A machine learning approach to volatility forecasting

TY - JOUR

T1 - A machine learning approach to volatility forecasting

AU - Christensen, Kim

AU - Siggaard, Mathias Voldum

AU - Veliyev, Bezirgen

PY - 2023

Y1 - 2023

N2 - We inspect how accurate machine learning (ML) is at forecasting realized variance of the Dow Jones Industrial Average index constituents. We compare several ML algorithms, including regularization, regression trees, and neural networks, to multiple Heterogeneous AutoRegressive (HAR) models. ML is implemented with minimal hyperparameter tuning. In spite of this, ML is competitive and beats the HAR lineage, even when the only predictors are the daily, weekly, and monthly lags of realized variance. The forecast gains are more pronounced at longer horizons. We attribute this to higher persistence in the ML models, which helps to approximate the long-memory of realized variance. ML also excels at locating incremental information about future volatility from additional predictors. Lastly, we propose a ML measure of variable importance based on accumulated local effects. This shows that while there is agreement about the most important predictors, there is disagreement on their ranking, helping to reconcile our results.

AB - We inspect how accurate machine learning (ML) is at forecasting realized variance of the Dow Jones Industrial Average index constituents. We compare several ML algorithms, including regularization, regression trees, and neural networks, to multiple Heterogeneous AutoRegressive (HAR) models. ML is implemented with minimal hyperparameter tuning. In spite of this, ML is competitive and beats the HAR lineage, even when the only predictors are the daily, weekly, and monthly lags of realized variance. The forecast gains are more pronounced at longer horizons. We attribute this to higher persistence in the ML models, which helps to approximate the long-memory of realized variance. ML also excels at locating incremental information about future volatility from additional predictors. Lastly, we propose a ML measure of variable importance based on accumulated local effects. This shows that while there is agreement about the most important predictors, there is disagreement on their ranking, helping to reconcile our results.

KW - accumulated local effect

KW - heterogeneous auto-regression

KW - machine learning

KW - realized variance

KW - volatility forecasting

UR - https://www.scopus.com/pages/publications/85178127509

U2 - 10.1093/jjfinec/nbac020

DO - 10.1093/jjfinec/nbac020

M3 - Journal article

SN - 1479-8409

VL - 21

SP - 1680

EP - 1727

JO - Journal of Financial Econometrics

JF - Journal of Financial Econometrics

IS - 5

M1 - nbac020

ER -