Estimating Style Weights of Mutual Funds by Monte Carlo Filter with Generalized Simulated Annealing

Abstract

This paper proposes a new approach to style analysis by applying a general state space model and Monte Carlo filter. Particularly, we regard coefficients of style indices as state variables in the state space model and employ Monte Carlo filter as an estimation method. Moreover, we utilize a generalized simulated annealing for estimating parameters, which seems the first attempt in particle filtering methods. Finally, an empirical analysis with actual funds' data confirms the validity of our approach.