Use this URL to cite or link to this record in EThOS:
Title: Spontaneous segregation of adaptive agents in auctions
Author: Aloric, Aleksandra
ISNI:       0000 0004 6349 4554
Awarding Body: King's College London
Current Institution: King's College London (University of London)
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
When a population of agents divides into subgroups, and subsequently interactions happen between largely fixed subgroups, we regard the population as segregated. The segregated state might confer benefits, e.g. a buyer who has a strong preference for a particular seller has shorter exploration time when buying. As it might also add vulnerability to a system, understanding how segregation emerges is essential. To investigate whether the segregation can arise spontaneously, as a consequence of repeated interaction and co-adaptation among the agents, a stylised model of double auction markets and traders is developed and investigated. We show that in a system with two discrete-time double auctions and a large population of adaptive traders a collaborative segregated state emerges. When the typical scale of market returns become higher than some threshold, the preferred state of the system is segregated: both buyers and sellers are segmented into subgroups that are persistently loyal to one market over another. The segregated state is stabilised by some agents acting cooperatively to enable trade and provides higher rewards than its unsegregated counterpart both for individual traders and the population as a whole. Realising that the agent’s adaptation is the key promoter of the segregation, we investigate the robustness of our findings in continuous double auctions with sophisticated trading strategies – adaptive agents still prefer to segregate. Accordingly, to create informed regulations e.g. in large financial systems, we believe it is necessary to investigate benefits and risks that segregation brings and consequently how to promote or suppress it.
Supervisor: Sollich, Peter Kurt ; Kuehn, Reimer Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available