Proposals for Master's theses
Please note our general information on theses. In addition, please note that most theses supervised by the institute have an empirical part. We therefore recommend that students interested in writing a Master thesis attend the modeling and research courses offered by the finance institutes (Financial Modeling, Research in Finance).
Predictability of stock market returns - how robust is it?
Whether aggregate stock market returns are predictable is a long-standing question. After Goyal and Welch had published their critical work, other researchers have proposed several approaches that seemed to work quite well. Now, these approaches are also viewed skeptically. In your paper, you shall first summarize the literature and then conduct a simulation study similar to the one in the paper below. You will use the same data, but test the robustness of machine learning approaches (Contact: Prof. Löffler).
Literature to get started: Rytchkov, O., & Zhong, X. (2019). Information Aggregation and P-hacking. Management Science.
Stock selection with fundamentals
Recent research shows that stock returns can be predicted based on simple regressions with firm fundamentals as predictors. In a first step, you will implement this strategy for a market (or several markets). In the second step, you will implement variants of the approach. Here you can consider several variants including machine learning approaches (Kontakt: Prof. Löffler).
Literature to get started: Bartram, S. M., & Grinblatt, M. (2018). Agnostic fundamental analysis works. Journal of Financial Economics, 128(1), 125-147. Not available online at journal yet but a pdf is here.
Managing the risk of quantitative equity portfolios
Investing based on characteristics such as value or momentum is very popular. In this thesis, you shall examine for a large number of possible strategies whether their risk-return profile can be enhanced through a risk management overlay. The idea is to predict future risk or low performance, and manage the exposure to the strategy according to the predictions. (Contact: Prof. Löffler).
Literature to get started: Barroso, P., & Santa-Clara, P. (2015). Momentum has its moments. Journal of Financial Economics, 116(1), 111-120.
Using image recognition for trading
Machine learning has made tremendous progress in image recognition tasks. In this thesis, you will explore whether standard image recognition approaches can be used to derive trading signals. You will explore different ways of converting financial data into images, and examine different assets and predictive information. Note that the literature on the application to trading is in its infancy, but it's something that experts from the field talk about (Contact: Prof. Löffler).
Literature to get started: Sim, H. S., Kim, H. I., & Ahn, J. J. (2019). Is deep learning for image recognition applicable to stock market prediction?. Complexity, 2019.
Opening the black box: credit scoring and explainable AI
In credit scoring, the scoring model is often a black box to the loan applicants whose creditworthiness is scored. This can raise a lot of concerns regarding transparency and fairness. In this thesis, you shall first review the literature on explainable or interpretable artifical intelligence (AI). In the major part of your thesis, you shall use real or simulated data to estimate credit scoring models and then explore how well explainable AI techniques would work in opening the black box.
Literature to get started:
Peters, C. (2019). Machine Learning Interpretability Techniques in Credit Risk Modeling.