Essays on Computational Finance

This thesis explores the behaviour of different types of individuals (i.e., traders, experimental subjects) in financial markets under two different environments. It consists of three chapters/papers. Chapters 2 and 3 focus on an electronic market environment where traders submit orders through an electronic platform (i.e., electronic (limit) order book). Chapter 4 develops a portfolio choice problem – implemented in a laboratory environment - where subjects choose under risk, the allocation of wealth between two assets. Chapters 2 and 3 develop a dynamic continuous-time model of trade in a single financial asset under two different setups. In chapter 2 under the Single-Market scenario, the set of preferences of all market participants is aggregated on a limit order book related to a ‘lit’ market. Under the Multi-Markets scenario, trading takes place on two limit order markets that operate in parallel: an open (transparent) ‘lit’ market and a ‘dark’ (opaque) market. The aim of this chapter is to study the effects of the latter market on the ‘lit’ market’s quality measures, trading behaviour and welfare of traders. The results show that the presence of the ‘dark’ market harms the quality of the incumbent exchange (i.e., reduced liquidity, quoted spread increase, the effective spreads of all market participants increase). All market participants across the two scenarios make higher profits when they participate in the ‘lit’ market under the Multi-Markets scenario. Chapter 3 looks into the recent regulatory interests on ‘dark’ trading (e.g., European Commission, 2010; Securities and Exchange Commission, 2010). To address the policy question a competitive setting is considered - with two different groups of agents - where two limit order markets operate simultaneously under two different scenarios. This chapter looks into whether the different price mechanisms applied to the ‘dark’ market - midpoint dark pool (‘mid-market’ scenario) and Volume-Weighted Average Price (VWAP) dark pool (‘VWAP-market’ scenario) - affect traders’ welfare (and behaviour) and the ‘lit’ market’s quality. The results suggest that across the two scenarios the ‘VWAP-market’ scenario benefits the ‘lit’ market’s quality while only a certain group of agents (speculators) make less profits in this environment as compared to the alternative scenario. The aim of chapter 4 is to study - in a laboratory environment - subjects’ decision making under risk in a financial asset setting, with contingent claims fixed across frames. By testing the fundamental and untested hypothesis imposed on the majority of the asset demand tests (i.e., Arrow-Debreu contingent claim setup); the study focuses on the application of revealed preference to the model of choice under risk and under uncertainty. Applying Afriat’s Theorem (Afriat, 1967) we check if the experimental data is consistent with the maximization hypothesis. The results suggest that the power of the test is close to perfect; while the test’s predictive success is within the positive range.