What VaR Actually Means for a 5-Stock Portfolio

2026-11-12 ยท hedgewing.ai Research

Value at risk was built for bank trading desks, and most explanations still read that way. But the concept is just as useful for an individual holding five stocks, provided you understand what it measures and, more importantly, what it leaves out. This article walks through the number for a concrete small portfolio.

The definition in one sentence

A one-day 95% VaR of 2.5% means that, based on the return distribution the model is using, you should expect to lose more than 2.5% of the portfolio on about 5% of days, roughly one day in twenty. The 99% VaR is the loss level exceeded on about one day in a hundred. VaR is a threshold, not an average and not a maximum: it says how often you cross a line, not how far past it you go. The glossary entries at /glossary/value-at-risk and /glossary/conditional-value-at-risk cover the formal versions.

A worked example

Suppose you hold five large-cap US stocks in roughly equal weights, worth $20,000 in total, each with an annualised volatility around 30%, which is about 1.9% per day. If the five were perfectly correlated, the portfolio's daily volatility would also be 1.9% and a normal-distribution 95% VaR would be about 1.65 times that, near 3.1%, or $620. If they were uncorrelated, the portfolio's daily volatility would fall to 1.9% divided by the square root of five, about 0.85%, and the VaR to roughly $280. Real five-stock portfolios sit between those cases, typically with pairwise correlations of 0.3 to 0.6, giving a VaR somewhere around $400 to $500. Correlation is the whole story: the same five stocks produce a VaR that differs by more than double depending on how they move together.

Why five stocks are a special case

With five holdings, one name is 20% of the portfolio, and single-stock events dominate the tail. An earnings miss that drops one stock 15% is a 3% hit to the portfolio from one position alone, which is already past the 95% VaR line. Fat tails, described at /glossary/fat-tails, are therefore not a theoretical concern here: the 99% VaR computed from normal assumptions will understate what a real bad day looks like, and conditional VaR, the average loss on the days beyond the line, is often the more honest number for a concentrated portfolio.

Three things VaR does not tell you

First, it does not tell you the size of the loss when the threshold is breached; that is what expected shortfall is for. Second, it does not tell you how long a drawdown will last or how deep it will go across consecutive bad days; maximum drawdown measures that. Third, it assumes the correlations and volatilities in the lookback window will hold, and they reliably rise together in a crisis, so VaR is lowest exactly when it is about to be most wrong. Treat VaR as an everyday guardrail and stress scenarios as the real test.

Using the number

The practical use of VaR for an individual is as a sizing check: if the 95% VaR is larger than the loss you could accept on an ordinary bad day, the portfolio is too large or too concentrated for your tolerance, regardless of how good each individual idea is. hedgewing.ai reports VaR at the 95% and 99% levels for stocks and tracked portfolios, alongside Sharpe, Sortino, factor exposures, and an HRP allocation that explicitly uses the correlation structure to reduce concentration. These are descriptive risk metrics, not predictions, and this article is not investment advice.

Create a free Hedgewing account to try hedgewing.ai yourself.