History of Probability & Statistics

Gerolamo Cardano, a compulsive gambler, wrote the first analysis of dice odds — though the book stayed unpublished for a century.

Asked why nine and ten come up differently with three dice, Galileo worked out the exact counts — an early exercise in probability.

A gambler's puzzle about dice odds, posed to Pascal, sparked the correspondence that turned games of chance into a mathematical science.

By analysing London's bills of mortality, Graunt founded demography and the statistical study of populations from raw data.

Solving how to divide the stakes of an interrupted game, Pascal and Fermat's letters founded the mathematical theory of probability.

His Ars Conjectandi proved the law of large numbers — that observed frequencies converge to true probabilities as trials pile up.

Christiaan Huygens wrote the first printed textbook on probability, introducing the crucial idea of mathematical expectation.

This influential text was the first to weigh both the probability and the magnitude of outcomes — the seed of the idea of expected value.

De Moivre discovered the normal curve as the limit of coin-tossing and gave an early central limit theorem — plus the maths of annuities.

Pierre Rémond de Montmort's analysis of card and dice games advanced combinatorics and inspired De Moivre's later work.

Resolving the St Petersburg paradox, he argued we value money by its utility, not its amount — founding decision theory and economics of risk.

His posthumous theorem showed how to update beliefs with evidence — the foundation of Bayesian inference used everywhere today.

Laplace built the analytic theory of probability, proved a general central limit theorem, and applied inference to astronomy and society.

Richard Price published Bayes's essay on inverse probability, giving a rule for reasoning from effects back to their probable causes.

Buffon's experiment estimating π by dropping a needle across ruled lines founded geometric probability and prefigured Monte Carlo methods.

Laplace's treatise unified a century of results and made probability a powerful mathematical science of uncertainty and error.

To track a lost asteroid, Gauss perfected least squares and the normal distribution of errors — the workhorses of modern data fitting.

Poisson's distribution of rare events — from wrongful convictions to radioactive decay — remains one of statistics' most useful models.

Quetelet applied the bell curve to human traits, inventing the 'average man' and the statistical study of society.

Legendre first published the method of least squares for fitting curves to data, a workhorse of regression to this day.

Irénée-Jules Bienaymé stated the inequality later named for Chebyshev and clarified the properties of variance in sums of variables.

Pafnuty Chebyshev's inequality and rigorous proof of the law of large numbers founded the influential Russian school of probability.

Galton discovered regression to the mean and correlation, turning the study of heredity into a quantitative statistical enterprise.

Pearson built much of the machinery of statistics — the correlation coefficient, the chi-squared test — and founded its first journal.

A Guinness brewer, Gosset invented the t-test for small samples and published it under the pseudonym that made 'Student's t' famous.

The founder of modern statistics gave us maximum likelihood, the design of experiments, analysis of variance, and significance testing.

Udny Yule extended correlation into multiple regression, giving social scientists a tool to disentangle several causes at once.

Aleksandr Lyapunov gave the central limit theorem its first rigorous general proof, explaining why the bell curve appears everywhere.

Émile Borel's measure theory and strong law of large numbers supplied the rigorous language on which Kolmogorov would build his axioms.

Norbert Wiener built a rigorous mathematical model of Brownian motion, founding the theory of stochastic processes in continuous time.

Fisher's book made randomisation and controlled comparison the gold standard of scientific evidence, from agriculture to medicine.

Kolmogorov grounded probability in measure theory, finally giving the subject rigorous axioms and unifying it with modern mathematics.

This short book's three axioms turned probability into a fully rigorous branch of mathematics used across every quantitative field.

Jerzy Neyman and Egon Pearson framed hypothesis testing as a decision between errors, defining the confidence intervals still taught today.

Bruno de Finetti argued probability is degree of belief, justified by coherent betting — the philosophical core of modern Bayesianism.

Abraham Wald recast statistics as decision-making under uncertainty and invented sequential analysis, born of wartime quality control.

Claude Shannon measured information in bits and linked it to probability and entropy, founding the theory behind all digital communication.

John Nash proved every finite game has an equilibrium, giving economics and probability a rigorous theory of strategic uncertainty.

Markov chains — processes where the future depends only on the present — underlie modern modelling from genetics to web search.

Ulam and von Neumann used random sampling on early computers to solve problems too hard to compute exactly — now vital across science.

Their systematic ARIMA methodology made forecasting from time series practical, shaping econometrics and operations for decades.

Applying stochastic calculus to option pricing, this model reshaped finance and made probability central to global markets.

Tukey pioneered exploratory data analysis and the fast Fourier transform, and coined the very words 'bit' and 'software'.

Bradley Efron's resampling method let computers estimate uncertainty from data alone, freeing statistics from restrictive formulas.

Vapnik's theory of generalisation and the support vector machine gave machine learning a firm probabilistic and geometric foundation.

Judea Pearl's causal graphs and do-calculus gave statistics a rigorous language for cause and effect beyond mere correlation.

As deep networks trained on massive datasets outperform hand-built models, probability and statistics become the engine of modern AI.