economic-forecasting

What problem does it solve?

This Skill helps you produce credible macroeconomic forecasts and compare competing models while avoiding look-ahead bias and quantifying uncertainty.

Core Features & Use Cases

• ARIMA and VAR forecasting: Use ARIMA for univariate series (e.g., GDP, CPI, unemployment) and VAR for multi-variable macroeconomic systems.
• ML ensemble forecasting with LightGBM: Train on lag and rolling-stat features plus calendar variables to forecast horizons using walk-forward validation.
• Model comparison and uncertainty visualization: Run the Diebold-Mariano test to evaluate predictive accuracy differences and generate bootstrap-based fan charts for forecast bands; optionally use Mincer-Zarnowitz regression for unbiasedness checking.
• Real-time data vintages via ALFRED/FRED: Retrieve time-series vintages to reduce look-ahead bias when working with updated macro releases.

Quick Start

Ask an AI to build an ARIMA vs LightGBM walk-forward forecast on your quarterly GDP-like series, run a Diebold-Mariano test on the resulting forecast errors, and output a fan chart for your target horizon.