Annualise, compare periods and spot cherry-picked start dates

Annualise returns correctly and spot when the choice of start or end date makes a track record look better or worse than it is.

Take the next return figure you are about to put in your workbook and look for three details printed near it. They are the length of the period it measures, whether it is cumulative (the total change over the period) or annualised (a yearly rate), and the dates it runs from and to. A figure can be accurate and still give you the wrong idea, and the misleading ones usually fall down on at least one of these three details. Returns posted on social media or in advertising are generally framed to make you feel something first, before anyone has looked at what is being measured.

In Marcus's group chat, one friend shared a screenshot of a fund that had risen 6% in three months (an invented example), with the comment "That's 24% a year". About a week later, another friend posted the page of a thematic fund (one that invests around a single theme, such as clean energy) showing 140% since launch (also an invented example). Both numbers may be correct as printed. Each was presented to draw a reaction from the chat before anyone looked closely at it.

Returns over less than a year

Annualising turns a return over any period into the yearly rate that, compounded, would give the same result. Over periods longer than a year this helps, since it puts records of different lengths on the same footing, so a three-year record can be set against a ten-year one.

Applied to less than a year, annualising makes up numbers. The calculation treats the months you saw as if they will repeat for the rest of the year, and a single period offers no evidence for that. In a volatile fund, a good quarter is largely noise (random ups and downs).

The friend's 24% contains two errors. Annualising three months at all is the first. Getting there by 6% x 4 = 24% is the second, and it fails even by its own logic, because returns compound. Compounding gives 1.06^4 - 1, roughly 26% a year, and that number is still invalid because three months is less than a year. GIPS (Global Investment Performance Standards), the standards fund managers report under, forbids annualising periods shorter than one year for this reason.

Your workbook follows the same rule: no annualising for anything under a year. Write down the raw return together with its real period, for example "6% over three months".

Reading cumulative and annualised columns

A cumulative return measures the total change across the whole period. An annualised return is the compound yearly rate that would produce that total. Many fund pages print both, sometimes in neighbouring columns, and promotional material usually puts whichever is larger up front.

A larger headline can belong to the weaker fund when it covers a longer stretch. Suppose one fund doubled over ten years (an invented example). Its cumulative return is 100%, and its annualised return is 2^(1/10) - 1, about 7.2% a year. A second fund rose 61% over five years (also invented), which works out to about 10% a year. Year for year the second fund did better, despite the smaller headline. Check the column header every time you copy a figure.

Labels help. A "since launch" figure, such as the 140% from Marcus's chat, is almost always cumulative. Anything covering more than a year should be marked "annualised" or "p.a." (per annum). When a figure over a year carries neither label, assume it is cumulative until you can confirm otherwise. Work out the annualised rate yourself using the formula from lesson 1.1: (ending value / starting value)^(1 / number of years) - 1.

Questioning the start and end dates

For the third check, find out who chose the start date and whether it sits at a low point. Where a track record begins does most of the work in its result. One index over one decade can show under 3% a year or over 15% a year, depending only on the window you pick.

Consider a made-up world equity index with these ten yearly returns in SGD: 15%, -9%, 24%, 7%, 16%, 12%, -16%, 20%, 8%, 13%. Across all ten years it compounds at about 8.3% a year. Begin at the end of year two, right after the first fall, and the eight years from year 3 to year 10 give about 9.9% a year. Take years three to five on their own and you get about 15.5% a year. Begin at the end of year six, just ahead of the second fall, and years 7 to 9 return under 3% a year over those three years. The same index can support a "track record" anywhere between under 3% and over 15% a year.

When a record begins at a market low, almost anything looks good, since the early years include the recovery. Common versions include funds launched shortly after a crash and strategy backtests (tests of a strategy on past prices) that begin in March 2009. Another is the friend who says their stock picks have "doubled since I started", having started at the bottom of 2020. Gains measured from a low did happen, but they tell you very little about skill.

End dates are chosen as well. Suppose a factsheet (the short summary document a fund publishes) goes to print right after a strong quarter. Its 1-, 3- and 5-year figures then look better than those on the same fund's factsheet from one quarter earlier, although nothing about the fund has changed.

Calculating rolling returns in a spreadsheet

The way around a chosen window is to stop depending on any one window. A rolling return is the return over every window of a set length within a history, such as every three-year or every five-year period. A spread of rolling returns gives a fairer picture than one figure, because it covers what an investor with bad timing earned as well as one with good timing.

To build three-year rolling returns:

Put the period-end values in one column. In the next column, divide each value by the value from three years earlier. Raise that result to the power of 1/3. Subtract 1. Drag the formula down the column to produce the rolling series.

On the made-up index, ten years of yearly figures produce eight three-year windows, with annualised returns between about 2.9% and about 15.5% a year. Look at the full range and weight the worst window most heavily, since that is the outcome you could end up living through. Monthly data, where you can get it, gives far more windows than yearly data and shows the spread more clearly.

Rolling returns do nothing to stop other people from cherry-picking. They keep you from fooling yourself, and they make plain how narrow any single headline figure is.

Finding the flattering period on your factsheets

Each period printed on a fund factsheet starts on a date someone else picked. Gather the factsheets for funds you own or are thinking of buying, note which periods each one shows, and find the period that flatters the fund most.

Pick a fund factsheet, annualise each cumulative figure covering more than a year, and mark any period under a year as not annualised. Write down which start or end date makes the fund look best.

Course

Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).