Size does matter: a study on the required window size for optimal-quality market risk models

IF 0.4 4区经济学 Q4 BUSINESS, FINANCE

Journal of Risk Model Validation Pub Date : 2022-01-01 DOI:10.21314/jrmv.2021.015

Mateusz Buczyński, M. Chlebus

引用次数: 0

查看原文本刊更多论文

规模确实很重要:对最优质量市场风险模型所需窗口大小的研究

当涉及到市场风险模型时，我们是应该使用我们拥有的全部数据，还是应该找到一个足够的子样本?我们针对三个不同的指标:WIG20、SP，对历史模拟、GARCH和CAViaR三种风险价值模型进行了不同固定移动窗口长度(从300到2000个观测值)的研究，以寻找最佳匹配截止点。结果表明，训练样本的大小大于900-1000个观测值并不能提高模型的质量，而小于该截止值的长度则不能提供令人满意的结果，并降低了模型的保守性。变化点检测方法提供了更准确的模型。将该算法应用于每个模型的重新计算，结果平均优于1。我们建议使用GARCH或CAViaR模型，重新计算窗口大小。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Journal of Risk Model Validation BUSINESS, FINANCE-

CiteScore

1.20

自引率

28.60%

发文量

期刊介绍： As monetary institutions rely greatly on economic and financial models for a wide array of applications, model validation has become progressively inventive within the field of risk. The Journal of Risk Model Validation focuses on the implementation and validation of risk models, and aims to provide a greater understanding of key issues including the empirical evaluation of existing models, pitfalls in model validation and the development of new methods. We also publish papers on back-testing. Our main field of application is in credit risk modelling but we are happy to consider any issues of risk model validation for any financial asset class. The Journal of Risk Model Validation considers submissions in the form of research papers on topics including, but not limited to: Empirical model evaluation studies Backtesting studies Stress-testing studies New methods of model validation/backtesting/stress-testing Best practices in model development, deployment, production and maintenance Pitfalls in model validation techniques (all types of risk, forecasting, pricing and rating)