EUROPT 2024
Abstract Submission

86. Analysis of Finance-Human Factor Interactions Using Various Indicators

Invited abstract in session FD-4: Optimal control and stochastic optimal control - theory, methods and applications 1, stream Optimal control and stochastic optimal control - theory, methods and applications.

Friday, 14:10 - 15:50
Room: M:M

Authors (first author is the speaker)

1. Betül Kalaycı
Financial Mathematics, Institute of Applied Mathematics
2. Vilda Purutcuoglu
Statistics, Middle East Technical University
3. Gerhard-Wilhelm Weber
Faculty of Engineering Management, Poznan University of Technology

Abstract

There are several empirical phenomena regarding the behaviour of individual investors, such as how their emotions and opinions influence on their decisions. The Sentiment term describes all of these emotions and opinions. In finance, stochastic changes may occur based on sentiment levels of investors. Machine Learning methods are well-known and useful tools for prediction problems, and they have already been used successfully to solve a variety of financial problems. In this study, apart from pure financial related challlenges, we focus on behavior of financial problems which is based on the investors' sentiment levels. Accordingly, the purpose of this study is to assess sentiment index predicting performance by utilizing two-stage hybrid models which are the composition of multivariate adaptive regression splines, random forest and neural networks. We additionally intend to define people's attitudes on the economy based on their levels of confidence. Hereby, we perform hidden markov model (HMM) to estimate the underlying state of changes in the Consumer Confidence Index and to examine its link with some macroeconomic indicators (CPI, GDP, and currency rate) at monthly intervals. Our goal is to observe and understand the transition between these phases, as well as to define a path through them. Furthermore, we apply volatility models to each subgroup generated via HMM in order to check whether forecasting outcomes can be improved.

Keywords

Status: accepted


Back to the list of papers