Measuring performance — course checkpoint

You began this course able to read a return. You finish able to interrogate one. Before the checkpoint quiz, here is the whole toolkit in one place — the pieces are far stronger assembled than they were one at a time.

The four questions, in order

1. Compared to what? A return means nothing without a benchmark that is investable, chosen in advance, and representative — matched on universe, return basis (total, not price), currency, and fee treatment. Active return = portfolio − benchmark, measured in percentage points. Most flattering performance claims are simply the wrong benchmark, and the four moves to watch are the price-return switch, the risk-class downgrade, the currency swap, and the chosen start date.

2. At what risk? Every risk-adjusted measure has the same shape — something earned over something endured.

• Sharpe = (return − risk-free) ÷ volatility. Annualise with mean × periods and standard deviation × √periods.
• Sortino = (return − target) ÷ downside deviation, dividing by all N periods.
• Information ratio = active return ÷ tracking error — the question that matters for active management.
• Tracking error and active share measure how far the portfolio actually strayed.

3. Through what experience? Drawdown = value ÷ running peak − 1; recovery from a fall of x requires a gain of x ÷ (1 − x). Time under water — peak to new peak — is the duration people actually live through. Calmar and the ulcer index price return against that pain. And money-weighted return answers "how did I do?" while time-weighted answers "how did the manager do?" — the €1,000-then-€9,000 example produced +9.5% a year and −15% a year from the very same fund.

4. How confident can we be? Beta measures market exposure; alpha is what remains after paying for it, and it is meaningless when R² is low. Attribution splits relative return into allocation and selection, revealing how many decisions the record actually contains. Then the statistics: t ≈ IR × √years, so an information ratio of 0.4 needs roughly 25 years to clear the usual bar — and 1,000 zero-skill managers still produce about 31 perfect five-year records.

The two habits that outlast this course

1. Never let a return travel alone. Pair it with the risk it was earned at, the benchmark it should be judged against, and the drawdown it passed through. A number without those three is a fragment.
2. Always ask how many independent observations it rests on. One lucky sector year recorded across two hundred positions is one observation. Sample size and the size of the edge relative to its noise decide together how much a record can prove — that is exactly what t ≈ IR × √years says.

And what none of it does

Every metric here is computed from history, under definitions someone chose, over a window someone selected. A strong Sharpe, a shallow drawdown, a positive alpha — each describes a period that has already happened. None identifies a good investment, ranks the future, or is a reason to buy or sell anything. Precision about the past plus humility about the future is the whole professional stance, and it is the thing worth keeping.

Before you sit it

Each of these is a minute at your desk. Any one that is not names the lesson to reopen first.

• Name two properties a benchmark needs before a record measured against it means anything — How do you choose a benchmark that can't be gamed?
• Compute a Sharpe ratio from a 9% return, 12% volatility and a 4% risk-free rate — How do you compute a Sharpe ratio without fooling yourself?
• Say which of time-weighted and money-weighted return the fund quotes, and which one you earned — Why is your return different from the fund's headline?
• Say roughly how many years of a track record it takes before skill separates from luck — How many years before a track record means anything?

Try it now

1. One fund, below, over its full history. On the 5Y range build a one-page scorecard: CAGR and annualised volatility from the measurements, Sharpe, maximum drawdown, longest time under water, and the active return against a benchmark you would defend.

2. Now move the window. Press MAX, pick a different five-year stretch entirely, and recompute three of those figures. Note which held steady and which moved most — that ranking is your practical guide to which metrics are stable and which are an artefact of where you started measuring.
3. Write one neutral paragraph describing the fund's record using at least four of the measures, containing no verdict and no forecast. That paragraph is the skill this course was built to teach.

Checkpoint quiz next. Nothing in this course was a recommendation to buy or sell anything — you have learned to measure results honestly, which is a discipline, not a signal.