Model predictive control in optimizing stock portfolio based on stock prediction data using Holt-Winter’s exponential smoothing

Abstract

Abstract This study aims to solve an optimization problem on stock portfolio. There are two main subtopics of this research, namely stock price prediction using Holt Winter’s Exponential Smoothing method and stock portfolio optimization using Model Predictive Control (MPC) method. The steps that have been taken are: collecting and analysing stock price data, determination of smoothing parameters and stock price prediction, calculation of stock price prediction returns, determining the stock portfolio model and system constraints, converting the objective function into a quadratic programming form, and initialization of optimization parameters and program simulation. Based on the simulation results, all control variables are within the predetermined constraints. The application of MPC in optimizing all capital in a portfolio based on stock price predictions can provide satisfactory results. This is represented by the decisions given by the MPC which resulted in the investor’s total capital increasing closer to the expected target.