Volatility Forecasting and Return Prediction under Market Regimes: Evidence from High-Frequency Chinese Equity Data¶
Bibliographic record. Follow the original-source link for the publication.
| Field | Value |
|---|---|
| Primary domain | Volatility |
| Other domains | — |
| Methods | Financial Ml, Research Methods |
| Facets | — |
| Authors | Xinyue Fang, Robert Ślepaczuk |
| Published | 2026-06-08 |
| Source | arXiv Quantitative Finance History |
| Identifiers | arxiv:2606.09478 |
| URL | Open original source |
Editorial synthesis¶
Why it matters¶
This preprint joins regime identification, volatility forecasting, return prediction, and implementation in one pipeline, while explicitly showing that better forecasts do not automatically become post-cost strategy value. (abstract:S5, abstract:S6, abstract:S7)
Main author claims¶
- On high-frequency CSI 300 data from 2005–2023, the authors report that regime-aware volatility models outperform baseline HARQ across forecast metrics. (
abstract:S2,abstract:S5) - The authors find weak return predictability concentrated in low-volatility regimes and state that volatility scaling, gating, thresholds, and turnover controls can improve defensive economic performance. (
abstract:S6,abstract:S7,abstract:S8)
Data, method, or discussion scope¶
The abstract describes a two-stage HARQ / Markov-switching GJR-GARCH and XGBoost framework with walk-forward out-of-sample estimation, but gives no result table, cost function, or gating parameters. (abstract:S2, abstract:S3, abstract:S4)
Main limitations¶
Evidence comes from one Chinese equity index and the return signal is weak and regime-dependent; economic improvement may be sensitive to regime definitions, thresholds, turnover, and cost assumptions. (abstract:S2, abstract:S6, abstract:S7)
Relationships¶
- None recorded.