Time-Varying Volatility Factors for Forecasting and Pairs Trading

Summary

This document describes a model-agnostic framework for forecasting asset volatility by extracting a compact set of time-varying factors from realized volatilities. The factors are intended to capture changing cross-asset co-movement while keeping computation manageable, addressing a limitation of static factor loadings and the expense of fully multivariate models. They can be added to statistical or AI-based forecasting methods to represent both market-wide and asset-specific dynamics.

The framework is evaluated on large-cap U.S. technology stocks and major cryptocurrencies at one-day and seven-day horizons. The paper reports improved predictive accuracy and economic value across linear and nonlinear models, and describes a pairs-trading strategy based on the forecasts as profitable with stronger risk-adjusted returns, especially in adverse conditions. The summary does not specify sample periods, benchmarks, costs, or validation procedures, so the reported improvements cannot be assessed independently from this description.

Key ideas

• Time-varying factors extracted from realized volatility can represent changing cross-asset co-movement.
• A compact factor set aims to reduce the computational burden of fully multivariate volatility models.
• The factors can augment both statistical and AI-based forecasting methods.
• The framework is evaluated on U.S. technology equities and major cryptocurrencies over short and medium horizons.
• The document reports benefits for volatility prediction and a forecast-based pairs-trading strategy, but gives few validation details.

Tags

# Time-Varying Factor-Augmented Models for Volatility Forecasting


# Time-Varying Factor-Augmented Models for Volatility Forecasting









Accurate volatility forecasts are vital in modern finance for risk management, portfolio allocation, and strategic decision-making. However, existing methods face key limitations. Fully multivariate models, while comprehensive, are computationally infeasible for realistic portfolios. Factor models, though efficient, primarily use static factor loadings, failing to capture evolving volatility co-movements when they are most critical. To address these limitations, we propose a novel, model-agnostic Factor-Augmented Volatility Forecast framework. Our approach employs a time-varying factor model to extract a compact set of dynamic, cross-sectional factors from realized volatilities with minimal computational cost. These factors are then integrated into both statistical and AI-based forecasting models, enabling a unified system that jointly models asset-specific dynamics and evolving market-wide co-movements. Our framework demonstrates strong performance across two prominent asset classes-large-cap U.S. technology equities and major cryptocurrencies-over both short-term (1-day) and medium-term (7-day) horizons. Using a suite of linear and non-linear AI-driven models, we consistently observe substantial improvements in predictive accuracy and economic value. Notably, a practical pairs-trading strategy built on our forecasts delivers superior risk-adjusted returns and profitability, particularly under adverse market conditions.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.