Analysis and design of competing double auction marketplaces

The double auction, a highly efficient market mechanism, has been widely used by both traditional and online exchanges. However, with the globalisation of the economy, these marketplaces increasingly need to compete with each other to attract traders and charge suitable fees to make profits. In this situation, a double auction marketplace needs effective market rules (also called market policies) to govern the trading activity of its buyers and sellers and the ability to set fees appropriately in order to make profits and, at the same time, keep existing traders and attract new ones. To this end, in this thesis, we analyse competing double auction marketplaces, and use insights from this analysis to design an effective competing marketplace for an international market design competition, which is called CAT.
\n
\nIn more detail, the design of a competing double auction marketplace consists of determining market policies, which govern traders’ interactions in the marketplace, and a charging strategy, which determines the fees charged to traders. In this thesis, we mainly focus on the latter since this is a significant determinant of the traders’ choices of marketplaces and the marketplaces’ profits. Now, the effectiveness of a certain charging strategy depends on the traders’ behaviour, both in terms of how the fees affect their market selection, as well as their bidding behaviour. Thus, in order to set an appropriate charging strategy, we need to obtain a fundamental understanding of the traders’ market selection and bidding strategies. In the context of multiple competing marketplaces, the optimal choice for a trader in terms of selecting a marketplace and submitting bids not only depends on its own preferences (i.e. type, which is usually privately known), but also on the behaviour of other traders and marketplaces, and the optimal choice of a marketplace in terms of setting fees also depends on the behaviour of traders and other marketplaces. Therefore we need to analyse the equilibrium strategies for traders and marketplaces. In so doing, we consider several settings. In particular, we consider the settings where traders can
\nonly enter one marketplace at a time (single-home trading) and can enter multiple marketplaces at a time (multi-home trading). Furthermore, we consider the setting where the traded goods are independent, substitutes or complements. In the analysis, we show how these different trading environments and different good properties affect the strategies of traders.
\n
\nIn more detail still, we first analyse a single-home trading environment with a small number of discrete trader types, where traders are assumed to use a truth-telling bidding strategy, i.e. submit their types as their shouts. For this setting, we first analyse the equilibrium market selection strategies of traders for given market fees. We derive the equilibrium strategies analytically and furthermore use evolutionary game theory to investigate the dynamics of the traders’ strategies. Our results show that when the same type of fees are charged by two marketplaces, all the traders will converge to one marketplace. However, when different types of fees are allowed (registration fees and profit fees), competing marketplaces are more likely to co-exist in equilibrium. Moreover, we find an interesting phenomena that sometimes all the traders eventually migrate to the marketplace that charges higher fees. We then go on to analyse the equilibrium charging
\nstrategies of the marketplaces. Specifically, we present two approaches: a static and a dynamic analysis. The former is based on the assumption that marketplaces set their fees once at the beginning and so the charging strategies are not affected by the changes in the traders’ market selection strategies. In the latter analysis, we tackle this limitation by using a co-evolutionary approach where we analyse how competing marketplaces dynamically set fees while taking into account the dynamics of the traders’ market selection strategies. From this analysis, we find that two initially identical marketplaces eventually charge the minimal fee that guarantees positive market profits for them. We also find an initially disadvantaged marketplace with an adaptive charging strategy can beat an initially advantaged one with a fixed charging strategy.
\n
\nBuilding on this, we use fictitious play (a computational learning approach) to extend the above analysis by considering continuous trader types, different trading environments and different good properties. Moreover, we consider two more types of fees (transaction and transaction price percentage fees), and instead of assuming that traders adopt a truth-telling bidding strategy, we analyse both the equilibrium market selection and bidding strategies. In more detail, we first analyse traders’ equilibrium bidding strategies in a single marketplace and investigate how these strategies are affected by the different fees. In so doing, we find that registration fees cause a bigger range of traders not to choose the marketplace; profit fees cause traders to shade a lot; transaction price percentage fees cause sellers to shade relatively less than buyers. Then we analyse how different trading environments and different good properties can affect traders’ equilibrium market selection and bidding strategies. We then analyse the effects of different types of fees on obtaining market profits and keeping traders in a single marketplace environment. We find that the transaction price percentage fee is the most effective in making profits and keeping traders. Finally, we analyse how competing marketplaces set fees in equilibrium and show that the marketplace will charge high profit fees since traders can shade.
\n
\nFinally, in addition to analysing the charging strategies, we also experimentally analyse how different market policies affect the performance of competing marketplaces in different environments where traders adopt different bidding strategies. Then, using the insights from analysing the equilibrium charging strategies and the market policies, we design a competing marketplace, which we entered into the 2010 CAT competition. This agent performed well and was ranked first in the second day’s competition and second in the third day’s competition.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.