Big Data Science in Finance

Chapter 8Applications: Unsupervised Learning in Option Pricing and Stochastic Modeling

Introduction

This chapter presents two applications for unsupervised learning: optimization of option pricing and, separately, optimization of Markov Chains. Perhaps two of the most popular financial applications, both inferences of options pricing and various Markov Chains components, can be dramatically sped up and improved with unsupervised learning, as this chapter illustrates.

Application 1: Unsupervised Learning in Options Pricing

The options data comprise trillions of data points per day. The options on the U.S. stocks alone number in millions, each with a different strike price, expiration, and action. In addition to the U.S. securities, there are options on commodity futures, currencies, and fixed income. All of the options have the capacity to be traded and, thus, convey public information about someone's belief about the markets.

Despite the richness of the options data, traditionally, these data have been aggregated into a handful of indicators, like

• call-put implied volatility spread (Bali and Hovakimian 2009; Cremers and Weinbaum 2010; and Yan 2011).
• risk-neutral skewness (Xing, Zhang, and Zhao 2010; Rehman and Vilkov 2012; Conrad, Dittmar, and Ghysels 2013; Stilger, Kostakis, and Poon 2016; and Bali, Hu, and Murray 2016).
• option to stock volume ratio (Roll, Schwartz, and Subrahmanyam 2010; and Johnson and So 2012).
• volatility of implied volatility (Baltussen, Van Bekkum, ...