Skip to content

From GARCH to Neural Network for Volatility Forecast

Bibliographic record. Follow the original-source link for the publication.

Field Value
Primary domain Volatility
Other domains
Methods Financial Ml
Facets
Authors Pengfei Zhao, Haoren Zhu, Wilfred Siu Hung NG, Dik Lun Lee
Published 2024-01-29
Source arXiv Quantitative Finance History
Identifiers arxiv:2402.06642
URL Open original source

Editorial synthesis

Why it matters

The item attempts to build an equivalence between GARCH and neural-network formulations and embed it in a single NN architecture, which is relevant for combining model interpretability with NN capacity; however, claims are still abstract-only. (abstract:S1, abstract:S2, abstract:S3, abstract:S4, abstract:S5, abstract:S6, abstract:S7, abstract:S8)

Main author claims

  • The authors note that econometrics and ML for volatility forecasting have evolved separately and propose establishing an equivalence between GARCH and NN models. (abstract:S2, abstract:S3, abstract:S4)
  • They introduce GARCH-NN and integrate GARCH model counterparts as components of an established NN architecture to inject stylized volatility facts. (abstract:S5, abstract:S6)
  • They claim the GARCH-LSTM/GARCH-NN approach yields enhanced outcomes compared to using stochastic models and NN models in isolation. (abstract:S7, abstract:S8)

Data, method, or discussion scope

Evidence scope is limited to stated objectives and aggregate experiment outcomes, without information on data frequency, metrics, backtest structure, or significance framework. (abstract:S4, abstract:S5, abstract:S6, abstract:S7, abstract:S8)

Main limitations

The abstract does not define the exact improvement criteria or identify conditions under which gains hold, limiting direct methodological selection. (abstract:S6, abstract:S8)

Relationships

  • None recorded.