From GARCH to Neural Network for Volatility Forecast
Authors: Pengfei Zhao, Haoren Zhu, Wilfred Siu Hung NG, Dik Lun Lee
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiment results validate that amalgamating the NN counterparts of the GARCH family models into established NN models leads to enhanced outcomes compared to employing the stochastic and NN models in isolation. (Abstract) and Experiment Settings, Validation of GARCH-NN Relation, Model Evaluation and Analysis are section titles. |
| Researcher Affiliation | Academia | Pengfei Zhao*1, Haoren Zhu*2, Wilfred Siu Hung NG2, Dik Lun Lee 2 1 Guangdong Provincial Key Laboratory of Interdisciplinary Research and Application for Data Science, BNU-HKBU United International College, 2Hong Kong University of Science and Technology, ericpfzhao@uic.edu.cn, {hzhual, wilfred, dlee}@cse.ust.hk |
| Pseudocode | No | The paper uses mathematical equations and diagrams (e.g., Figure 1) to describe models but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any explicit links to source code repositories or statements about the availability of their own implementation code in supplementary materials. |
| Open Datasets | Yes | We select five asset types covering stock indexes, exchange rates, and gold prices1, which are widely traded equities by investors from all over the world. ... 1Data were downloaded from https://finance.yahoo.com/ |
| Dataset Splits | Yes | We split the complete dataset into training, validation, and testing parts and the split ratio is roughly 8:1:1. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory specifications) used for running the experiments. |
| Software Dependencies | No | We implement the NN counterparts in Py Torch. The vanilla GARCH models use the SLSQP optimizer. ... The stochastic GARCH models and their NN counterparts are trained on the simulation data by the standard ARCH package (Sheppard et al. 2022) and Py Torch, respectively... The paper mentions software like PyTorch and ARCH package but does not provide specific version numbers for these dependencies. |
| Experiment Setup | Yes | For GARCH-LSTM, we tune the initial learning rate to 1e-2 from the range [3e-2, 3e-4] and use GJR-GARCH as the kernel function. For these models, we input the returns series to the encoder and input the volatility series to the decoder to generate future volatility. We set the input length and label length to 126 and use grid search to obtain the optimal hyperparameters. |