An artificial market model for the forex market

Abstract As financial markets have transitioned toward electronic trading, there has been a corresponding increase in the number of algorithmic strategies and degree of transaction frequency. This move to high-frequency trading at the millisecond level, propelled by algorithmic strategies, has brought to the forefront short-term market reactions, like market impact, which were previously negligible in low-frequency trading scenarios. Such evolution necessitates a new framework for analyzing and developing algorithmic strategies in these rapidly evolving markets. Employing artificial markets stands out as a solution to this problem. This study aims to construct an artificial foreign exchange market referencing market microstructure theory, without relying on the assumption of information or technical traders. Furthermore, it endeavors to validate the model by replicating stylized facts, such as fat tails, which exhibit a higher degree of kurtosis in the return distribution than that predicted by normal distribution models. The validated artificial market model will be used to simulate market dynamics and algorithm strategies; its generated rates could also be applied to pricing and risk management for currency options and other foreign exchange derivatives. Moreover, this work explores the importance of order flow and the underlying factors of stylized facts within the artificial market model.