LQ45 Stock Portfolio Optimization by considering Return Predictions using the Holt Winter Method

Abstract

A portfolio represents a selection of assets owned by individuals or groups with the goal of generating profit. Stocks are among the various forms of investments available. Investors typically consider two key factors: expected returns and associated risks. Portfolio optimization endeavors to maximize returns while minimizing risks. With the progression of time, portfolio optimization strategies increasingly incorporate return predictions using machine learning techniques. In this study, the Holt-Winter method is employed to forecast stock prices and returns, known for its accuracy in estimating seasonal time series data. Previous research primarily utilized the Mean- Variance method for portfolio optimization, but the outcomes were deemed unsatisfactory. Therefore, this study adopts a novel approach by integrating the Mean-Variance model with forecasting. Performance testing of the optimal portfolio emphasizes sensitivity, aiming for substantial average returns, minimal standard deviation, and a high Sharpe ratio. Comparative analysis with the LQ45 index portfolio reveals that the sensitivity-driven portfolio exhibits superior performance, yielding higher values in terms of average returns, standard deviation, and Sharpe ratio.