how do you calculate value at risk

The simplest method, parametric VaR, assumes returns are normally distributed and multiplies three things: the portfolio value, the volatility over the horizon, and a confidence multiplier (about 1.65 for 95% and 2.33 for 99%). Volatility is scaled to the horizon by multiplying the annual figure by the square root of the time fraction. There are also historical VaR, which reads the loss threshold straight from past returns, and Monte Carlo VaR, which simulates thousands of scenarios. All three answer the same question; they differ in how they estimate the distribution of losses.

what is the difference between 95% and 99% var

The confidence level sets how deep into the tail of losses you are measuring. A 95% VaR is the loss you should not exceed on 95% of days, breached about one day in twenty; a 99% VaR reaches further into the bad tail, breached only about one day in a hundred, so it is always a larger number for the same portfolio. Neither tells you how bad the loss is on the days the threshold is broken — only how often a loss of at least that size should occur. Higher confidence means a rarer, bigger threshold, not a worst case.

what are the limitations of value at risk

VaR’s biggest flaw is that it says nothing about the size of losses beyond the threshold — it tells you a bad day happens perhaps 1% of the time but not how catastrophic that day could be. Parametric VaR also assumes returns are normally distributed, which badly understates how often extreme moves actually happen, so real tail losses exceed VaR more often than the model implies. Over-reliance on VaR contributed to major financial blow-ups, which is why it is now paired with stress tests and expected shortfall, which does estimate the average loss in the tail.

What is Value at Risk (VaR) and how is it calculated

How much could this portfolio lose? Value at Risk answers it with a single number — a loss threshold, at a chosen confidence, over a chosen horizon. It became the risk industry’s common language precisely because it is so compact. That compactness is also its danger: the one thing VaR deliberately hides has caused some of finance’s worst disasters.

One number: value, volatility, confidence

A one-day 95% VaR of ₹20,000 means: on a normal day, you should not lose more than ₹20,000, and a bigger loss is expected roughly one day in twenty. The simplest way to compute it, parametric VaR, multiplies the portfolio value by its volatility over the horizon and by a confidence multiplier (about 1.65 for 95%, 2.33 for 99%), scaling annual volatility to the horizon with the square root of time.

Loading interactive demo…

Set the portfolio value, its volatility and the horizon, switch the confidence between 90/95/99%, and watch the Value at Risk change — note how much bigger the 99% figure is than the 95% one.

Worked example

A ₹10 lakh portfolio at 20% volatility

One-day horizon

Portfolio20% annual volatility₹10,00,000

Daily volatilityScaled to one day20% ÷ √252 ≈ 1.26%

95% VaROne day in twenty1.65 × 1.26% × ₹10L ≈ ₹20,700

99% VaROne day in a hundred2.33 × 1.26% × ₹10L ≈ ₹29,300

The catchOn breach days it can be far worseNeither caps the loss

The 99% VaR is larger than the 95% because it reaches deeper into the tail of bad days. But both are only thresholds: they tell you how often a loss of at least that size should happen, never how large the loss is on the days the threshold breaks. That missing piece is where the real danger hides.

Check yourself

Your portfolio’s one-day 99% VaR is ₹50,000. On a terrible day the market gaps and you lose ₹2,00,000. Did the VaR model “fail”?

Simple bhasha mein

Normal din ka sabse bada nuksaan

VaR ek number: "95% confidence, 1 din — ₹10 lakh portfolio (20% vol) ka nuksaan lagbhag ₹20,700 se zyada nahi hoga," matlab 20 mein 1 din se zyada. Par VaR ye kabhi nahi batata ki us bure din nuksaan kitna bada hoga — sirf kitni baar hoga. Markets ki fat tails ke kaaran asli tail-loss aksar zyada. Isiliye ise stress-test aur expected shortfall ke saath use karo — akele VaR pe crash ka bharosa mat karo.

What to remember

VaR is the loss a portfolio should not exceed over a horizon at a chosen confidence.
Parametric VaR = portfolio × horizon volatility × confidence multiplier (~1.65 at 95%).
A 99% VaR is larger than a 95% VaR because it reaches deeper into the loss tail.
VaR says how often a loss occurs, never how bad it is beyond the threshold.
It assumes normal returns and understates fat tails — pair it with stress tests and expected shortfall.

You reached the endMark it done and keep your streak going.

Up nextExpected shortfall: how bad the bad days really are Previous: When is a system actually dead?

Finished this lesson?

Mark it done to track your progress through the curriculum.

Common questions

Short, direct answers to what people ask about this topic.

what is value at risk: Value at Risk, or VaR, is a single number summarising how much a portfolio could lose over a set period at a given confidence level. A one-day 95% VaR of ₹20,000 means that on 95% of normal days the loss should not exceed ₹20,000 — or equivalently, that a loss beyond ₹20,000 is expected on about one day in twenty. It is the standard risk metric used by banks, funds and regulators because it compresses a portfolio’s risk into one figure that anyone can grasp and compare.
how do you calculate value at risk: The simplest method, parametric VaR, assumes returns are normally distributed and multiplies three things: the portfolio value, the volatility over the horizon, and a confidence multiplier (about 1.65 for 95% and 2.33 for 99%). Volatility is scaled to the horizon by multiplying the annual figure by the square root of the time fraction. There are also historical VaR, which reads the loss threshold straight from past returns, and Monte Carlo VaR, which simulates thousands of scenarios. All three answer the same question; they differ in how they estimate the distribution of losses.
what is the difference between 95% and 99% var: The confidence level sets how deep into the tail of losses you are measuring. A 95% VaR is the loss you should not exceed on 95% of days, breached about one day in twenty; a 99% VaR reaches further into the bad tail, breached only about one day in a hundred, so it is always a larger number for the same portfolio. Neither tells you how bad the loss is on the days the threshold is broken — only how often a loss of at least that size should occur. Higher confidence means a rarer, bigger threshold, not a worst case.
what are the limitations of value at risk: VaR’s biggest flaw is that it says nothing about the size of losses beyond the threshold — it tells you a bad day happens perhaps 1% of the time but not how catastrophic that day could be. Parametric VaR also assumes returns are normally distributed, which badly understates how often extreme moves actually happen, so real tail losses exceed VaR more often than the model implies. Over-reliance on VaR contributed to major financial blow-ups, which is why it is now paired with stress tests and expected shortfall, which does estimate the average loss in the tail.

Value at Risk: the loss you should not exceed on a normal day

One number: value, volatility, confidence

Common questions