Al and traditional forecasting of currency return using Diebold Li factors from commodities

Abstract

This research paper investigates out of sample forecasting AI compared to traditional models. The data used are oil crude futures prices and the currency pair NOK/USD. The crude oil futures prices is transformed to the commodity basis and termstructure of crude oil futures prices. The Diebold Li factors is extracted from the termstructure. These factors and the currency return are used with various models to forecast currency return. Encouraging results for the use of AI in out of sample forecasting is found.