What problem does it solve? Monte Carlo studies often fail on reproducibility: hidden failed replications, non-reproducible random seeds, missing Monte Carlo standard errors, and unsupported claims about estimator performance. This Skill enforces a rigorous workflow for designing, executing, and reviewing simulation studies so results about bias, coverage, size, and power are statistically defensible. ## Core Features & Use Cases - Structured study design: Records the research question, estimand, truth formula, DGP parameters, estimator grid, replication count, and seed before any code runs. - Reproducible execution: Integrates with the waterology CLI and TORC to register, run, watch, and reproduce simulation workflows with declared inputs and outputs. - Statistical rigor: Enforces L'Ecuyer-CMRG RNG streams, per-replication raw rows, failure tracking, and MCSE reporting alongside bias, coverage, and rejection rates. - Use Case: Compare two estimators for a regression coefficient across sample sizes: define the DGP and truth, run 1,900 replications per scenario, then report bias and 95% coverage with MCSEs to determine whether differences exceed noise. ## Quick Start Ask the assistant to design and run a Monte Carlo simulation study in R comparing your estimators under a specified data generating process.
