Case Study: How Alternative Data Predicted Retail Earnings Surprises

An advanced case study built around Meridian Quant, a fictional Mumbai systematic fund, and its analyst Priya Nair, who is asked one question: can data that arrives before a quarter ends tell us whether DMart, Trent, Titan or V-Mart will beat or miss consensus? Each chapter is one stage of the project: mapping the retail P&L to what alternative data can actually observe, assembling an Indian data stack from GST and UPI aggregates, Google Trends, app-download panels, scraped store locators and price trackers, engineering point-in-time features, building and backtesting a revenue nowcast against Bloomberg and Refinitiv consensus, walking through three real earnings quarters where the model called a beat, called a miss, and got it wrong, and finally turning the forecast into a sized, risk-controlled event trade that survives crowding, signal decay and SEBI's insider-trading line. Built for aspiring quant researchers, prop-desk applicants and systematic traders who want to see a complete alternative-data pipeline in an Indian context, with the failures left in.