Modeliranje volatilnosti financijskih vremenskih nizova zasnovano na nadziranom učenju

undergraduate thesis

• Title (english): Modelling financial time series volatility based on supervised learning
• Author: Filip Aleksić
• Mentor: Stjepan Begušić (mentor)
• Committee member: Stjepan Begušić (predsjednik povjerenstva)
• Committee member: Lovre Mrčela (član povjerenstva)
• Committee member: Tessa Bauman (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-08-26, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad se bavi predikcijom volatilnosti i procjenom matrice kovarijance koristeći metode strojnog učenja. U uvodu je naglašena važnost razumijevanja i predviđanja volatilnosti financijskih instrumenata za investitore i financijske analitičare. Teorijska osnova uključuje objašnjenje volatilnosti, kovarijance, korelacije, beta koeficijenata i Markowitzove teorije portfelja. Metodologija rada opisuje prikupljanje i pripremu podataka, te primjenu različitih modela strojnog učenja, uključujući linearnu regresiju, slučajne šume, XGBoost i unaprijednu neuronsku mrežu. Rezultati pokazuju da modeli strojnog učenja, posebno XGBoost, mogu efikasno predvidjeti volatilnost i doprinijeti preciznijoj procjeni matrice kovarijance. Ovi modeli su evaluirani na osnovu njihove sposobnosti minimiziranja volatilnosti i maksimiziranja povrata portfelja. Zaključeno je da napredni modeli strojnog učenja imaju potencijal značajno poboljšati upravljanje portfeljima i donošenje investicijskih odluka.

Abstract (english)

This thesis deals with the prediction of volatility and the estimation of the covariance matrix using machine learning methods. The introduction highlights the importance of understanding and predicting the volatility of financial instruments for investors and financial analysts. The theoretical foundation includes explanations of volatility, covariance, correlation, beta coefficients, and Markowitz's portfolio theory. The methodology section describes data collection and preparation, and the application of various machine learning models, including linear regression, random forests, XGBoost, and a multilayer perceptron neural network. The results show that machine learning models, especially XGBoost, can effectively predict volatility and contribute to a more accurate estimation of the covariance matrix. These models are evaluated based on their ability to minimize volatility and maximize portfolio returns. The conclusion is that advanced machine learning models have the potential to significantly improve portfolio management and investment decision-making.

• Keywords: kovarijanca; volatilnost; predvidanje; strojno ucenje
• Keywords (english): kovarijanca; volatility; prediction; machine learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:990226
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2024-10-12 19:57:57