independent trials, the coin has no memory. After 10 heads, the probability of the next head is still 50%. This is the Gambler’s Fallacy.
Small samples can provide directional insights, especially when combined with other evidence. The law says they’re unreliable, not meaningless.

Central Limit Theorem

The finding that sample distributions approach normality as sample size increases—works with LLN to explain why statistics work.

Regression to the Mean

The observation that extreme results tend to be followed by more average results—a practical consequence of the Law of Large Numbers.

Gambler's Fallacy

The mistaken belief that past random events influence future ones—the exact opposite of what the Law of Large Numbers actually says.

​
One-Line Takeaway

Trust the pattern, not the noise. In the long run, outcomes converge to their probabilities—but you need enough data for the convergence to become visible.