A density functional methodology for the computation of accurate density forecast

Abstract

The development objective of the project was to use numerical and simulation techniques to construct density forecasts, with attempt to solve an open problem proposed in the Berkowitz (2001) article. The policy objective of the project, to which its immediate objectives contribute, was to introduce a unified statistical framework tool to evaluate, and compare the adequacy of possibly misspecified GARCH forecasting models. The project had four immediate objectives: (i) developing accurate, efficient, and robust' numerical algorithm to estimate the probability density, with particular reference to problems that involve density forecast; (ii) answering the questions rose in pervious research; (iii) investigate the question of when the problem requires evaluating the whole distribution or only the tail of the distribution; (iv) providing essential reference, for academic researchers interested in empirical finance, as well as practitioners such as risk and portfolio management in the finance industry.