Predicting VIX with Adaptive Machine Learning

71 Pages Posted: 16 Jun 2021 Last revised: 1 Mar 2023

See all articles by Yunfei Bai

Yunfei Bai

IEEE

Charlie X. Cai

University of Liverpool Management School

Date Written: March 1, 2023

Abstract

We used an automated machine learning framework to investigate economic factors that predict the CBOE implied volatility index (VIX), analyzing a comprehensive list of 278 variables for the first time. Our study tested multiple classification models, including Naïve Bayes, Logistic Regression, Decision Tree, Random Forest, Adaptive Boosting, Multi-Layer Perceptron, and an Ensemble model that combined all methods. Based on the validation stage and an 11-year out-of-sample period, we found Adaptive Boosting to be the most effective classification model. We demonstrated the economic importance of this out-of-sample predictability by simulating a long-short strategy. We then focus on understanding the sources of predictability and the limitations of the method. Our tests revealed that both nonlinear methods and comprehensive economic variables were significant predictors of VIX, with the weekly US jobless report and some S&P 500 members’ technical indicators emerging as the most important sources of predictability. While VIX spikes remained unpredictable, our algorithms could adapt quickly by extracting new information from the data to recover from losses in trading strategy tests. Our new evidence contributes to both machine learning applications in finance and practical volatility forecasting, offering alternative research designs and insights for traders and investors.

Video abstract: https://youtu.be/L_FQ6dFuYr0

Keywords: Machine Learning, AutoML, Explainable AI, VIX, Predictability, Forecasting, Quantitative Trading, Big Data, S&P 500, Futures, US markets

JEL Classification: G0, G17, C52, C55, C58

Suggested Citation

Bai, Yunfei and Cai, Charlie Xiaowu, Predicting VIX with Adaptive Machine Learning (March 1, 2023). Available at SSRN: https://ssrn.com/abstract=3866415 or http://dx.doi.org/10.2139/ssrn.3866415

Yunfei Bai

IEEE ( email )

Charlie Xiaowu Cai (Contact Author)

University of Liverpool Management School ( email )

University of Liverpool
Liverpool, L69 7ZA
United Kingdom

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
2,379
Abstract Views
6,467
Rank
11,428
PlumX Metrics