What it means
Value at risk tells you a loss level, for example that a portfolio should not lose more than $700,000 on 95 days out of 100. Its weakness is that it says nothing about the other five days, so two portfolios with identical value at risk can carry very different disaster profiles.
Conditional value at risk closes that gap by averaging all losses beyond the threshold. If the worst 5% of outcomes average a loss of $920,000, that is the figure, and it will always be at least as large as the value at risk number sitting behind it.
Banks, funds and insurers use it for capital planning and limit setting because it responds sensibly when tail losses grow. Adding a small position that can occasionally lose an enormous amount barely moves value at risk but visibly moves conditional value at risk, which is exactly the behaviour a risk manager wants from a measure.
It is calculated in three main ways: from historical returns, from a modelled distribution, or from Monte Carlo simulation, which generates thousands of possible future paths. Historical calculation is the easiest to explain to a board, though it can only show tail events that have already happened within the sample period.
The measure carries the same health warnings as any risk statistic, since it depends on the confidence level chosen, the holding period and the quality of the underlying data. Quoting it without stating those three inputs makes the number close to meaningless, and comparing two firms' figures without checking them is worse than useless.
In practice
Real-world examples.
Example
A pension fund sets its hedging budget using conditional value at risk rather than value at risk, because it cares about the size of a shortfall it would have to fund, not just how often one occurs. The switch raises the modelled worst case from $18,000,000 to $24,000,000 and justifies buying more protection.
Example
A commodity trading firm compares two strategies with identical value at risk of $500,000. Conditional value at risk is $650,000 for one and $1,400,000 for the other, so the risk committee cuts the second book's limit despite the headline numbers matching.
Example
An insurer models catastrophe exposure with Monte Carlo simulation and reports conditional value at risk at the 99% level to its board. The figure drives how much reinsurance it buys for the coming year.
Think of it
“CVaR is the average loss when things go really bad-expected shortfall beyond VaR.
Formula
Calculation
Conditional value at risk = the average of all losses worse than the value at risk threshold
Take a fictional trading book simulated over 100 equally likely daily scenarios at a 95% confidence level. The five worst scenarios produce losses of $1,200,000, $1,000,000, $900,000, $800,000 and $700,000, so value at risk is the least severe of those five, $700,000.
Conditional value at risk averages all five: ($1,200,000 + $1,000,000 + $900,000 + $800,000 + $700,000) / 5 = $4,600,000 / 5 = $920,000. The desk should therefore plan for a bad day costing $920,000 on average, not the $700,000 the value at risk headline suggests, a difference of $220,000 per event.Case study
Seen in the real world.
The following is an illustrative and entirely fictional example. Kestrel Lane Capital, an invented multi strategy fund, reported daily value at risk of $2,000,000 at 95% confidence for three years and never breached its board limit. Its risk reports stopped at that single number, and the board treated a quiet run of results as evidence the portfolio was safe.
A new risk officer recalculated the same book using conditional value at risk and found the average loss in the worst 5% of simulated days was $5,400,000, nearly three times the value at risk figure. The gap came from a small options position that lost almost nothing most of the time and an extraordinary amount very rarely, a shape value at risk could not see.
In the fictional outcome, Kestrel Lane did not close the position but halved it and set a separate tail limit expressed in conditional value at risk terms. Reported returns fell slightly, and the board gained a number that actually described what a bad month would feel like.
Watch out
Common mistakes.
- Treating conditional value at risk as a worst case loss, when it is an average of bad outcomes and individual losses can be far larger.
- Comparing figures from two institutions without checking that the confidence level and holding period match.
- Relying on historical calculation over a short, calm sample period, which produces a comfortable number simply because nothing bad has happened yet.
Questions
People also ask.
Is conditional value at risk always higher than value at risk?
Yes, because it averages the losses beyond the value at risk point, so it can equal it only in unusual cases and never falls below it.
Which confidence level should a business use?
95% and 99% are the common choices, with 99% giving a more severe figure that suits capital planning and 95% suiting day to day limit monitoring.
Does this measure work for non financial risks?
The maths applies to any loss distribution, so operational and insurance teams use it too, but it needs enough credible scenario data to be meaningful.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%