University of Twente Student Theses


Machine Learning Applications in Financial Advisory

Vulpoiu, Ruxandra (2018) Machine Learning Applications in Financial Advisory.

This is the latest version of this item.

[img] PDF
Abstract:Financial advisory is complex task, prone to subjectivity and mistakes. Extensive knowledge and experience is required in staying ahead of the market and even a single mistake can wipe out carefully attained profits. As it is a data intensive task with no specific methodology we look into machine learning as a promising approach for creating a decision support tool for financial advisers. We limit the study to assessment of future performance of stocks and creation of a diversified portfolio of stocks. More specifically, we use basic regression and ensemble classification algorithms to predict analyst rating estimated as an average value from multiple advisory houses. In a second part of the study we use clustering to create a mix of dissimilar stocks. Our research shows that aggregating stocks per economical activity, using complex models and extensive feature engineering improves results in analyst rating prediction when using variables from technical financial analysis. Moreover, the clustering technique is successful in implementing the diversification constraints and several portfolios created with this method have excellent financial performance in the first year of holding. Nevertheless, our study proves that integration of machine learning techniques with constraint based recommender systems is possible and a promising approach.
Item Type:Essay (Master)
Ortec Finance, Rotterdam, The Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science MSc (60300)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page