Beginner to Ninja: Math for Causal Inference
A 100-day bootcamp that builds every piece of mathematics behind modern causal inference, from probability axioms to double machine learning.
Every day is one 2-hour session: a short, rigorous theory block, fifteen exercises that climb from mechanics to proofs to applied economics and public-health problems, a Python notebook that checks the answers by simulation, and an interactive simulator in the Simulation Lab.
Progress is saved in this browser. 45 of 100 days are published.
How each day works
The definitions, theorems and the one idea that links today's maths to causal inference.
Mechanical computation: apply the formulas until they feel automatic.
Derive the results yourself, step by step, from what you already know.
Economics and public-health problems, then a Python notebook and a simulator.
The six phases
Axioms, conditional probability, expectations, distributions, the CEF, LLN/CLT, estimators, intervals, tests, MLE, the delta method, Bayes and missing data.
Vectors and matrices, rank, projections, eigenvalues, calculus and optimisation, and OLS derived in matrix form.
The Rubin causal model, SUTVA, estimands, randomisation inference, unconfoundedness, propensity scores and doubly robust estimation.
DAGs, d-separation, structural causal models, the back-door and front-door criteria, do-calculus and mediation.
Instrumental variables and LATE, difference-in-differences with staggered adoption, regression discontinuity and bounds.
Matching and balancing, synthetic control, influence functions, double/debiased ML, heterogeneous effects and policy learning.
From the Simulation Lab
All 45 simulators →A probability tree and a 1,000-person grid: see how prevalence drives the meaning of a positive test.
Drag to rotate the joint density; change the correlation and see the contours and the regression line.
Draw thousands of sample means from skewed or bimodal parents and watch the normal shape appear.
The null and alternative sampling distributions, the rejection region, α, β and power.