Developing a Simulated Microcosm of a Contemporary Betting Exchange for the Discovery of Transferable and Profitable Algorithmic Betting Strategies

Abstract

Listen

Translate article

Since the dotcom boom of the early 2000s the betting industry has been transformed by the rise of betting exchanges, from a market dominated solely by traditional bookmakers, to one in which punters have a variety of wagering options. These exchanges bring together populations of bettors with varying and opposing views in the same way as traditional exchanges operating within the world's major financial markets. As a result the rise of algorithmic trading within the financial markets has been similarly observed within betting markets. The design and implementation of new financial trading algorithms is an arduous task due to the difficulty in obtaining high-quality data that can be used to evaluate a strategies' performance. As such Synthetic Data Generators (SGDs), such as Cliff's Bristol Stock Exchange (2012), have been developed to provide a test-bed in which researchers can develop trading algorithms and investigate market dynamics. This thesis therefore describes the design, implementation, and subsequent evaluation, of a novel asynchronous implementation of the Bristol Betting Exchange (BBE), the rationale and high-level design of which can be found in Cliff 2021. BBE provides an agent-based simulation model of a contemporary sports-betting exchange which will enable both researchers and industrial users to develop algorithmic betting agents and investigate previously unexplored betting market dynamics.