What is Value at Risk (VaR)?
Reviewed by Karthikeyan Anbarasan (Founder, TradeLyser) · Updated 19 Oct 2026 · Editorial policy
Three ways to calculate it
Historical
Parametric
Monte Carlo
Worked example on a ₹10 lakh portfolio
- Collect 250 daily P&L figures for the portfolio, about one trading year (Figure 1).
- Historical 95%: 5% of 250 is 12.5 days, so the cut-off sits between the 12th and 13th worst. Taking the 13th worst, −2.1%, gives about ₹21,000. Tools differ on this rounding.
- Parametric 95%: the daily standard deviation is 1.2%, so VaR = 1.65 × 1.2% × ₹10,00,000 = ₹19,800.
- Parametric 99%: 2.33 × 1.2% × ₹10,00,000 = ₹27,960.
- Longer horizon: under the usual square-root-of-time rule, 10-day 99% VaR ≈ ₹27,960 × √10 ≈ ₹88,400. The rule assumes days are independent, which real markets often are not.
The historical figure is higher than the parametric one because the real sample has a heavier left tail than a normal curve: five days lost more than 3%. The last tile is the expected shortfall (also called conditional VaR), the average loss on the days beyond the VaR line. It answers the question VaR leaves open: when a bad day comes, how bad is it on average.
How traders use it
For a trader holding several positions overnight, a daily VaR figure turns a list of positions into one number to compare with a limit. If your plan allows a worst normal day of ₹20,000 and the open book’s 95% VaR is ₹31,000, size comes down before the close. VaR is less useful for a single intraday position with a stop; there, risk per trade already gives the planned loss.
- Record the 1-day 95% VaR of the open book at the close each day, and the method used.
- Next day, mark whether the actual loss exceeded it. Over 250 days, a sound 95% VaR should be exceeded about 12 or 13 times.
- Many more breaches than that means the model understates risk; fewer means it may be too cautious.
- Note the size of each breach. If breaches average far beyond the VaR, track expected shortfall alongside it.
Limits and the term it is confused with
Your portfolio VaR
NSE VaR margin
Reference guide
| Context | Value | Reading |
|---|---|---|
| VaR read as a loss exceeded on about 1 day in 20 (at 95%) | Correct reading | Expect 12 to 13 such days in a 250-day year |
| VaR read as the most you can lose | Wrong reading | VaR says nothing about how large the losses beyond it can be |
| Parametric VaR on a book of short options | Misleading | Option P&L is not a straight line in price; use historical or scenario methods |
Questions
What is Value at Risk in simple terms?
A loss figure that a portfolio should exceed only rarely over a set period. A 1-day 95% VaR of ₹19,800 means losses above ₹19,800 are expected on about 1 day in 20.
How is VaR calculated?
Three common ways: historical (read the cut-off from sorted past P&L), parametric (z × volatility × value, with z = 1.65 at 95% and 2.33 at 99%), and Monte Carlo simulation.
What is a good VaR for a portfolio?
There is no standard figure. Set a limit from the daily loss you can accept, compare the book’s VaR with it, and check how often real losses exceed it.
What is the difference between VaR and expected shortfall?
VaR is the cut-off loss at a confidence level. Expected shortfall is the average loss on the days beyond that cut-off, so it shows how bad the bad days are.
Is NSE VaR margin the same as portfolio VaR?
No. NSE VaR margin is an exchange margin rate for each stock in the cash market, collected by brokers with an extreme loss margin. Portfolio VaR is a risk estimate you compute for your own holdings.