Development of Comparative Forecasting Models of Daily Prices of Aggressive Pension Mutual Funds by Univariate Time Series Methods

Abstract

The primary goal of the individual pension system is to enhance retirees' living standards by generating supplementary income through the investment of their savings during retirement. This involves guiding individuals to invest their savings in pension mutual funds. This research aims to develop comparative forecasting models using Autoregressive Integrated Moving Average (ARIMA) and Exponential Smoothing techniques for the daily prices of pension mutual funds categorized as aggressive risk. The study utilizes data from 2020 to 2023 pertaining to a pension mutual fund provided by a Turkish pension company. The dataset is split into a training set (75%) and a validation set (25%). Mean Absolute Percentage Error (MAPE) is employed to gauge the error measurement values of the training and validation sets of the developed forecasting models. The findings reveal that for the validation sets, ARIMA model performs best for the İş Bank participation index funds, whereas Exponential Smoothing forecasting models yield the lowest MAPE values for equity, group equity, and secondary equity funds. This research can serve as a decision-making tool for the effective management of high-yield pension mutual funds and aid pension companies in enhancing the appeal of their product offerings to customers.