Thesis Institute · policy futures
Forecasts on every consequential cell of government data
Three coupled forecast types fall out of the integrated stack: government data points on published statistics, policy state forecasts on formal settings and encoded parameters, and conditional forecasts on outcomes given policy states. Thesis shows the target agent workflow: call public data and the PolicyEngine microsim, then publish calibrated uncertainty with an audit trail behind it.
How forecasts are generated
Every forecast cell is opened by the Thesis Institute analyst agent, which decomposes the question, calls the PolicyEngine microsim against scenarios drawn from encoded statutes and MICROPLEX synthetic populations, checks official public data sources, integrates external baselines (CBO, FOMC SEP, JCT, BLS, Census, ONS, ABS, Statistics Canada), and produces a calibrated distribution. Open each forecast to read the full streaming reasoning.