Do Finance Researchers Address Sample Size Issues? – A Bayesian Inquiry in the AI Era.

Abstract

Artificial intelligence is found to be applied in all forms and levels of business. The process results in a huge volume of data generated in businesses, especially in the field of finance such as financial markets and financial institutions. The scenario became intense due to enormous digitization, during the post the Covid phase. Empirical researchers use algorithm-based techniques to handle this big data to infer evidence. However, the inherent issue of analyzing large data that tend to move toward statistical significance cannot be ignored. This can effectively be addressed through Bayesian statistics which is driven by conditional probability. Since analyzing larger data sets using traditional statistical methods or algorithm-based methods are prone to the risk of small evidence resulting in statistical significance, it is crucial to incorporate Bayesian methods in any empirical study, especially in finance. However, there is not much information about the ways and means of using Bayesian statistics in financial research, especially in the time of AI. Hence, this study attempted to review the literature with this objective. The results found two broader applications of Bayesian statistics i.e., prediction in financial markets and credit risk models. The methods and contexts identified in the literature will be helpful for researchers to widen the Bayesian application in finance and address the sample size issue effectively.

Keywords: Bayesian statistics, Artificial Intelligence, Naïve Bayes, Sample size issue

JEL Classification: G10, C11