Parametric VaR: Loss From Volatility and a Curve

Key Takeaways

• Parametric VaR multiplies portfolio volatility by a z-score to find the loss at a confidence level.
• The 95 percent z-score is about 1.645 and the 99 percent z-score is about 2.326.
• Its core flaw is the normal-curve assumption, which understates fat-tailed crash risk.
• It scales fast across large portfolios using a covariance matrix, which suits firm-wide risk reporting.

What It Is

Parametric VaR assumes that portfolio returns are drawn from a specified probability distribution and uses that distribution's mathematics to find a loss threshold. The classic version uses the normal distribution, which is fully described by two numbers: the mean return and the standard deviation, or volatility.

Because a normal curve has a fixed shape, the loss at any confidence level is just the volatility scaled by a known multiplier called a z-score. This is why the approach was the heart of the original RiskMetrics framework: with a covariance matrix of asset returns, you can compute VaR for a portfolio of thousands of positions in one calculation.

The Intuition

If you accept that returns are normal, you do not need to look at every historical loss. The whole tail is implied by the volatility. A wider spread of returns means a larger possible loss at the same confidence level.

The z-score is the bridge. It says how many standard deviations below the mean a given probability sits. At 95 percent confidence, 5 percent of the curve lies more than 1.645 standard deviations below the mean. Multiply volatility by that z-score and you have the VaR. The elegance is real, and so is the catch: markets are not actually normal.

How It Works

For a single position or a portfolio treated as one return series, the formula is:

VaR = (z * sigma - mu) * portfolio value

Where:

z     = z-score for the confidence level (1.645 at 95%, 2.326 at 99%)
sigma = return volatility over the horizon
mu    = expected return over the horizon (often set to zero for short horizons)

For a multi-asset portfolio, sigma comes from combining each asset's volatility with the correlations between them through a covariance matrix:

portfolio variance = w' * Sigma * w

Here w is the vector of position weights and Sigma is the covariance matrix. The square root of portfolio variance is the portfolio volatility you feed into the VaR formula. Over short horizons the mean term is often dropped because daily expected return is tiny next to daily volatility.

Worked Example

You hold a 1 million dollar portfolio with a daily return volatility of 1.2 percent. You want the 1-day 99 percent VaR and you set the expected daily return to zero.

The 99 percent z-score is 2.326.

VaR = 2.326 * 0.012 * 1,000,000 = 27,912 dollars

So parametric VaR says you should lose more than about 27,900 dollars on roughly 1 day in 100. To convert to a 10-day horizon under the independence assumption, you scale volatility by the square root of 10:

10-day VaR = 27,912 * sqrt(10) = 88,266 dollars

That square-root-of-time scaling is convenient but assumes returns are independent across days, which crisis periods routinely break.

Common Mistakes

1. Trusting the normal assumption in the tail. Real returns have fat tails. A move the normal curve calls a once-in-a-century event can show up several times a decade, so parametric VaR understates extreme loss.
2. Using a stale covariance matrix. Correlations spike toward 1 in a crash, when diversification fails most. A matrix estimated in calm markets will overstate the benefit of spreading risk.
3. Square-root-of-time scaling without thought. It assumes independent, identically distributed returns. Volatility clustering and autocorrelation make the scaled number unreliable for longer horizons.
4. Applying it to optionality. Parametric VaR assumes returns are roughly linear in the risk factors. Portfolios with options or convexity need delta-gamma adjustments or a different method.
5. Ignoring estimation error in volatility. The whole result hinges on the volatility input. A poorly estimated sigma propagates straight into a misleading VaR.

Frequently Asked Questions

What is parametric VaR in simple terms? Parametric VaR estimates potential loss by assuming returns follow a normal curve and scaling the portfolio's volatility by a multiplier. It gives a fast loss threshold without sorting through past returns one by one.

How does parametric VaR affect investment decisions? It lets risk managers compute firm-wide loss limits quickly using a covariance matrix. Because it relies on a normal curve, prudent users add fat-tail adjustments or stress tests before sizing positions.

What is a real-world example of parametric VaR? A 1 million dollar portfolio with 1.2 percent daily volatility has a 1-day 99 percent VaR of about 27,900 dollars, found by multiplying volatility by the 2.326 z-score and the portfolio value.

How can investors avoid misusing parametric VaR? Do not treat the normal-curve output as a worst case. Supplement it with historical or Monte Carlo VaR, refresh the covariance matrix often, and stress test correlation breakdowns.

How is parametric VaR different from historical VaR? Parametric VaR assumes a distribution shape, usually normal, and computes loss from volatility and a z-score. Historical VaR makes no assumption and reads the loss directly from sorted past returns.

Sources

1. J.P. Morgan/Reuters. RiskMetrics Technical Document, 4th ed. (1996). https://www.msci.com/documents/10199/5915b101-4206-4ba0-aee2-3449d5c7e95a
2. CFA Institute. Measuring and Managing Market Risk. https://www.cfainstitute.org/insights/professional-learning/refresher-readings/2026/measuring-managing-market-risk
3. Basel Committee on Banking Supervision. Minimum Capital Requirements for Market Risk. https://www.bis.org/bcbs/publ/d457.htm
4. Investopedia. Value at Risk (VaR). https://www.investopedia.com/terms/v/var.asp

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.