Machine Learning the Cross-Section of Intraday Option Returns (Machine Learning /Big Data)
- Contact:
- Project Group:
BSc and MSc
- Startdate:
Now
- Over the past decade, according to the Futures Industry Association (FIA) annual statistics, the global volume of options traded on exchanges surged from $9.42 billion contracts in 2013 to $137.3 billion contracts in 2023, representing an increase of over 1457%. Approximately 60% of these contracts are based on individual stocks and stock indices, highlighting the growing prominence of options in financial markets.
- Given the dynamic nature of this market, investors are increasingly focused on a critical question: how can option returns be predicted, and what specific characteristics influence such predictability? While prior research, such as the work by Bali, Beckmeyer, Moerke, & Weigert (2021), has concentrated on predicting monthly option returns, the high-frequency nature of options trading raises interest in predicting intraday returns. This project aims to extend this line of research by focusing on intraday option return prediction, drawing on insights from existing studies on intraday stock returns to compile a novel set of characteristics that impact intraday option returns.
- At the same time, machine learning models have garnered significant attention in financial forecasting, particularly in the derivatives market, due to their exceptional predictive power. Models such as decision trees and neural networks are particularly well-suited for capturing the nonlinear relationships and interactions among a vast array of variables, enabling them to model complex interactions in predicting intraday option returns. Furthermore, the real-time nature and high volume of intraday option return data favor machine learning models, which can process large datasets rapidly and generate forecasts over short time intervals. This capability is especially critical in intraday trading, where option prices fluctuate rapidly, and AI algorithms can produce highly accurate predictions within a very short time frame, allowing investors to swiftly adjust their trading strategies.
- This project seeks to advance the study of intraday option return prediction through machine learning and high-dimensional data processing. By providing new tools for investors and contributing to the academic understanding of the predictability of intraday option returns and their influential characteristics, it holds potential to offer significant insights into this evolving field.