MARKOV CLUSTERING FOR PORTFOLIO CONSTRUCTION UNDER STOCHASTIC ENVIRONMENT

Tri Handika, Arief Wibisono Lubis

Abstract

Until recently there were still many new investors andfinancial consultants who
face dificulties in stocks portfolio construction, both in terms of selection and
deciding how large portion ofeach asset in the portfolio. It takes relatively longer
time and hence they constantly strive to achieve faster portfolio construction
because timely information can mean the difference between a deal struck or
missed, which translates to substantial profit or loss. This paper aims to analyze
the efficiency ofMarkov clustering processes for portfolio construction in order to
speed up assets selection based on correlation principle. Furthermore, portfolio
optimization for selected assets will be achieved with Markovian modeldriven by
a Brownian motion process under stochastic environment. We compare the
performance ofthe constructed portfolio to LQ45, Kompcisioo, and Bisnis2y indices
using Sharpe Ratio, and the results show that it outperforms these benchmark
indices. Hence, investors might use Markov clustering technique in the stocks
selection as an alternative since it is more efficient in terms oftime and in this case
proven to provide better reward to risk taken by the investors.

Full Text:

PDF