在金融市场波动愈发频繁和剧烈的金融背景下，金融风险管理和预防是投资者和决策者共同关注的问题，其中，准确度量极端情形下的金融风险也愈发重要和必要。VaR是常用的衡量金融风险的统计量，其影响因素之一是收益分布函数。本文拟通过研究极端收益率数据的分布特点，选择使用极值理论模型POT对极端数据进行建模和拟合，并在模型的参数估计中应用Bayes统计理论，使用Gibbs抽样计算模型参数的Bayes估计值。并通过实证检验思路的可行性。模型的结果经过Kupiec检验效果良好，说明使用基于Bayes参数估计的极值理论模型应用于金融风险度量是可行的。

In the financial context of increasingly frequent and violent financial market fluctuations, financial risk management and prevention are common concerns of investors and policymakers, among which financial risks in extreme cases of accuracy measurement are becoming more important and necessary. VaR is a commonly used statistic to measure financial risk, and one of its influencing factors is the return distribution function. In this paper, by studying the distribution characteristics of extreme yield data, we choose to use extreme value theory model POT to model and fit extreme data, apply Bayes statistical theory in the parameter estimation of the model, and use Gibbs sampling to calculate the Bayes estimate of model parameters. And through the empirical test of the feasibility of the idea. The results of the model are well tested by Kupiec test, indicating that it is feasible to apply the extreme value theory model based on Bayes parameter estimation to financial risk measurement.