A complex network approach to the bargaining problem

U ovom radu predstavljen je model kompleksne mreže konkurentnih agenata. Statistička i topološka svojstva modela su ispitana i uspored¯ena s poznatim modelima kompleksnih mreža. Dani model primjenjen je na problem pogodbe. U tu svrhu je razvijen model pogodbe zasnovan na konkurentnim agentima. Model je implementiran i simuliran kako bi se analizirala globalna dinamika. Rezultati pokazuju da logaritamski prinosi iz modela mogu imati teške repove, u ovisnosti o parametrima modela. Nadalje, pokazano je da broj agenata i grupiranje konkurencije u mrežama utječu na racionalnost simuliranog tržišta. Prinosi su uspoređeni s onima sa Zagrebačke burze i rezultati ukazuju na to da predloženi model odražava specifične globalne fenomene sa tržišta.In this thesis a complex network of competing agents is presented. The statistical and topological properties of the network are investigated and compared to the common complex network models. The network model is applied to the bargaining problem. For this purpose, a bargaining model relying on networks of competing agents is proposed. The model is implemented and simulated in order to test the global dynamics. The results show that the logarithmic day returns from the model may exhibit heavy tails, depending on the model parameters. Furthermore, it was shown that the number of agents and competition clustering in the networks affect the rationality of the simulated market. The returns were compared to those from the Zagreb Stock Exchange and the results show that the proposed model reflects specific global phenomena observed in the market