How the numbers work

Returns

Returns are chained, never added. Twelve months of 1% is 12.68% for the year, not 12%, because each period grows on the last one’s result.

Windows shorter than a year are shown as the plain compounded return. Windows of a year or more are converted to a per-year rate so that a three-year and a five-year figure can be compared directly — that conversion is what p.a. on a column header means. It is also why a five-year p.a. figure looks smaller than the total return over those five years: one is a yearly rate, the other is the whole journey.

Why two figures can differ — Two conventions account for almost every apparent disagreement. First, one summary table compounds through twenty-four months before it annualises, where the others switch at twelve — which is why its column reads 2 Yrs Ave. Return. Second, the drawdown chart never lets the running peak fall below the starting value, while the maximum drawdown figure in the tables does; on a portfolio that fell early, the two legitimately differ.

Risk measures

What each one is actually telling you.

Reading capture ratios

When the market rose, how much of the rise did the portfolio get? When it fell, how much of the fall did it take? Periods are split by what the benchmark did, not the portfolio.

The ideal is high up-capture with low down-capture: a portfolio at 90% up and 60% down gives up a little in rallies to lose much less in falls. Read the pair together — either number alone can flatter a portfolio, and a capture ratio built on three down periods is not evidence of much.

A name clash worth knowing — Capture ratios are calculated by one component: the performance consistency and risk table. The two charts named Capture Vs Benchmark do something quite different — they plot the portfolio’s best or worst eight periods against what the benchmark did in those same periods, and compute no ratio at all.

Weights and contribution

Contribution is how much of the portfolio’s return one holding was responsible for. A holding matters through both how well it did and how much of it you owned, which is why a small holding that soared can contribute less than a large one that did moderately well. Where dollar-level data exists, contribution is calculated from actual movements instead — more accurate when money flowed in or out during the period.

Look-through is the real exposure to a company once you look inside the funds holding it. If four funds each own the same bank, look-through combines them into one line rather than leaving four hidden ones.

Every excess, difference or vs column is a plain subtraction in percentage points — never a ratio.

Conventions

• Income and growth — Total return split into what was distributed and what came from price movement. Growth is whatever is left after income, and the split renders only where distribution data exists.
• Fiscal-year reporting — Performance can be presented on a fiscal year rather than a calendar one, configured to whichever convention applies in your market. A part-finished year shows as year-to-date. Where clients are taxed and reported to on a fiscal year, calendar figures make them do a mental translation.
• Quartiles — Where the portfolio sits in the spread of what it is compared against, in four bands. Q1 is the best. Rankings are on return alone, so a top-quartile portfolio is not necessarily a well-run one — it may simply have taken more risk.
• Annualising conventions — Some conventions are configurable at account level — whether volatility uses the sample or population calculation, and whether annualising counts whole periods or actual days elapsed. The latter matters for portfolios that started mid-period. Yours are set during onboarding and applied consistently.

Full methodology — This page covers the concepts. Complete calculation methodology — the formulas, conventions and edge cases behind every figure — is available to clients and their consultants on request, and we will walk through any of it with your investment committee. Ask through Help & Support.

What a component needs configured

Some components are comparisons and cannot be calculated without something to compare against. This is the reference behind Why a chart is empty.