日在险价值
出自 MBA智库百科(https://wiki.mbalib.com/)
日在险价值（Daily Value at Risk，DVaR）
VaR(Value at Risk)按字面解释就是“在险价值”，其含义指：在市场正常波动下，某一金融资产或证券组合的最大可能损失。更为确切的是指，在一定概率水平（置信度）下，某一金融资产或证券组合价值在未来特定时期内的最大可能损失。
日在险价值（DVaR）采用历史模拟法，利用两年的历史数据进行计算，同时利用返回检验（Back-testing）方法进行校验。而年在险收益（Annual Earnings at Risk，AEaR）主要度量年收益对市场利率变动的敏感性，置信区间为99%，时间跨度为一年，主要用来度量结构性利率风险和结构资产管理风险。
| 本条目包含过多不是中文的内容，欢迎协助翻译。 若已有相当内容译为中文，可迳自去除本模板。 |
Daily Value at Risk is a financial services firm's estimate of the volatility it expects of its investments in the capital markets. Specifically, it is the most money the firm thinks it will lose in a single day at a 99% confidence level. Stated otherwise, excluding the very worst 1% of days, what's the most money a firm expects to lose in a single day? Value at risk is a useful relative measure of how aggressively a company is investing its own assets. Larger value at risk indicates a company has more to lose - but in all likelihood, more to gain as well.
While companies will sometimes lose more money in a single day than their "daily value at risk", this should occur infrequently if the company is modeling its risk correctly. If the value at risk is calculated at a 99% confidence interval, losing more money than the daily value at risk should occur around 1 in 100 days, or about 3 or 4 days out of the year. However, firms calculate daily value at risk by modeling out past market behavior. Assumptions can vary between firms and flawed assumptions can lead to flawed results. Additionally, if markets begin behaving in ways they haven't behaved in the past, the historical assumptions underlying Value at Risk models may no longer apply.
Although the industry standard is to calculate average daily value at risk at a 99% confidence level, banks can calculate this number at difference confident intervals - for example, at a 95% confidence interval, which would result in a smaller value at risk - but less confidence in whether this decline will be exceeded on any given day. It is also possible to calculate Value at Risk over larger timeframes, such as weekly or monthly.
