Abstract

Stock market crashes are rare but can cause major problems for investors, financial institutions, and policymakers. The standard risk models used tend to assume that stock returns are normally distributed, leading to an underestimation of extreme movements in the stock market. The purpose of this study is to analyse the performance of probability distribution models developed by Extreme Value Theory (EVT) on extreme negative returns of the NIFTY 50 index to check if the prediction of stock market crashes can be improved. The study investigates the characteristics of events with large market losses and provides an estimate of the likelihood of such events based on the Peak Over Threshold (POT) and the Generalised Pareto Distribution (GPD). The results demonstrate that the NIFTY 50 returns are fat-tailed and considerably deviated from normal distribution, implying that extreme losses are not all that rare. Contrary to traditional models like Valueat- Risk (VaR), EVT models can provide more precise estimates of the probability and intensity of extreme market events, resulting in a more holistic approach to risk analysis of financial products. This study contributes to the existing literature on financial risk management in emerging markets by demonstrating the usefulness of EVT in understanding extreme market behaviour and forecasting crashes. The results have implications for investors, portfolio managers, risk managers, and policymakers, as they help them make informed investment decisions, build better risk management strategies, and make the equity market more resilient.

Keywords: Extreme Value, Theory (EVT), Stock Market Crashes, NIFTY 50, Generalised Pareto Distribution (GPD), Tail Risk, Financial Risk Management