Postgraduate Research

• Algorithmic trading and financial forecasting

We use state-of-the-art machine learning (ML) algorithms to develop financial forecasting tools for different problems, such as volatility forecasting and future cash flow growth. We also use ML algorithms to create novel and competitive trading strategies. The derived trading strategies can use information not only from financial markets, but also from news outlets and social media, by taking advantage of sentiment analysis techniques.

• Directional changes

This novel approach approaches financial data from an event-based perspective, rather than the traditional way of sampling prices at fixed intervals (e.g. daily closing prices). The advantage of this approach is that it can capture significant points in price movements that the traditional physical time methods cannot, and as a result derive novel and more profitable trading strategies.

• High frequency finance

High frequency data is the real-time record of all trading activities and their associated characteristics observed in an electronic exchange system. These “tick-by-tick” data sets provide deeper insights into the price formation process at the micro level and allow us to develop trading strategies that take advantage of the “richness” of the information that tick data provides. The use of machine learning algorithms is challenging in this domain due to the large volume of data that causes computational times to explode. Research in this area can look into how we can effectively utilise ML algorithms without excessive computational costs.

• Weather derivatives

These are financial instruments used as part of a risk management strategy to reduce risk associated with adverse or unexpected weather conditions. We create novel algorithms to predict the value of the underlying asset (e.g. temperature, rainfall) and we then use these predictions to develop pricing models for weather derivatives contracts.

Our work in Algorithmic Game Theory (AGT) focuses on the study of decentralised systems in which (artificially intelligent) self-interested agents interact with each other aiming to optimise their personal objectives.

The most important questions related to such systems are whether the dynamics induced by the interaction between the agents converges to a stable outcome, known as equilibrium, how quickly such an equilibrium is reached and how efficient it is in terms of a social objective function.

Computational social choice (COMSOC) deals with computational aspects of problems in which a set of agents have preferences over different outcomes and the objective is to decide an outcome that is acceptable by the agents and satisfies possible constraints.

For example, one such important problem is that of allocating set of resources to the agents so that each of them considers the allocation to be fair.

A. Deligkas, J. Fearnley, A. Hollender and T. Melissourgos. Tight Inapproximability for Graphical Games. AAAI 2023

P. Kanellopoulos, M. Kyropoulou, H. Zhou. Debt Transfers in Financial Networks: Complexity and Equilibria. AAMAS 2023

M. Kampouridis, P. Kanellopoulos, M. Kyropoulou, T. Melissourgos, A.A. Voudouris. Multi-agent systems for computational economics and finance. AI Commun. (2022)

P. Kanellopoulos, M. Kyropoulou, H. Zhou. Forgiving Debt in Financial Network Games. IJCAI 2022

A. Deligkas, J. Fearnley, A. Hollender and T. Melissourgos. Pure-Circuit: Strong Inapproximability for PPAD. Focus 2022

A. Deligkas, J. Fearnley, A. Hollender and T. Melissourgos.  Constant Inapproximability for PPA. STOC 2022

P. Kanellopoulos, M. Kyropoulou, H. Zhou. Financial network games. ICAIF 2021

A. Filos-Ratsikas, E. Micha, and A. A. Voudouris. The distortion of distributed voting. Artificial Intelligence, vol. 286, article 103343, 2020. https://doi.org/10.1016/j.artint.2020.103343

A. Brabazon, M. Kampouridis, M. O’Neill. Applications of Genetic Programming to Finance and Economics: Past, Present, Future. Genetic Programing and Evolvable Machines (Invited Article), vol. 21, Springer, pp. 33-53, 2020. https://doi.org/10.1007/s10710-019-09359-z

B. de Keijzer, M. Kyropoulou and C. Ventre. Obviously Strategyproof Single-Minded Combinatorial Auctions. Proceedings of the 47th International Colloquium on Automata, Languages and Programming (ICALP), 2020.

G. Amanatidis, G. Birmpas, A. Filos-Ratsikas, A. Hollender, and A. A. Voudouris. Maximum Nash welfare and other stories about EFX. Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI), 2020.

IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (IEEE CIFER)

Workshop on the Distortion and Information-Efficiency Tradeoffs (DIET)