Multi-agent systems for computational economics and finance
What is it about?
Environments where different entities compete over limited resources are found in almost all levels of our societies. In these environments, the inherent agents' freedom to manipulate the system to their benefit, imposes great unpredictability on the interactions' outcome. Using theoretical tools, we study in depth the properties and quality of outcomes in these environments, and design competition systems that result to desirable outcomes (this usually is considered to be the maximiser of social happiness). On the practical side, we apply artificial intelligence techniques to develop new strategies striving to maximize the return in complex financial markets.
Why is it important?
This is a survey of recent developments in the intersection of economics, finance, theoretical computer science and artificial intelligence. We study various problems related to computation and quality of solution concepts in multi-agent environments. Such concepts include Nash equilibria, voting mechanisms, fair division of divisible and indivisible items, and optimal adaptation to financial competition. We present our extensive contribution to these fields both from a theoretical and an applied perspective. Our findings can be utilized in a multitude of applications spanning across a wide range of economic environments: auction design, online advertisement, design of supply chains, centralised/distributed security systems, voting settings, fair split of inheritance, design of optimal routing networks, portfolio management in financial markets, financial trading, and many more.