How to Read a Fund Return Without Reading a Forecast

A published scheme return describes one stretch of time that has already finished. The charge was taken out of the values the figure was built from, so the figure arrives net. Nobody can hold an index, so an index figure carries no cost at all. Set the two side by side and two different things have been compared. Neither says anything about the year ahead.

A return is printed as a percentage, and a percentage looks like a fact about the world rather than a fact about a document. A return is the second kind of thing. Somebody picked two dates, took the value of one unit on each of them, and divided. Everything inside the number was put there by that choice, and nothing else got in at all. Read a return as a description of a document rather than as a description of what is coming, and most of the mistakes made in this area simply stop happening.

One asset manager and two schemes run through everything below. Girnar Asset Management Limited, an invented company, is the manager here. The equity scheme it runs, called the Girnar Large Cap Equity Fund, is open ended; it carries net assets of Rs 4,200 crore and has 120.00 crore units in issue. Set the units figure beneath the assets figure and one unit works out at Rs 35.00 on the nose, a division done here rather than a number lifted from anywhere. A second scheme also belongs to the same manager, the Girnar Broad Market Index Fund, and it follows a broad index that is left unnamed. Kalyani Bhagat runs the equity portfolio; Sohail Merchant runs operations. Five further parties matter and none is named: whoever acts as trustee, whoever holds the securities, whoever keeps the folio records, whoever audits, and whoever distributes. Each appears here as a role rather than as a name.

Three matters are settled already, and nothing below rebuilds them. Already established is what an expense ratioA scheme ongoing charge, put as a yearly percentage of what it holds after what it owes. does, and that it bites into a scheme assets daily instead of turning up as a bill for someone to settle. Also established: a single value per unit gets struck each day from the scheme own accounting, and a scheme names a measure it stands to be compared with. All three arrive here as settled, and only one job follows from them: learning what has to be written next to a return figure before it is allowed to do any work.

So what is that percentage a record of, exactly?

Of one stretch of time that has already finished, on one scheme, computed from that scheme's own value per unit at two dates. Two dated values and one division are the entire content. So before the number is read, what kind of number it is has to be written down, and there are four things to write down every single time.

The figure itself. The length of the window it covers. The basisWhether a figure is stated before the charges come out or after them. Two figures on different bases cannot be set against each other. it sits on, meaning whether the charge is inside the number or still outside it. And the scheme it belongs to. A reader who cannot write all four has not been handed a return at all, only a number.

Any other measurement is already treated this way. A shopkeeper says rice went up eleven per cent. The next question is not whether eleven is a lot. The next question is: since when? Eleven per cent since last week and eleven per cent since the year before last are not the same claim, and neither of them is the same claim as eleven per cent for one variety in one shop. Nobody finds that hard with rice. The same question very often goes unasked when the percentage is printed beside the name of a scheme.

Writing the four out takes about eight seconds, and nothing else about reading a return is cheaper. Writing them also does what a caution never manages. The missing annotation becomes obvious. A basis that was never given cannot be written down. The blank on the sheet of paper is what marks it as missing, and the finding is the work.

Why is a return with no window stated beside it not yet a figure?

Because a return is a rate, and a rate carries no meaning until what it is a rate over is stated. The periodThe stretch of time a return covers, from the first date to the second. A return quoted without one is not yet a quantity. is not decoration attached to the figure. The period is one of the two things the figure was actually built out of, the other being the scheme. Take it away and the arithmetic that produced the number is no longer reconstructable by anybody.

Of the four annotations the period is the simplest to verify. A headline also drops it most readily, and the pairing is a poor one. Similarity of appearance is not a property the arithmetic has any interest in, so two returns computed over windows of different lengths cannot be compared with each other however similar the two numbers look. A figure covering one year and a figure covering three years are answers to two different questions, and setting them beside each other produces a difference that measures nothing.

There is a household version of this that most households have lived through. Somebody at a wedding says the caterer went up by forty per cent. The whole table reacts. Then it turns out the comparison is against a wedding in the same household eleven years ago. Forty per cent across eleven years and forty per cent across one year are not remotely the same statement, and the reaction quietly deflates. The number never changed. The change was that somebody finally supplied the window.

The Girnar Large Cap Equity Fund figure used throughout this guide covers one year. The window is written as one year every time the figure appears, including inside the drawings. Which windows must be published, and how long each has to run, are matters the Securities and Exchange Board of India (SEBI) settles, and where that stands today is published at sebi.gov.in.

Why does a scheme figure arrive net, and net of what exactly?

Net of the scheme's own running charge, and the reason is mechanical rather than conventional. The value per unit is struck from net assets after that day's expenses have been set against them. Do that on the first date and do it again on the second date, and the charge has been taken out of both ends before either end became a number. A return computed from those two ends is therefore a net returnA figure built from values whose charges had already gone, leaving nothing further to deduct. before anybody decides to call it one.

Establish the size of what is sitting inside, rather than asserting that something is. A ratio of 1.65 per cent, struck on net assets, runs for a year on the equity scheme. Against net assets of Rs 4,200 crore that comes to Rs 69.30 crore for the twelve months, or Rs 69,30,00,000/- spelled all the way out. Spread across a 365 day year that is about Rs 0.1899 crore a day, or roughly Rs 18,98,630/- daily; the exact daily figure is Rs 693/3650 crore, and the rounded version is shown purely because a reader can carry it around more easily. Across 120.00 crore units, one day of it is about Rs 0.001582 a unit, exactly Rs 231/146000.

Arrive at the same per unit number by a second path and it comes out identical. Splitting 1.65 per cent across 365 days gives 33/7300 of a per cent for one day, roughly 0.0045205 per cent. Applied to a unit worth Rs 35.00 that gives Rs 231/146000 once more. The two routes are one expression written in a different order, so they cannot possibly disagree. The first divides the total by the units at the end, the second divides by the units at the start, and dividing at a different moment does not change a product. So the second route is not a check but the same sum twice.

Here is a check that genuinely can fail, and it fails. Take that daily amount and let it run for 365 days on a base that shrinks a little each day, the way a real accrual does, and the year adds up to about 1.6365 per cent rather than 1.65 per cent. The residue is minus 0.0135 percentage points and it does not cancel. Nothing about that is a fault in the figures. The residue is the first sight of something that returns near the end: a charge that accrues daily does not undo itself by simple addition or simple subtraction.

The mistake here is quiet and reasonable looking, and that is why it survives so long. A reader who has been taught that fees matter goes looking for the place to apply them, finds a published return, and subtracts. Out comes 11.75 per cent, and that figure describes nothing whatsoever. The return already was after fees, so 11.75 per cent is not the return after fees. It is the return after fees twice.

Why does the measure a scheme is compared against carry no costs?

Because there is nothing there to charge. A benchmark indexA stated yardstick a scheme puts itself alongside. Being a calculation and not a holding, it cannot be bought and costs nothing to keep. is a calculation, not a holding. No registrar keeps folios for it, no custodian settles anything for it, nobody manages it in the sense a portfolio is managed, and above all nobody can buy it and then be charged for having done so. The figure it produces is therefore a gross returnA figure from which nothing has been removed at all. An index number is the plainest case, and nothing was ever there to remove. in the strongest sense available: not a figure with the costs added back, but a figure that never had any.

The absence of cost is not a criticism of index figures and not a flaw in how they are built. An index is a calculation, and the consequence is that every scheme against benchmark gap ever printed sets a net figure beside a costless one. It works much like two quoted prices for the same wedding hall. One quote includes the decorator, the generator and the cleaning staff. The other quote is what the hall itself would cost if nobody had to be paid to run it. The second number is not dishonest. The bare hall rate is just not a price anybody can pay.

For the equity scheme this shows up on one line of an ordinary document. For one year the scheme came to 13.4 per cent, net. Over that identical stretch, the measure it names for comparison came to 12.1 per cent, and no cost of any description sits inside that. Both numbers are correct. Both cover the same length of window. Neither is on the other one's basis, and the document that prints them together typically says nothing about that at all.

What actually happens when the two get set beside each other?

The result is a difference that looks like a comparison and is not one. Taking the costless figure away from the net one leaves 1.3 points. One of the two numbers had a year of charges removed before it existed and the other had nothing removed at any point. The subtraction is arithmetically correct, and the result compares nothing.

Put both sides on one footing and the size of the distortion becomes visible. If the equity scheme figure is lifted to what it would have read with no charge inside it, the difference against a costless 12.1 per cent stops being 1.3 points and lands somewhere near three. The mismatch is not a rounding nuisance at the edge of the answer. The mismatch is worth more than the headline gap itself, and the basis therefore belongs with the mechanism rather than in small print.

The word near in that sentence is load bearing, and the section below the simulation explains why it cannot be sharpened. Depending on how a charge that accrued day by day is undone, the like for like difference lands anywhere between about 2.95 points and about 3.20 points. A spread of about a quarter of a point is not sloppiness. The spread is the honest width of the answer, and a figure printed to two decimals would be inventing a precision that the arithmetic does not contain.

One more check belongs here, and it comes back unresolved. Leaving it open is the correct outcome rather than a gap. Income the scheme collected is part of what a scheme figure records. An index, though, may be built from price movement by itself, or built with income put back in, and the two conventions land on different numbers for one and the same year. Which convention the equity scheme stated measure uses is not on record. So the check stands open, and no figure above depends on which convention the index uses.

What does that mismatch look like in motion?

On paper the pair look the same, so a reader can be told twenty times over that one number is net while the other holds no cost whatever, and the subtraction still gets made. So the mismatch is worth a shape instead of a sentence. The published figure of 13.4 per cent, net, over one year stays exactly where it is and never moves. The costless marker at 12.1 per cent over the same year never moves either. The only thing that moves is the charge assumed to be sitting inside the published figure, and the shaded band is the size of the mismatch that assumption creates.

The vanishing point is the only setting where the two bases actually coincide, and that makes it the most useful one on the control. Everywhere else the band has a width, and its width is precisely the charge that was assumed. The band is a picture of a mismatch between two ways of stating a figure, and it is not a picture of performance, of skill, or of anything a scheme did. Nothing in it moves because the scheme moved. The band moves because the assumption about the charge moved.

How much evidence is one scheme over one year, really?

Saying the count out loud is what breaks the spell. A single scheme across a single year gives exactly one observationA single measurement of a single thing over a single stretch of time. One of them is a data point, not a pattern.. Not a track record, not a tendency, not a range. One.

One number turns up dressed in the full visual authority of evidence while carrying hardly any of its weight, and the least expensive habit on offer anywhere here is speaking the phrase one scheme, one year within the same breath as the number itself. The phrase costs four words. The phrase is not modesty and not hedging, but an accurate description of how much was handed over.

The same reasoning is applied correctly in ordinary life and then dropped around percentages. One vegetable seller in one market on one Tuesday charged sixty rupees a kilo. Nobody would call that the price of the vegetable. Other sellers, other days, and some idea of whether that Tuesday was unusual would all be wanted first. The number itself was perfectly real. Sixty rupees was simply one number, and the honest description of one number is one number.

What could this figure not possibly have known?

Anything at all about the year that follows it. Look at how the figure was assembled and the reason is not mysterious. Two values per unit were taken, one at each of two dates, and one was compared with the other. Both of those numbers already existed at the moment the return was computed. No arrangement of two numbers from the past reaches forward past the later of the two, so a completed return contains no information whatever about the period ahead of it.

The shape of that claim is narrower and harder than the usual warning. The claim is not that the future is unknowable in principle, and not that the figure is unreliable, and certainly not that the figure is wrong. The claim is purely arithmetical. The inputs to that division were two dated values, both already past, and nothing about dividing one by the other produces a fact about a date that has not arrived.

The same is true of an electricity bill. Last month a household used a certain number of units and the bill records it exactly. The bill is completely accurate, and it is not a prediction of next month. Nobody expected a record of one month to be an announcement about the next, so nobody feels cheated. A return figure is the same kind of document. Only the percentage sign makes a return feel like a forecast.

What does the figure genuinely support, then?

Quite a lot, and refusing is only half of what a return figure is good for. Start with the plain thing: the figure supports a description. Over the stated year, on a net basis, the equity scheme moved by 13.4 per cent. The sentence is true, checkable against the scheme's own books, and exactly the sort of statement that belongs in a record of what happened.

Then comes the part that is actually useful. The figure also supports a why. Converting a percentage into a question instead of into a verdict is, in the end, the only skill on offer here. What was the scheme holding through that year, and how did that differ from the measure it is stated against? Was the difference wide in a few weeks and absent the rest of the time, or steady? Did anything about how the scheme was run change during the year? The questions the number earns are all answered somewhere other than in the number.

Where support stops is equally definite, so it is worth writing the boundary explicitly. The figure does not support a statement about the coming year. The figure does not support a ranking against any figure whose basis was never stated. Better needs a purpose and a horizon and a set of costs, none of which is inside a single percentage, so the figure does not support the sentence this scheme is better than that one. And the figure does not support any instruction. Whether a holder should act is a question about that holder and belongs to wealth and advice rather than to reading a document.

What would a comparison on one single basis require?

One of two things, and only one of them is actually available. The first would be the equity scheme figure with no charge inside it, letting a costless number meet a costless number. A costless equity figure cannot be had exactly, and the add-back is taken up below. The second is an index in a form a person could actually buy. Such a form carries charges, and a net number can then meet another net number. Option two is no thought experiment. A tracking scheme exists, it publishes its own net figure, and putting one net figure beside another net figure is the only honest move on the table.

Girnar Asset Management Limited runs one. Its Girnar Broad Market Index Fund tracks a broad index, and across the stated year that index came to 12.40 per cent, carrying nothing by way of cost, as index figures never do. On net assets the scheme takes 0.20 per cent a year. A tracker shadowing that index flawlessly while taking the same amount would therefore have shown 12.20 per cent, net, across those twelve months. The Girnar Broad Market Index Fund actually returned 12.12 per cent, net, over that year. The tracking differenceHow far a tracking scheme figure sits from the index behind it, both measured across identical dates. is therefore minus 0.28 points.

Take the decomposition one step and then stop. The rest is covered separately. Of those 0.28 points, 0.20 is the charge. The remaining 0.08 is everything else, and it is left named rather than opened up here. The residue does not cancel. To say only that a tracker slips by the amount it charges is to stop one step short: charges account for the bulk of that slippage and, on these figures, plainly not the whole of it. A check row carrying its own sign is how the leftover gets shown instead of glossed over.

Held side by side, the two subtractions look remarkably alike. The mismatched one gives 1.3 points. The honest one gives 1.28 points. The two numbers are almost the same size, and shown only the answers, nobody would guess that one of them is a real comparison and the other is not. The near identity is worth dwelling on. The size of a difference tells nothing at all about whether the difference was legitimate to compute.

Can the costless equivalent be recovered exactly from a net figure?

No, and the reason is the same daily accrual that made the figure net in the first place. An add-backPutting a charge back onto a figure that already had it taken out, in order to see what the figure would have been without it. looks like a one line correction and is not one. The charge did not come off in one amount but in 365 small daily amounts, each of them measured against a base that had already been reduced by the amounts before it and that was moving for its own reasons at the same time.

Worked three ways, the answers refuse to meet. Adding the ratio straight back gives about 15.05 per cent. Undoing the accrual the way it actually happened, day by day, lands near 15.29 per cent. Dividing by what survived the year of charging lands near 15.30 per cent. The three answers spread across about a quarter of a percentage point. The disagreement is wide enough that quoting any one of them as the equity scheme costless figure would be inventing precision the record does not contain.

So only one honest course remains. The figure used is 13.4 per cent, net, over one year. That one is a record rather than a reconstruction. About 15.05 per cent appears solely as an approximate adding back and is called approximate on every appearance. Surviving all three routes is not a number but a finding: put both sides of the equity scheme comparison on one basis and the honest gap sits somewhere near three points rather than at the 1.3 points the document shows.

What does the whole annotation look like run over one record?

Here is the whole discipline applied end to end, with every figure annotated as it goes past. Read the middle column as the arithmetic and the right column as what came out of it, and notice that no line carries a bare number.

The last line is one year, one scheme, on one basis. A completed interval has no reach past its own second date, so the line settles nothing at all about what either way of running a scheme delivers in general, and nothing at all about the year that follows. It is a description, correctly annotated, of something that has already happened.

One caution has to be carried rather than resolved, and it is about measuring sticks and not about numbers. The measure the equity scheme states it is compared against returned 12.1 per cent over the year. Across that same stretch, the broad index behind the tracking scheme came to 12.40 per cent. The two are different indices: one is a stated measure for a scheme marketed on large capitalisation, the other is a broad market index. The record never said they were the same thing, so which stick each figure belongs to is named every time, and the two are never run together. The subject each index measures is what holds them apart, so the distinction holds even though the pair of numbers now look obviously distinct. If a pair of indices happened to print the same number, that would still leave two of them; and printing different numbers is not what establishes that there are two.

Girnar Asset Management Limited, the Girnar Large Cap Equity Fund, the Girnar Broad Market Index Fund, Kalyani Bhagat and Sohail Merchant are invented.
Educational material. Not advice on any investment, tax, budget or market position.