Paper: arXiv 2504.15985
Authors: Markus Bibinger, Jun Yu, Chen Zhang
Abstract
A multivariate fractional Brownian motion (mfBm) with component-wise Hurst exponents is used to model and forecast realized volatility. We investigate the interplay between correlation coefficients and Hurst exponents and propose a novel estimation method for all model parameters, establishing consistency and asymptotic normality of the estimators. Additionally, we develop a time-reversibility test, which is typically not rejected by real volatility data. When the data-generating process is a time-reversible mfBm, we derive optimal forecasting formulae and analyze their properties. A key insight is that an mfBm with different Hurst exponents and non-zero correlations can reduce forecasting errors compared to a one-dimensional model. Consistent with optimal forecasting theory, out-of-sample forecasts using the time-reversible mfBm show improvements over univariate fBm, particularly when the estimated Hurst exponents differ significantly. Empirical results demonstrate that mfBm-based forecasts outperform the (vector) HAR model.
Complexity vs Empirical Score
- Math Complexity: 9.5/10
- Empirical Rigor: 7.2/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper relies heavily on advanced stochastic calculus, Gaussian process theory, and estimation with asymptotic normality proofs, indicating high mathematical complexity. It provides out-of-sample forecasting comparisons, a time-reversibility test, and a reference to R code for implementation, demonstrating solid empirical implementation.
Research Flowchart
flowchart TD
A["Research Goal<br>Model & Forecast<br>Realized Volatility"] --> B{"Key Methodology<br>Multivariate fBm mfBm Model"}
B --> C["Data: Multivariate<br>Realized Volatilities"]
C --> D["Parameter Estimation<br>Consistent & Asymptotic Normal Estimators"]
D --> E["Time-Reversibility Test<br>Validates Model Assumption"]
E --> F["Optimal Forecasting<br>Derives Formulae for mfBm"]
F --> G{"Outcomes"}
G --> H["Theoretical Insight<br>MfBm with Varied Hurst Exponents<br>Reduces Forecast Error"]
G --> I["Empirical Result<br>Outperforms Univariate fBm<br>& Vector HAR Model"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a multivariate fractional Brownian motion model with component‑wise Hurst exponents and a novel moment‑based estimation method for forecasting realized volatility.
Source sentence
A multivariate fractional Brownian motion (mfBm) with component-wise Hurst exponents is used to model and forecast realized volatility. We
Assumes 5
- The data‑generating process is assumed to be a time‑reversible mfBm. model
Source sentence
When the data-generating process is a time-reversible mfBm, we derive optimal forecasting formulae and analyze their properties.
- Mild regularity conditions are required for consistency of the estimators. statistical
Source sentence
Under mild regularity conditions, we establish the consistency of these estimators.
- Component‑wise Hurst exponents are assumed to be below 0.5 in the empirical application. model
Source sentence
The Hurst exponents are all below 0.5, the correlations of increments across components are approximately 0.3, and the asymmetry parameters are close to zero.
- Correlation parameters must satisfy positive semi‑definiteness of the covariance kernel. model
Source sentence
To ensure the covariance kernel of mfBm remains positive semi-definite, certain parameter restrictions must hold.
Show 1 more
- Asymmetry parameters are assumed to be negligible (close to zero) for volatility series. data
Source sentence
For realized volatilities, estimated asymmetry parameters are typically close to zero.
Data
- Daily realized volatilities of Dow Jones 30 constituents market-prices proprietary
Source sentence
Our analysis focuses on daily realized volatilities of the Dow Jones 30 (DJ30) constituents from March 28, 2013, to August 21, 2021—a span of eight years, using data provided by Risk Lab.
- Simulated bivariate fBm sample paths synthetic unknown
Source sentence
Figure 1: Simulated sample paths of fBm with H1 = 0.1 (top), and H2 = 0.4 (bottom), which are correlated with ρ = 0.8 and zero asymmetry parameter.
Costs
Not applicable
Finds 4
- Out‑of‑sample forecasts using the time‑reversible mfBm improve predictive accuracy over univariate fBm forecasts. predictive-accuracy
Source sentence
out-of-sample forecasts using the time-reversible mfBm show improvements over univariate fBm, particularly when the estimated Hurst exponents differ significantly.
- mfBm‑based forecasts outperform the (vector) HAR benchmark. performance
Source sentence
Empirical results demonstrate that mfBm-based forecasts outperform the (vector) HAR model.
- The proposed estimator yields substantially lower bias for the asymmetry parameter compared to existing methods. statistical
Source sentence
our estimator for η1,2 demonstrates substantially improved accuracy, particularly when η1,2 = 0.
- When correlations are non‑zero and Hurst exponents differ across components, the multivariate model yields theoretical efficiency gains over the univariate counterpart. theoretical
Source sentence
we establish theoretical results on optimal forecasts, demonstrating that the multivariate model outperforms its univariate counterpart when correlations are non-zero and Hurst exponents differ across components.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 38 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.