Guides

Risk metrics have a credibility problem: they are precise, backward-looking, and easy to mistake for predictions. None of them tells you what will happen. What they do — and this is genuinely valuable — is describe what you have already lived through, in numbers you can compare across time and against alternatives.

Maximum drawdown

The largest peak-to-trough fall in value, measured from every previous high.

It is the most psychologically honest number on the page, because it corresponds to something you actually experienced: the worst it ever felt. A portfolio that returned 12% a year with a 22% maximum drawdown and one that returned 12% with a 55% drawdown are not remotely the same investment, even though the return column says they are.

The asymmetry is worth internalising. A 50% fall requires a 100% gain to recover. A 20% fall requires 25%.

What to do with it. Compare your worst drawdown to what you believe you could tolerate without selling. If your portfolio has fallen 40% and you held on, you have real evidence about yourself. If it has never fallen more than 8%, you have no evidence at all — and you should be careful about assuming you would be calm at −40%.

One technical note: drawdown must be computed on returns, not on raw account value. Otherwise withdrawing money to buy a car registers as a catastrophic loss. BensTrack computes it on the time-weighted return index for exactly this reason — see time-weighted vs money-weighted returns.

Volatility

The standard deviation of daily returns, scaled to a year by multiplying by the square root of the number of trading days.

Roughly, a portfolio with 15% annualised volatility will spend about two thirds of its years within 15 percentage points either side of its average return. For reference: a broad equity index sits around 15–18%, a single large-cap stock 25–35%, and a small speculative position or a cryptocurrency 60–100% or more.

Volatility is the metric most worth being sceptical about, for three reasons. It treats upside and downside identically, so a portfolio that doubles is "risky". It assumes returns are roughly normally distributed, and real markets have fat tails — the 2008 and 2020 moves were statistically impossible under that assumption and happened anyway. And it is entirely backward-looking; volatility is lowest right before it stops being low.

It is still useful for one thing: comparison. Your volatility versus the index, or versus your own portfolio last year, tells you whether your risk profile has drifted. A number that has crept from 14% to 28% means your portfolio has changed character, whether or not you meant it to.

Concentration

The share of your portfolio sitting in its largest position. Unglamorous, and probably the metric most likely to matter.

Concentration is not automatically bad — most large fortunes were built through it. But it needs to be deliberate. There is a considerable difference between "I hold 35% in this company because I understand it better than anything else I own" and "I hold 35% in this company because it went up a lot and I never rebalanced". The second is a decision you never actually made.

Rough guidance, assuming this is money you need:

• Under 10% in any single stock — a bad outcome is annoying, not structural.
• 10–25% — acceptable for a position you have genuinely researched and can defend.
• Over 25% — your portfolio's outcome is now mostly one company's outcome. Be certain that is what you want.

And look past the ticker. Six different semiconductor companies is one bet with six names on it. Real concentration lives at the level of the underlying economic exposure — sector, geography, currency, and interest-rate sensitivity.

The metrics your dashboard cannot show

The most important risks are the ones that resist quantification.

• Liquidity risk — can you sell at a fair price on a bad day? A small-cap that trades thinly has a spread that widens exactly when you need it not to.
• Behavioural risk — will you actually hold? The best strategy you abandon at the bottom is worse than a mediocre one you keep.
• Permanent capital loss — volatility is temporary; a business going bankrupt is not. Every risk metric on this page treats a stock that fell 60% and recovered identically to one that fell 60% and never came back.
• Correlation that appears under stress — diversification is calculated in calm markets and evaporates in crashes, when everything falls together.

How to actually use the panel

1. Check concentration quarterly. It drifts on its own, and it is the easiest thing to fix.
2. Check maximum drawdown after any large market move, and ask honestly whether you behaved the way you expected to.
3. Check volatility yearly, as a drift detector rather than a forecast.
4. Ignore all three for the purpose of predicting next year. That is not what they are for.

Risk management is mostly position sizing and the ability to wait. These numbers are the instrument panel; they do not fly the plane.