Series · Economics · 4 parts
Why Simulate Demand
Bias requires truth. A simulator lets you set the truth, hand an estimator only the observable view, and measure how far off it lands — the one measurement real market data can never give you.
Parts
Why Simulate Demand at All?
Bias requires truth. A simulator lets you set the truth, hand an estimator only the observable view, and measure how far off it lands — the one measurement real market data can never give you.
Research Foundations — How Demand Curves Have Been Estimated
Five method families the simulator is built to replicate and stress-test: Uber's regression discontinuity, ZipRecruiter's randomized pricing, Double Machine Learning, BLP / conjoint, and Amazon-scale forecasting.
Why Inventory Belongs in the Simulator
Stockouts censor sales below true demand. The doom loop, elasticity bias, and forecast-evaluation corruption — why keeping true demand in the oracle changes every downstream number.
Model Playbook — The Estimators, in Python and R
Log-log OLS, binary and mixed logit, Double ML, regression discontinuity, BLP, censored-demand corrections, and probabilistic forecasting — each with runnable Python and R.
Includes 1 interactive reportExperimentation Design — Randomization Units, Sample Sizes, and Guardrails
How to run a pricing experiment that actually identifies a demand curve: per-user, per-session, switchback, and geo randomization; power; and analytical guardrails (SRM, A/A, balance, contamination).