From 1st to 10th August 2024, Petnica Science Center (PSC) organized a specialized seminar in computational finance with a significant involvement of teachers and students from the Department of Mathematics and Informatics (DMI), Faculty of Sciences, University of Novi Sad, Serbia. The seminar was headed by Diana Šantavec and Aleksandra Simić. Diana is currently a computer science student at DMI, teaching associate at PSC and quantitative software developer at fintech company FIS, while Aleksandra is an economist specialized in asset liability management and market risk analysis also working in FIS. The seminar was intended for bachelor and master students interested in applied financial mathematics and informatics, aspiring to work as researchers in fintech, banking and insurance financial sectors. Leading, internationally recognized researchers and experts were involved in seminar activities, notably Miloš Božović (full professor at University of Belgrade, Faculty of Economy), Branko Urošević (full professor at Union University, School of Computing), Drago Inđić (managing director of Oxquant and industrial professor at UCL Institute of Finance and Technology, UK), Bojana Milošević (associate professor at University of Belgrade, Faculty of Mathematics), Jasna Atanasijević (associate professor at University of Novi Sad, DMI and director of economic advisory at EY for Serbia, Montenegro and Bosnia & Herzegovina) and Miloš Savić (associate professor at University of Novi Sad, DMI and member of PSC executive board).
The seminar was realized using a teaching model based on lectures providing in-depth overviews in the field, practical workshops devoted to specific topics and team work on research projects following the standard pedagogical principles of PCS – focused mentorship and learning through research. In total, 16 students participated in the seminar, approximately half of them studying computer science/information technologies and the second half studying economics. Such a diversified group of students created a specific working environment in which young, talented people with different backgrounds performed very intensive, interdisciplinary teaching and research activities together with their lecturers and mentors coming both from academia and industry.
After a general introduction to computational finance, lectures and workshops covered topics in fundamental mathematics for computational finance (numerical methods, probability, statistics, regression analysis), Python programming for computational finance and good software development practices, history of financial markets and crises, monetary systems and policies, law regulations for financial transactions and instruments, valuation of financial instruments and predictability of their prices, optimization of financial portfolios, evaluation of bonds, the Black–Scholes options pricing model, AI and machine learning techniques in computational finance, insurance industry business, digital currencies and tokens, fintech and hedge funds, and financial assets in practice.
Diana Šantavec, co-head of the seminar, briefly summarize main ideas of the seminar, her impressions of this year’s realization and goals for next editions:
“The Computational Finance Seminar was created with the idea of merging three domains under one big umbrella. Mathematics, finance, and programming were unified in a ten-day seminar where we successfully managed to combine theoretical concepts with practical experience. Sixteen student participants from various universities complemented our initial vision and helped us achieve it. The two-day project work with the support of assigned mentors is perhaps where the participants most felt and applied the knowledge they had already gained. For this reason, in the coming years, we would like to focus more on project work. Our first experience was very significant because it allowed us to understand how to improve certain aspects. For example, we now have a better insight into how we can create a stronger connection between programming and economics in the coming years. For us, this is a very important segment, and we will work on ensuring that it is realized as efficiently as possible so that the participants have strong support for their future achievements. Additionally, we would focus on expanding the material covered in the seminar, both by introducing new topics and by extending the time allocated for topics that we believe should be covered in more detail. Another idea is to open the seminar to foreign students and make it international. We would be very pleased to connect students with different knowledge and experiences, and also to establish collaboration with industry and academic experts who are accomplished in various markets. For us, this is just the beginning of something very beautiful and useful, and as organizers, we thoroughly enjoy the whole process. We believe that the enthusiasm with which we approach the seminar can only lead us to even more innovative ideas and to expanding the project in a way that we truly believe in.”
As a part of mentored exercises, one of the group of three students was given the responsibility of creating a tool for market analysis of the Serbian banking industry, and Jasna Atanasijević from DMI and EY served as their mentor. For all three team members, the full day (till late at night) guided exercise provided a chance for intensive learning and teamwork. Python code was used to scrape and parse the data, convert it into useful indications of market position and financial performance, and visualize the data using a graphical user interface.
Jasna also gave a lecture about money. A wide range of topics, including the definition and measurement of money, inflation, interest rates, exchange rates, monetary systems, the creation of money, and monetary policy, were discussed in the lecture. The students appeared highly interested in expanding their knowledge of economics outside the domain of finance. Their inquisitiveness about the subject led to a lengthy conversation that covered a wide range of matters, from the definitions of basic terms to more complicated ideas like how a negative interest rate can exist, what kinds of currency substitution can coexist (in an economy like the one in Serbia that has been strongly eurized), and is it possible to predict a foreign exchange rate. She summarizes her impression as follows:
“Teaching to a group that was so diverse in terms of years of prior study (from none to master level) and knowledge in finance and economics was a difficult challenge from the start. The participants’ desire to learn more about social science ideas in general and finance in particular, which enables the analysis of (big) data to support decisions and provide answers to a variety of intriguing problems, however, inspired me. The experience has strengthened my conviction that, while they are difficult to set up, programs that combine social science and programming on the one hand and mathematics and programming on the other are absolutely needed. The Petnica Science Center is a great venue for these seminars, while the long tradition of programs in applied mathematics and informatics at DMI has been proven as a valuable resource. Diana (co-head of the seminar), as well as some of the participants, have been our students.”
Miloš Savić from DMI gave a lecture on artificial intelligence and machine learning with applications in finance. After presenting the historical overview of AI and the dominant role of search algorithms, domain knowledge and big data at different stages of AI development, the lecture was focused on contemporary machine learning algorithms in unsupervised and supervised learning settings with emphasis on anomaly detection as one of the most important AI applications. Savić presented a case study on anomaly detection in tax datasets describing a hybrid machine learning method combining representational learning based on autoencoders and clustering enhanced with relevant domain knowledge. This case study illustrated the importance of combining high-recall and high-precision anomaly detection models in order to provide an accurate set of internally validated anomalies.
Reflecting on his teaching experience at the seminar, Miloš emphasizes the following:
“Computer programming is today an essential skill for researchers and experts in many fields. Various scientific disciplines (e.g., computational finance, computational physics, computational sociology, computational linguistics, bioinformatics, etc.) rely on computer science methods to analyze and investigate important and intriguing phenomena in natural and social sciences. Consequently, contemporary educational systems and institutions, both formal and informal, should cultivate interdisciplinary study approaches to real-world problems that have a significant impact on human well-being. I am glad that two educational institutions to which I am affiliated, DMI and PSC, recognized the importance of interdisciplinarity in which computer science is not simply taken as a set of black box tools. Especially when applying data mining, data analysis, AI and machine/deep learning methods in finance – students, researchers and industry practitioners should be familiar with internal concepts and algorithms of computer science methods in order to properly use and maximally benefit from existing programming libraries, frameworks and tools”.
About PSC. Petnica Science Center is an independent non-profit organization that has been working since 1982 as a unique national center dedicated to advanced additional extracurricular education and support for young people who show a particularly high level of interest in natural and social sciences and modern technologies. More information about PSC is available at its official web page www.petnica.rs.
