Portfolio risk analysis: volatility, VaR, drawdown and correlations for normal humans

Two portfolios up 8% can hide opposite realities. The four risk measures explained in plain language — honest limits included.

Your portfolio is up 8% over the past year. Good news? Impossible to say. Performance is the number everyone stares at, yet it is the one that tells you the least: two portfolios both up 8% can hide opposite realities — a quiet ride, or a rollercoaster that nearly came off the rails three times. The difference is not written in the return. It is written in four risk measures.

Volatility, VaR, drawdown, correlations: four words that sound like a finance lecture, four ideas that are actually simple. This article explains each one with a concrete image, is honest about what each one cannot see — every measure has a blind spot — and, most importantly, shows how to read them together.

Performance is a rearview mirror

Performance describes what happened along one path — the one that got realised. Risk asks the other question: what could have happened, and what can still happen to you? Someone who bets everything on red and wins shows a superb return and made a terrible decision. Judging a portfolio by its return alone is judging a driver by the fact that he got home, without ever watching how he drives.

This is also why the regulatory line is not empty boilerplate: past performance is no guarantee of future results. Risk measures do not predict the future either — but they describe how your portfolio behaves, and behaviour has a certain inertia that a single return figure does not.

Volatility: the amplitude of the ride

Two roads lead to the same town: a flat motorway, and a mountain road full of hairpins. Same destination, incomparable journeys. Volatility measures exactly that: the typical amplitude of your portfolio's swings, expressed as an annualised percentage. At 12% volatility, an ordinary year's oscillations stay contained; at 38%, the same oscillations become lurches — bright-red days that test your nerves.

In practical terms: the higher the volatility, the rougher your ordinary days, in both directions. It is the most widely used measure because it compresses the general mood of the ride into a single number.

95% VaR: your typical bad day

VaR (Value at Risk) at 95% answers the question every investor asks without daring the jargon: “how much can I lose on a bad day?”. The definition fits in one sentence: it is the daily loss you only exceed 1 day in 20, historically. Nineteen days out of twenty, your daily loss — when there is one — stays below that threshold. Convert the percentage into money at the size of your portfolio and you get the cost of your “ordinary bad day”.

Beware the classic misreading: VaR is not the worst case. It says nothing about how bad the one-day-in-twenty actually gets — that day can exceed the threshold by a little or by a lot. It is a vigilance floor, not a loss ceiling. Its virtue is concreteness: volatility is an abstraction smeared across a whole year, while VaR speaks the language of one single bad day.

Drawdown: the fall that makes people sell at the bottom

Maximum drawdown is the worst decline from a peak to the trough that followed — the depth of the wave measured from its crest. Of all the measures it is the most psychological: what makes investors capitulate is not volatility, it is watching their portfolio sit 40% below its high and selling right there, at the exact worst moment. A drawdown you never anticipated almost always becomes a drawdown you endure.

The arithmetic is cruel and purely mechanical: after a 50% fall, you need a 100% rebound just to get back to even. To see what real drawdowns do to a concrete portfolio, our stress test replays 2008, 2020 and 2022 — three crises with very different shapes.

Correlations: do your holdings move together?

Fifteen lines in a portfolio feels reassuring. But if all fifteen rise and fall as one block, you do not own fifteen bets — you own one bet, in fifteen copies. Correlation measures exactly that: two lines that always move in the same direction correlate near 1; two independent lines, near 0. Diversification only exists when correlations are low. That trap deserves an article of its own: this portfolio looks diversified — it isn't.

Two portfolios, same return — an illustrative example

Let us put it all side by side. The two portfolios below are fictional, built purely for illustration — no real data here, only orders of magnitude kept mathematically consistent with each other:

Same +8% at the finish line. But portfolio B went through −45% on the way: its typical bad day wipes out three times more than portfolio A's, and its owner had to watch nearly half the capital evaporate without selling. How many investors actually hold through that ride? The final return showed none of it. That is the whole point of risk analysis: the two journeys were never comparable, and a single number — the return — made them look identical.

What each measure cannot see

None of these measures is an oracle, and selling them as one would be dishonest. Their blind spots, one by one:

• Volatility counts rises the same as falls: an asset that climbs fast shows “bad” volatility. It measures agitation, not direction.
• Historical VaR only sees the past inside your sample: a risk that never materialised there is invisible to it. And it says nothing about how large losses get beyond the threshold.
• Maximum drawdown is the worst episode of a single realised path. A short history simply has not had time to show its own — the number looks reassuring because it is young.
• Correlations are unstable: they climb towards 1 precisely during crises. Diversification measured in calm weather melts exactly when you need it.

Reading them together

In isolation, each measure tells a fragment. Together they sketch the full silhouette of your risk, and the reading collapses into four plain questions:

• Volatility — does my portfolio's ordinary amplitude suit me, or am I living beyond my nerves?
• 95% VaR — is my typical bad day, converted into money, bearable without panic?
• Drawdown — have I lived through, or at least simulated, my worst fall? Could I cross it without selling at the trough?
• Correlations — are my holdings genuinely several bets, or one bet in disguise?

Four honest answers beat ten performance charts. They also change the nature of your decisions: you stop adjusting a portfolio because it went up or down, and start adjusting it because its behaviour no longer matches what you can hold. That shift — from reacting to prices to responding to behaviour — is what separates analysing a portfolio from merely watching it.

What OplynQ does

OplynQ computes these four measures on your real, consolidated portfolio — every account merged — and sums them up in a risk score that is explained: not a grade falling from the sky, but a number shipped with what pulls it up or down. An AI copilot then answers your questions with your own figures as evidence. And when data is missing or a history is too short, you see “—” and the reason: an absent measure is more honest than an invented one.