Frequently asked questions (FAQ)

In order to achieve this, performance statistics must be calculated on an absolutely uniform basis, and must be as closely to real-life returns as possible.

Performance figures are not particularly designed to reveal the investment returns achieved by a particular individual investor in one fund. This is much better achieved by calculating the market value of a specific investment and comparing it to the entry cost. Entry and exit points often have a dramatic effect on performance calculations. Typically, standardised performance statistics run from the first day of the month to the last day of the month at the end of the period. Investors do not necessarily enter and exit funds on the first and last of any month, so performance figures seldom give an accurate indication of a particular investor's returns.

Standardised performance statistics are useful, however, in comparing funds or sectors. Indeed, it is only by having standardised figures calculated on a uniform basis that such comparisons are possible.

What do the percentages in performance statistics mean?
Not all data vendors calculate performance figures in exactly the same way. It is therefore important to look carefully at table notes and column headers to ensure that you understand what you are looking at.

Profile's performance statistics are presented mainly as the equivalent of annual compound returns. "Compounding" assumes the re-investment of returns, and that growth in subsequent years is earned on the original capital as well as any amount reinvested.

A percentage of 10% over three years means that the fund delivered a return of 10% per annum each year for three years. At this rate of return, R1 000 grows to R1 331 by the end of the third year. The absolute return, as it is called, is therefore 33.1%.

Other statistics available include rolling period returns, risk-adjusted returns (such as Sortino), and returns in rand values.

Why does Profile use annual returns rather than absolute returns?
If we imagine an investor looking at a performance figure, the question we need to answer is: what does the figure mean to the investor? Ideally, we want to present a figure that is comparable, easy to interpret, and fairly universal. We feel annual compound returns best fit the bill, partly because interest rates on savings accounts and fixed deposits are generally quite well understood.

An investor looking at a unit trust fund which has achieved 5% per annum over a certain period can make the following statement: To earn the same return, I would need to have a fixed deposit earning annual compound interest of 10% per annum.

Depending on the inflation and interest rate environment, most investors will immediately have a pretty good idea as to whether this is a good or bad return on investment.

We feel the same is not necessarily true of absolute returns. Is 30% over three years a good return? As a first reaction, many people will say yes. 30% sounds high. But in fact, 30% is only 9.1% per annum – not a particularly good return in the equity markets

The other advantage of annual returns is that they allow comparisons across periods. Absolute returns are obviously comparable across funds for the same period (1 year, 3 years, etc), but it is not that easy to see how 3 year returns compare to 5 year returns. A fund has returned 30% over 3 years and 40% over 5 years. How does that compare? While a bit of thought will no doubt lead to the conclusion that average annual performance over 5 years was not as good as over 3 years, this is easier to see if the performance stats are presented as 9.14% p.a. and 6.96% p.a.

How are dividends / distributions treated in performance calculations?
In South Africa, unit trust management companies typically declare dividends or distributions on the last day of a month, and then pay those dividends anything from a few days to a month later. Some funds declare on the last day of one month and pay on the last day of the following month. This is usually done twice a year, sometimes quarterly.

Performance statistics are by their nature time-based (ie, over one year, or six months, or three years). Sounds simple. But a year is 365 days (unless it’s a leap year), and a day is 24 hours.

If we are calculating a one year performance figure to end December, to include exactly 365 full days we must assume the investor held the investment from 09:00 on 1 Jun to 16:00 on 31 Dec. But if you buy units in South Africa on 1 Jan (assuming it wasn’t a holiday), you would get the price at the end of the day, and any market movement on that day is lost to the investor. In practice, therefore, you would have a period of 364 days.

For this reason, we always buy units using the price on the last day of a month. It’s as if you placed the order on 31 December, which means you were "in" the market from 09h00 on 1 January (again, ignoring the fact this is always a public holiday). For the end point of the calculation we use the price of the following 31 December, which is like giving the sell order on that day and receiving the price at the end of that trading day (which is how SA management companies work).

This system isn’t perfect (some periods do end up being 366 days), but it can be defended as at least "a full year" in every instance.

All well and good, but what about dividends and other distributions?

An investor in a typical SA equity fund will receive two distributions a year; this must therefore be true for the performance calculation as well. But take a fund (eg, the Allan Gray Equity Fund) which declares dividends on 31 December and 30 June. If our calculation period runs for a calendar year, the 30 June declaration is not at issue. But should the investor qualify for the first or last December distribution? Or both?

Given the logic above, in reality the investor who places a buy order on 31 December with Allan Gray is buying cum div, and will therefore receive the distribution. When he sells the following 31 December, he is giving up his right to receive the distribution (because he needs to be registered that night with the fund as a unit holder to qualify, and he will be de-registered before the distribution run commences because he has sold out).

For purposes of calculating performance statistics, we therefore follow the same logic. In the example above, we would use the first December distribution, not the second.

When and at what price are dividends / distributions re-invested for purposes of performance calculations?
To simplify the calculation of performance statistics, some data houses always reinvest the dividend the day after declaration. So if a dividend is declared on 31 December, the program will use the income received to buy further units on 2 January (presuming that is the next trading day). Obviously the price on 2 January will be used as the purchase price.

In reality, it seldom works like this. An investor in the Allan Gray Equity Fund, for example, will in reality only get paid on 31 January – a whole month later. If the distribution is reinvested, it will be at the price ruling on 31 January. In the meantime the market might have moved up or down by a significant percentage, meaning the investor would get less or more units or re-investment (compared to 2 January).

In order to keep performance statistics as accurate and realistic as possible, we re-invest on the re-investment date and at the re-investment price applicable on that day.