| Summary: | The stock market is a high-risk business that involves a large amount of money, and bad market decision- making can lead to catastrophic losses. The series of stock markets occurs at a higher rate than expected, known as extreme events. Assessing the behaviour of extreme stock market data is challenging and requires accurate risk management strategies to account for the heavy-tailed characteristics of these extreme data. Hence, this study aims to explore and assess the risk in stock market data using the extreme value theory. Additionally, the extreme value theory will be used with Generalized Extreme Value and Generalized Pareto distributions to model the distribution of daily loss probability. At the end of the study, the Value at Risk and Expected Shortfall will be employed together with the extreme value knowledge to fully capture the extent of potential losses in extreme market conditions. This study will be illustrated using daily stock market observations, namely MAY 1155.KL and PB 1295.KL is one of the top two financial sector stocks by top market capitalization in Malaysia stock market exchange. The results show that the risk measure estimate increases for both stock market data as the quantile increase in the Generalized Extreme Value model. As the confidence level increase, the stock PB 1295.KL has a higher risk of market returns than MAY 1155.KL. It is also found that the Generalized Extreme Value model is the best in fitting and estimating the risk for the stock market in the finance sector in Malaysia with low mean square error and mean absolute deviation error values. This study is important as incorporating the extreme value theory with Value at Risk and Expected Shortfall allows for more accurate modelling of the tail behaviour of stock market returns. Extreme value theory focuses on occurrences that substantially influence portfolios, such as market collapses or unexpected downturns. Investors and institutions can make more informed decisions and apply appropriate risk management measures by tracking these tail risk events. © 2023 IEEE.
|
|---|