Data Science
Time Series · R · Forecasting

Forecasting S&P 500 Sector Earnings with Survey Data and Time Series Models

By Samuel Alarcón · Econometrician & Data Scientist

Sector-level dispersion often tells a more useful story for asset allocation than index-level earnings forecasts. This case study covers how I built a sector-level S&P 500 earnings forecasting model in R, combining time series methods with Blue Chip Financial Forecasts survey data.

The modelling approach

The model accounts for shared macro drivers across sectors while allowing sector-specific earnings sensitivities to diverge — useful for spotting which sectors are expected to outperform or lag the broader index.

Backtesting and validation

Every forecast vintage is backtested against realized earnings, with visual diagnostics showing where the model over- or under-shoots by sector and forecast horizon.

View on GitHubDiscuss a similar project