A value at risk calculator estimates how large a loss a portfolio could suffer over one period at a chosen confidence level, using its own history. Enter past returns, a confidence level such as 95 percent, and optionally the portfolio value. The calculator reads the value at risk, or VaR, off the sorted history and converts it to a money amount.
What value at risk is
NYU Stern's V-Lab describes value at risk as a statistical measure of the potential loss in value of an asset or portfolio over a defined period at a given confidence level. Its 95 percent VaR is the loss with a 5 percent chance of being exceeded, so a one-day 5 percent VaR of one million dollars means a 0.05 probability of losing one million dollars or more in a day, a loss expected on about one day in twenty. The 5 percent VaR is also called the 95 percent VaR, and VaR is conventionally reported as a positive number.
The VaR formula used here
VaR = −inf{x : F(x) > 1 − confidence}
where F is the share of past returns at or below x
The mathematical definition is that the VaR is the negative of the infimum of the values x for which the distribution function is above α, equivalently the (1 − α) quantile of the loss. The calculator applies that definition to the distribution of the returns you enter, where F(x) is the share of the observations at or below x. Historical simulation is named among the nonparametric ways of estimating VaR; this is the simplest form, with no assumption about the shape of returns.
How to use the value at risk calculator
- Enter past returns, one per period, in percent, with losses as negative numbers.
- Set the confidence level, or leave 95 percent.
- Optionally enter the portfolio value to see the VaR in money.
- Press Calculate.
Worked example
Take 20 daily returns, from −4.1 percent to +2.2 percent, as in the placeholder. At 95 percent confidence, α is 0.05, and with 20 observations 5 percent of them is one observation, so the smallest value where the distribution function exceeds 0.05 is the second-worst return, −3.2 percent. The VaR is therefore 3.2 percent, or 3,200 on a portfolio of 100,000. At 99 percent it is the worst return, 4.1 percent. The returns are made up for the illustration.
Reading the result
A 3.2 percent one-day VaR at 95 percent confidence means that, in this history, losses of 3.2 percent or more happened on about one day in twenty. It does not say the loss cannot be larger: the worst day in the data was 4.1 percent, and the next could be worse. The calculator shows the worst loss in the data and how many observations lie at or beyond the VaR so that you can see how much data stands behind the figure.
The limits of VaR
Criticisms of VaR include a bet that wins 100 dollars with probability 127 in 128 and loses 12,700 dollars otherwise: the 1 percent VaR is zero, because the loss has a probability below 1 percent, even though the loss is large. In the same way, a single-branch bank with a tiny daily chance of a robbery would not show it in a one-day 1 percent VaR. VaR tells you where the tail begins, not how deep it is.
How much data to use
The more observations you have, the more stable the estimate, particularly at high confidence levels. At 99 percent confidence you need at least 100 observations for the VaR to rely on more than the worst one. Use returns for the same period length as the VaR you want, such as daily returns for a one-day VaR, and use a history that includes the kinds of conditions you want to guard against. Very short histories can miss rare events entirely.
Related tools
For the spread of returns, the standard deviation calculator is a first step, and the portfolio beta calculator measures sensitivity to the market. The average calculator gives the mean of the returns, and the percentage calculator helps convert prices to returns.
A second example at a different confidence level
Using the same 20 returns at 90 percent confidence, α is 0.10, which is two observations out of 20, so the smallest value where the distribution function exceeds 0.10 is the third-worst return, −2.6 percent. The VaR is 2.6 percent, lower than at 95 percent, because a lower confidence level excludes more of the bad outcomes. The more confidence you ask for, the larger the loss, and the fewer observations stand behind it.
Choosing the period
VaR depends on the period of the returns. A one-day VaR uses daily returns, a one-week VaR uses weekly returns, and so on. The calculator does not rescale one period into another, because that would need assumptions that the history may not support. If you need a weekly figure, enter weekly returns. Make sure the figures are returns for the same period and in the same units, and that a loss is a negative number.
Common mistakes
- Entering losses as positive. Losses must be negative numbers.
- Mixing period lengths. Use returns for the period you want the VaR for.
- Using too little data. High confidence levels need many observations.
- Reading VaR as the worst case. Losses beyond it are possible.
What next
Use this often? Add Value at Ris to your Home screen.
Frequently asked questions
How does a value at risk calculator work?
It sorts your past returns and reads off the loss at the chosen confidence level.
What does a one-day 5% VaR of $1 million mean?
A 5% probability of losing $1 million or more in a day, about one day in 20, according to NYU Stern's V-Lab.
Is 5% VaR the same as 95% VaR?
Yes. The two names are used for the same measure.
Does VaR show the worst possible loss?
No. It does not say how large losses beyond it can be.
How many observations do I need?
More is better; at 99% confidence, use at least 100.
Was this tool helpful?
What went wrong?
We only receive the reason you pick, never what you typed into the tool.
Thanks for your feedback.