Carolina Feher da Silva, Gaia Lombardi, Micah Goldsmith Edelson, Todd Anthony Hare, Rethinking model-based and model-free influences on mental effort and striatal prediction errors, Nature Human Behaviour, 2023. (Journal Article)
 
A standard assumption in neuroscience is that low-effort model-free learning is automatic and continuously used, whereas more complex model-based strategies are only used when the rewards they generate are worth the additional effort. We present evidence refuting this assumption. First, we demonstrate flaws in previous reports of combined model-free and model-based reward prediction errors in the ventral striatum that probably led to spurious results. More appropriate analyses yield no evidence of model-free prediction errors in this region. Second, we find that task instructions generating more correct model-based behaviour reduce rather than increase mental effort. This is inconsistent with cost–benefit arbitration between model-based and model-free strategies. Together, our data indicate that model-free learning may not be automatic. Instead, humans can reduce mental effort by using a model-based strategy alone rather than arbitrating between multiple strategies. Our results call for re-evaluation of the assumptions in influential theories of learning and decision-making. |
|
Julian V. Fischer, Currency Hedging Strategies for Bond Portfolios, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)

This thesis analyses the impact of currency hedging on international bond portfolios from the point of view of a Swiss investor. Four of the most important regions including their currency were included in this analysis, namely the US Dollar, Euro, Pound Sterling, and Japanese Yen. The focus of the thesis is on the mitigation of risk stemming from the currency exposure and the comparison of different hedging approaches. The leading question of this thesis is whether a Cash-Flow-Matched hedging approach is superior to the benchmark used by most index providers or to no hedging at all. All hedging approaches use currency forward contracts as hedging instruments. The results show that the benchmark is the best hedging method among these three. Superiority is not only shown in terms of the mean return but also in terms of higher moments. The monthly structure of hedging and reporting forms the base of the success.
The results are robust across all regions and bond-maturity buckets, showing that currency hedging is absolutely crucial in the setting of international bond portfolios. Additionally, more active hedging strategies were carefully introduced and compared to the three main approaches. An effort was undertaken to stay as close as possible to the philosophy of risk management. Overall, the results are less robust, pointing towards a slightly speculative nature of the introduced hedging strategies. Nonetheless, promising results were reported with rather simple quantitative trading rules, especially for the US region. A simple momentum approach is superior to all other approaches in terms of mean return.
The development of the currency exchange rates including the recent turbulence underlines the importance of currency hedging. Throughout the considered period, the Swiss Franc was very strong compared to the currencies of the other regions, generally playing in the favour of hedging. The analysis is strongly affected by the point of view as well, which is tailored to an institutional Swiss investor such as a Swiss pension fund. |
|
Prasun Saurabh, "DevOps pipeline for vision-based security attacks for Cyber-Physical Systems (CPS)", University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)
 
Unmanned Aerial Vehicles (UAVs), commonly known as drones, have revolutionized various industries, including agriculture, photography, delivery, and security. The UAV's ability to fly autonomously and perform various missions with ease is largely attributed to the advancement in vision algorithms. However, as these UAVs become more prevalent in civilian airspace, their reliability and security become crucial concerns.
One of the key components of a UAV is its onboard stereo camera, which enables the UAV to navigate through its environment. However, stereo cameras are vulnerable to vision-based security attacks, which can cause the UAV to crash or malfunction. The safety and reliability of UAVs heavily depend on the performance of their vision-based navigation systems. To ensure that these systems are robust and secure, it is essential to evaluate their resilience to different types of attacks and identify potential vulnerabilities. In order to address this issue, a platform was developed that can inject vision-based adversarial attacks into the UAV system to determine its vulnerability. This platform, called AerialShield, is an extension of Aerialist and is capable of carrying out different kinds of vision-based adversarial attacks on a UAV platform. AerialShield generates several adversarial test cases by mutating important parameters to attack the system. Through experiments, it was found that the PX4 Avoidance system, which is used for obstacle avoidance, is prone to adversarial attacks.
The experiments conducted by AerialShield showed that the PX4 Avoidance system is very sensitive to even a little noise in the stereo camera in a real-world-like simulated environment, leading to crashes. Moreover, the UAV was found to be less resilient to noise at a lower altitude as compared to a higher altitude. This highlights the importance of testing UAVs in various environments and altitudes to ensure their reliability and security. Our experiments have shown that several factors, such as UAV altitude, environmental complexity, and the level of noise injected into the camera, can significantly impact the system's performance.
While UAVs offer numerous benefits, their reliability and security are critical for their safe integration into civilian airspace. AerialShield's ability to carry out vision-based adversarial attacks on UAVs provides valuable insight into their vulnerability, allowing for improvements to be made to ensure their safe and secure operation. |
|
Anton Crazzolara, A Recommender System for Reviewable Code Changes, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)
 
Modern Code Review is an essential step of software development processes in industrial settings and open-source projects. It is usually supported by various tools to help reviewers during the process. Nonetheless, a significant part of the review time is still spent on understanding submitted changes. The challenge of understanding code changes could be improved by new tools designed for change authors to help them create more reviewable changes.
In this study, I collected information on different aspects relevant to the design of such tools, including their responsibilities and the associated implementations. I present Cres, a tool designed for identifying oversized commits and helping developers divide them into smaller commits. Cres was implemented following two different approaches, resulting in a web application and a pair of Git hooks. Both approaches were evaluated in interviews with expert developers to provide ideas and advice for the design of future tools. |
|
Emmanuel Mamatzakis, Steven Ongena, Pankaj C. Patel, Mike Tsionas, A Bayesian policy learning model of COVID-19 non-pharmaceutical interventions, Applied Economics, Vol. forthcoming, 2023. (Journal Article)

This article examines the impact of non-pharmaceutical interventions on the initial exponential growth of the infected population and the final exponential decay of the infected population. We employ a Bayesian dynamic model to test whether there is learning, a random walk pattern, or another type of learning with evolving epidemiological data over time across 168 countries and 41,706 country-date observations. Although we show that Bayesian learning is not taking place, most policy measures appear to assert some effect. In particular, we show that economic policy variables are of importance for the main epidemiological parameters derived from the policy learning model. In an empirical second-stage application, we further investigate the underlying dynamics between the epidemiological parameters and household debt repayments, a key economic variable, in the UK. Results show no Bayesian learning, although a higher transmission rate would increase household debt repayments, while the recovery rate would have a negative impact. Therefore, suboptimal learning is taking place. |
|
Baris Özakar, Time-Aware Centralities and Embeddings of Nodes for Influence Prediction in Evolving Socio-Financial Networks, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)
 
Financial markets are complex and constantly evolving systems, where investors make decisions based on market conditions, company performance, and global economic trends. However, recent studies suggest that peer effects can also play a significant role in shaping investment decisions. Peer effects refer to the influence that one's peers have on their decision-making, and in the context of financial decision-making, can cause investors to follow trends in herding behavior. This influence process can result in cascading behavior, where the actions of a few investors can trigger a chain reaction of buying or selling, leading to significant price movements. The impact of peer effects has been amplified by social networks that have revolutionized the way we communicate and share information.
In this thesis, we investigate the role of peer effects in financial markets and their impact on cascading behavior. Using a real-life evolving socio-financial network, we aim to quantify the extent to which individual investors influence the generation of cascading behavior, with a particular focus on the spatio-temporal features of individual investors within the network. We formulate a prediction task that forecasts the influence of individual users by utilizing various centrality measures and time-aware node embeddings. We evaluate the effectiveness of these centrality measures and time-aware node embeddings in predicting the influence of users in generating cascades of trades through the network. Our study contributes to a better understanding of the spatio-temporal factors that facilitate cascading behavior in financial markets, highlighting the need to understand their impact in various contexts, including real-life socio-financial networks.
|
|
Stefan Pohl, Three Essays on Empirical Financial Intermediation, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Dissertation)

|
|
Thiago Miguel Ottavio Semadeni, Der Mosambikkredit: Die Verschuldung eines ganzen Landes aufgrund der Credit Suisse, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

Die Credit Suisse steckt ein weiteres Mal im Schlamassel, doch sie nimmt es verhältnismäßig gelassen. Diese Arbeit gibt Aufschluss über den Ablauf der 2.07 Milliarden US-Dollar Kredite von der Credit Suisse London und der russischen Bank VTB London an halbstaatliche mosambikanische Firmen und zeigt die prekären Folgen für das Land auf. Weiter werden auf verhaltensethischer Ebene die Umstände und persönliche Charaktereigenschaften von den Drahtziehern dieses Skandals untersucht. Den Verantwortlichen fehlt es schlussendlich an ethischem Gedankengut und Empathie. Für die Zukunft muss die Credit Suisse ihre Unternehmenskultur verbessern und ändern, denn diese trägt ebenfalls zur Entstehung einen solchen Skandals bei. |
|
Amos Calamida, RadEval: A radiology-aware model-based evaluation metric for report generation, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)
 
In our work, we propose a novel automated radiology-specific evaluation metric that can be used for evaluating the performance of machine-generated radiology reports. We utilize the existing successful COMET metric architecture, which we adapt and optimize for use in the radiology domain. Using this architecture, we train and publish four medically-oriented model checkpoints using various combinations of encoders and corpora of radiology reports. One of the model checkpoints is trained using RadGraph, a radiology knowledge graph, and the thereof-derived RadGraph F1 and RadCliQ scores are integrated into our contributed parallel corpora to enhance their quality. Our results show that the developed metric exhibits a moderate to high correlation with established metrics such as BERTscore, BLEU, and S_emb score, indicating its potential effectiveness as a radiology-specific evaluation metric. |
|
Melvin Samson Steiger, Sentence-like Segmentation of Swiss German Audio Transcripts for Dependency Parsing, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)
 
Dependency parsers tend to struggle with parsing transcribed spoken language as they are trained on properly structured, written text.
Spoken language lacks the structure of properly written text and exhibits typical phenomena like disfluency, repetition, and truncation of words and sentences. This research examines the problem of parsing spoken language for Swiss German audio transcripts from ArchiMob corpus.
Swiss German, an umbrella term for the German (Alemannic) dialects spoken in Switzerland, lacks orthographic and grammatical standardization, shows a high degree of variation among the various dialects and differs substantially from Standard German. The lack of standardization is due to the situation of diglossia in Switzerland. As Swiss German is mainly an oral language or restricted to informal writing, many resources lack structure and exhibit a high variability in terms of morphology, spelling and vocabulary. The combination of variation in Swiss German, its lack of standardization and the unstructuredness of spoken language render parsing transcribed Swiss German challenging. Accordingly, pre-trained (German) dependency parsers struggle with Swiss German audio transcripts and little data is available to train them.
This research tackles the problem of parsing spoken language by re-segmenting Swiss German audio transcripts into sentence-like units (SLUs) and examines the impact of re-segmentation on dependency parser performance. Therefore, our experiment setup includes two evaluation steps, one for re-segmentation and one for dependency parsing. We frame the re-segmentation as a binary classification task aiming to predict tokens marking an SLU-boundary. For this purpose, we fine-tune a pre-trained German Bert model to predict such boundaries. These predicted SLU-boundaries are used to re-shape the input for the dependency parser. We show that the re-segmentation into SLUs leads to an improvement of the Labeled Attachment Score (LAS) over a baseline. Moreover, we demonstrate that the performance in the SLU-boundary classification task correlates with the parser performance. To engage in such a supervised learning setting, a test set composed out of roughly 200 SLUs was manually created and annotated with dependency labels for the two folded evaluation. With our work, we contribute to processing spoken Swiss German by showing a way of inducing more structure. |
|
Matej Gurica, Sinking in masses of online reviews: An analysis of the effects of various levels of AI support on the performance of response authors, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)
 
Online reviews have become a ubiquitous source of information for consumers, and response authors play a critical role in addressing customer feedback. However, the vast amount and complexity of online reviews pose a significant challenge for those who respond to them. By implementing artificial intelligence (AI), the possibility of enhancing
both the efficiency and quality of online review responses arises. In this thesis, it is examined how the use of various degrees of AI assistance affects the performance of response authors across a variety of efficiency and quality metrics.
In order to address this question, a survey was conducted to gather data on the perceived quality of review responses composed by novice and professional authors in four distinct settings: without any AI support (B setting), with partial AI support (I setting), with AI-powered response generator support (G setting), and with fully automated AI-generated responses (G-AI setting). In addition, data on the efficiency of the response authors was recorded during writing sessions in each setting.
The results show that the use of AI significantly improves the efficiency of response authors. In particular, the G setting greatly reduced the writing time for both professional and novice authors. Furthermore, the use of GPT-3, an advanced AI language model, resulted in significantly higher quality responses than those which a competing AI system and most of the other work configurations were able to produce.
Based on these findings, a work configuration is proposed which combines the strengths of AI systems with human authors to optimize the online review response process. This proposed configuration aims to maximize efficiency and response quality while minimizing the workload on human authors.
In conclusion, this thesis provides valuable insights into the potential of AI support to enhance the performance of response authors in the context of online reviews. The proposed optimal work configuration provides a practical solution for businesses and individuals looking to optimize their response authoring process. |
|
Tianshuai Lu, Reducing Gender Bias in Neural Machine Translation with FUDGE, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)
 
Gender bias appears in many neural machine translation models and commercial translation software. The problem is well known and efforts to reduce such discriminatory tendencies are underway. But gender bias is still not fully solved. This work utilizes a controlled text generation method, Future Discriminators for Generation (FUDGE), to reduce the so-called Speaking As gender bias emerging when translating from English to a language that openly marks the gender of the speaker. The model is evaluated with BLEU and MuST-SHE, a novel gender translation evaluation method. The results demonstrate improvements in the translation accuracy of the feminine terms. |
|
Nikola Maric, Analyse von verschiedenen Stablecoins, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

Die vorliegende Arbeit betrachtet acht Stablecoins, die aufgrund ihrer Eigenschaften aus-gewählt wurden, um diese auf Stabilität und Korrelationen zu untersuchen. Die Stab-lecoins werden durch bestehende Studien und anhand einer Stabilitätsanalyse betrachtet. Das Ziel der Arbeit ist es, Schluss-
folgerungen zu Stabilität, Korrelation und Unterschie-den der ausgewählten Stablecoins und ihrer Kategorisierung zu erhalten. Die Ergebnisse zeigen, dass die Fiat-besicherten Stablecoins die beste Performance bezüglich der Stabi-lität leisten. Jedoch ist auch ersichtlich, dass diese mit dem kleinsten Grad an Innovation folgen. Bei den algorithmischen Stablecoins ist es vice versa: Diese zeigen die unsicherste Stabilität mit der höchsten Volatilität, dafür den stärksten Innovationsgrad. |
|
Robin Wegmüller, Systematic exposure assessment using NLP and textual reporting data, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)
 
Financial research recently started using the latest breakthroughs in natural language processing.
This thesis develops an innovative framework for systematic exposure assessment: it measures the
notion of exposure by leveraging language models and gathering a priori expert expressions to quantify
the use of corpus- and domain-specific vocabulary. Focusing on risks and emerging technologies
in earnings calls, we apply our method to reveal insightful market behaviors. A case study then
shows that exposures from earnings conferences reflect the content from corresponding 10-Ks. Since
our approach solves major issues of other NLP techniques, we strongly encourage other researchers
to explore further applications.
|
|
Bekir Cagman, Financial institutes ESG profile: A textual analysis, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)

|
|
Sébastien Zurbriggen, Evaluating the Predictive Power of Dictionary News Sentiment and Commercially Provided Sentiment on Equity Volatility, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Master's Thesis)

|
|
Jorge Dörig, Momentum-Effekt unter Berücksichtigung der Marktkapitalisierung im amerikanischen Aktienmarkt, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

In dieser Arbeit wird die Markteffizienz mithilfe einer Momentum-Strategie anhand verschiedener Marktkapitalisierungs-Indizes getestet. Dabei basiert die Strategie auf dem Aspekt, dass Gewinneraktien Gewinner bleiben und Verliereraktien Verlierer bleiben. Durch die Anwendung dieser Strategie anhand eines Small-, Mid- und Large-Cap-Index soll zudem ein möglicher Unternehmensgrösseneffekt erkannt werden. Die empirische Studie der Arbeit konnte anhand des amerikanischen Aktienmarktes für den Zeitraum zwischen 2002 bis 2022 zeigen, dass es durch die Implementierung von auf technischer Analyse basierenden Strategien möglich ist, Überrenditen zu erzielen. Darüber hinaus konnte mit der Differenzierung von verschiedenen Marktkapitalisierungs-Indizes keine Evidenz für den Unternehmensgrösseneffekt bei Small-Cap-Unternehmen aufgebracht werden. |
|
Andrea Alessandro Mancuso, Will Hedge Funds be around in the future? An analysis and prediction of the presence of Hedge Funds in the financial market, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

This thesis analyses the current situation of the hedge fund industry to predict its survival in
the coming years. The focus is to understand the problems that hedge funds face today and
whether there are positive aspects in connection to a hedge fund investment.
The thesis starts with a historical overview of the industry and continues by presenting its
clientele, its strategies, and the fee structure used by hedge funds. It continues by comparing
hedge funds with passive investment alternatives, which are extremely popular today. The
comparison is qualitative and quantitative, analysing the last 20 years using data from
Bloomberg. The last chapter dives into several topics that influence hedge funds and their
relationship with the public.
The results suggest that hedge funds can offer their clients value that goes further than just
returns, such as great diversification effects and the ability to invest in complicated financial
structures and products, proving that there are multiple reasons why an investment in them can
be a good idea. This, combined with many external reports that have shown a growing trend
for hedge funds in the last couple of years, proves that they will continue to be an important
part of the market in the future. The thesis shows various points that should be further
developed by hedge funds, such as the fee structure and the relationship with the public |
|
Florian Schwaller, Neo-Banking – das Modell der Zukunft für den Schweizer Bankkunden, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

Die Studie widmet sich der Analyse demographischer Faktoren und ihrer Auswirkungen auf die Aktivität von Schweizer Neo-Bankkunden. Dabei erfolgt die Messung der Aktivität anhand von Transaktionsdaten, welche sowohl Kartenzahlungen als auch Banküberweisungen umfassen. Die Resultate aus einem Datensatz von über 50'000 Kunden einer Schweizer Neo-Bank erlauben unter Einbezug von statistischen Methoden die Konklusionen, dass der Zusammenhang zwischen dem Alter und der Nutzung von Karten- und Bankzahlungen negativ ist. In der Tendenz findet die Studie zudem eine höhere Aktivität auf Karten- und Bankzahlungen bei männlichen Kunden. Daraus lässt sich schliessen, dass demographische Merkmale einen bedeutenden Einfluss haben können, wie aktiv ein Kunde die Banking-Funktionen einer Neo-Bank nutzt. Demographische Faktoren sollten demnach für die Identifikation geeigneter Zielgruppen und die Ausarbeitung von Marketing-Strategien berücksichtigt werden. |
|
Granit Latifi, Evaluation der Predictive Performance verschiedener Value-at-Risk Modelle mittels Backtesting am Beispiel des Swiss Performance Index, University of Zurich, Faculty of Business, Economics and Informatics, 2023. (Bachelor's Thesis)

In dieser Arbeit wird die Vorhersagekraft von vier verschiedenen Modellen zur Ermittlung
des Value-at-Risk (VaR) empirisch untersucht. Der VaR stellt dabei eine Verlust-Vorhersage
für ein Portfolio dar, welche mit einer bestimmten Wahrscheinlichkeit nicht unterschritten
wird. Die Vorhersagen werden rollierend für 1500 Handelstage berechnet, wobei diese
die Periode vom 16.01.2017 bis zum 30.12.2022 umfassen. Das bedeutet, dass die Vorhersage
für den morgigen Zeitpunkt auf den am Vortag verfügbaren Informationen beruht. Das
Aktienportfolio besteht aus einer Long-Position im Swiss Performance Index (SPI). Die
Schätzung der Modellparameter, die Berechnung der Vorhersagen und das Backtesting erfolgen
mit dem Statistikprogramm R. Zur Beurteilung der VaR-Vorhersagen werden diese
mit den entsprechenden Renditen verglichen und mit vier Backtesting-Methoden evaluiert.
Aus der Analyse geht hervor, dass das GJR-GARCH(1,1)-Modell mit schiefer t-Verteilung
der Renditen die besten Vorhersagen liefert. Dieses Modell berücksichtigt auch die meisten
empirisch beobachtbaren Eigenschaften der Renditen. Das historische Simulationsmodell
wird in der Praxis zwar am häufigsten verwendet, führt aber aufgrund der trägen Anpassungsfähigkeit
der VaR-Vorhersagen zu ungenügenden Ergebnissen im Backtesting. |
|