Artificial Neural Networks in Financial Modelling
Alex LaPlante, Managing Director, Research, Global Risk Institute
Alexey Rubtsov, Research Associate, Global Risk Institute
In this paper, we discuss Artificial Neural Networks (ANNs), one of the most common and complex Machine Learning tools. After a non-technical introduction to ANNs, we examine the risks associated with the application of this algorithm to financial data.
The last decade has witnessed tremendous growth in the adoption of Machine Learning (ML) tools in the financial services industry, with global investment in Artificial Intelligence start-ups rising from $282 million in 2011 to $2.4 billion in 2015. Concurrently, financial regulators and market participants have been raising concerns regarding the risks that such a widespread and rapid adoption may entail. In this paper we examine one of the most common tools of ML, called Artificial Neural Networks (ANNs), and discuss associated risks of applying these tools to financial modelling.
Results of successful applications of ANNs in finance have been widely reported in academic and nonacademic literature. In particular, they have been employed in credit-risk evaluation (Baesens et al., 2003); bank failure prediction (Tam and Kiang, 1992); assessment of the risk of management fraud in financial statements (Green and Choi, 1997); prediction of yield curve dynamics (Kondratyev, 2018), assets pricing (Gu et al., 2018), implied volatility movements (Cao et al., 2019), etc.