Joint Modeling of Multi-Scale Stock Price using Hierarchical Hidden Markov Models

• 발행기관: 한국자료분석학회
• 저자명: Hee-Young Kim Seokju Ki Yongjin Cho
• 간행물 정보: 『Journal of The Korean Data Analysis Society (JKDAS)』Vol.26 No.5, 1275~1288쪽, 전체 14쪽
• 주제분류: 자연과학 > 통계학
• 파일형태: PDF
• 발행일자: 2024.10.31

국문 초록

Hidden Markov models (HMMs) are often applied in quantitative finance to capture the stylized facts of financial returns and to infer hidden state of financial markets, which are set as a hidden state. In the financial technical analysis, these states are known as Bearish, Bullish, and sideway. And these hidden states are important for decision making of stock trading. A lot of articles using HMMs to stock price data have considered Gaussian component distributions. However, in this article, in order explain the most likely sequence with heavy-tailed and leptokurtic in reality, and to capture both short-and long-term trends, we use hierarchical HMMs (HHMMs) with coarse-scale observation process and fine-scale observation process to daily stock price of SK Hynix Inc, from January 1, 2000 to December 30, 2023. Also, we used t-component distributions, not Gaussian distributions. It is shown that HHMMs reproduce various stylized facts of such a long series of daily log returns better than the common Gaussian HMMs.

목차

1. Introduction

2. HMMs and HHMMs

3. Application and illustrations

4. Conclusion

Acknowledgements

References

키워드

Hierarchical hidden Markov model Hidden Markov models Stock prices Regime switching Daily log returns