Model dubokog učenja za optimizaciju portfelja

master's thesis

• Title (english): Deep learning model for portfolio optimization
• Author: Dora Ercegović
• Mentor: Stjepan Begušić (mentor)
• Committee member: Stjepan Begušić (predsjednik povjerenstva)
• Committee member: Zvonko Kostanjčar (član povjerenstva)
• Committee member: Juraj Petrović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-10, Croatia

Abstract

U ovom radu dan je teorijski uvod u problem optimizacije portfelja, te mean-variance optimizacije kao jedan od standardnih pristupa rješavanju ovog problema. Zatim je predstavljeno područje dubokog učenja, uz naglasak na LSTM mreže. Predstavljena je implementacija modela dubokog učenja, s LSTM arhitekturom, koji iz podataka o povratima uče optimalne alokacije imovina s ciljem maksimizacije Sharpeovog omjera, odnosno omjera povrata i rizika portfelja. Modeli su ispitani na povijesnim tržišnim podatcima uz statističku analizu rezultata u okviru financijskih mjera performansi. Kao referentna točka za procjenu kvalitete rada modela korišteni su portfelji dobiveni mean-variance optimizacijom.

Abstract (english)

This paper provides a theoretical introduction to the problem of portfolio optimization, and mean-variance optimization as one of the standard approaches to solving this problem. Next, the field of deep learning is introduced, with an emphasis on the LSTM network. The implementation of a deep learning model, with LSTM architecture, which learns optimal asset allocations from return data with the aim of maximizing the Sharpe ratio, i.e. the ratio of return and portfolio risk, is presented. The models were tested on historical market data with a statistical analysis of the results in the framework of financial performance measures. Portfolios obtained by mean-variance optimization were used as a benchmark for evaluating performances of the models.

• Keywords: optimizacija portfelja; mean-variance optimizacija; duboko učenje; LSTM
• Keywords (english): portfolio optimization; mean-variance optimization; deep learning; LSTM
• Language: croatian
• URN:NBN: urn:nbn:hr:168:833243
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-10-12 20:33:02