Firms use many kinds of data for forecasting future sales—one of the key activities in the management of a business—and combine various methods in order to utilize different types of information. A recent study by Nikolay Osadchiy, assistant professor of information systems and operations management; Vishal Gaur (Cornell); and Sridhar Seshadri (UT Austin) focuses on financial stock market data in developing a new methodology for firm-level sales forecasting, testing it against standard benchmarks such as forecasts from equity analysts and time-series methods. Applying their method to the forecast of total annual sales for US public retail firms, the researchers find their market-based approach achieves an average 15 percent reduction in forecasting error compared with more typical forecasting methods. Their model, they write, can also be applied to hedging operational risk with financial instruments.
Nikolay Osadchiy Associate Professor of Information Systems & Operations Management