Loss Functions in Financial Sector: An Overview

Abstract

The loss function or cost function is a principal component of all optimizing problems such as statistical theory, decision theory, policy making, estimation process, forecasting, learning, classification and financial investment. In this study various types of loss functions and their uses were studied. In this study linex loss function, the analyst's loss function, the optimal prediction under asymmetric loss, loss functions for forecasting financial returns, loss function and forecast biasedness, loss functions for estimation and evaluation, loss functions for probability forecasts and volatility forecasts have been studied. Further loss function for testing Granger-causality and loss function for uncertainty and policy aggressiveness have been also studied. This study suggested that among asymmetric loss functions linex loss function is more useful and to be applied in many financial situations.