How to distinguish a Fund Fact From a Valuation Assumption

Ask five questions of any fund figure. Did cash actually move? If not, is the number a computation the contract fixes, or is it somebody's estimate of value? Which denominator is it divided by? Which date does it accrete to? And what would change it? Three kinds of number come out, and a figure mixing two of them is the one that misleads.

A private fund's report opened anywhere shows a column of rupee figures. The figures are set in one typeface. The figures line up on one right margin. Each carries the same three letters in front of it. And they are not the same kind of statement at all. Some of them describe something that happened: money left one account and arrived in another, on a date, and a bank statement somewhere records it. Others describe something that somebody currently believes: a position nobody has sold, measured by a person, on a date, under a method. Nothing printed beside a figure says which kind it is, and telling the kinds apart takes a procedure.

Why is this the first skill and not a refinement?

Consider first a case away from funds entirely. A household has been running for twenty years on one salary. Two numbers describe its position. The first is what landed in the bank last month, and the statement could be printed and the line pointed at. The second is the flat's value, and that figure sits nowhere except in the heads of the people who live in it and the neighbours who have opinions. Asked for it, they will give a figure to the lakh. The figure will sound exactly as firm as the salary. The two are not the same kind of thing, and everybody in that household knows it the moment somebody actually has to sell.

A private fund is that household at a much larger scale, writing quarterly. The fund reports what it has drawn from its investors and what it has paid back to them, and those are movements. The same report gives what the positions it has not sold are worth, and that is a measurement. Both print identically. The two sit in the same column and the document is not going to help, so a reader who cannot separate them cannot read a fund document at all.

Two cautions. Mixing the kinds in one column is not an accusation against anybody. Nobody is hiding anything by putting a measured figure next to a received one, and there is no version of a report from a fund holding unsold positions that could avoid doing it. And this is not valuation. How an unsold holding is measured, who performs that measurement and what the word independent buys are a separate subject, covered on their own. The order to ask the questions in is the subject below, along with what each answer rules out.

Question 1: did cash actually move, and between whom?

The first question is the cheapest and it does most of the work. The answer is physical. Either an amount left one bank account and arrived in another on a date, and somewhere there is a record of it that two parties both recognise, or it did not. There is no middle position and no judgement involved. A street vendor who takes Rs 3,400 in the tin by evening does not need a method to know it: the notes are in the tin.

Take Nilgiri Growth Partners Fund II, invented, at its record date, the end of its Year 9 Q2. The fund has drawn Rs 4,80,00,00,000 from its investors across seventeen capital calls, and has paid back Rs 4,38,00,00,000 across four distributions. Of the money drawn, Rs 4,00,00,00,000 went into nine companies, Rs 70,20,00,000 was taken as management fee and Rs 9,80,00,000 covered fund expenses, and those three add to the Rs 4,80,00,00,000 exactly. One holding, Sahyadri Diagnostics Private Limited, invented, was sold in the fund's Year 7 Q2 for Rs 2,03,00,00,000, and somebody wrote that cheque. Every one of those figures is recorded cashA figure that corresponds to money that actually moved between two parties., meaning a bank statement could in principle be put beside it and pointed at.

The first question settles a great deal at once, and it takes very little thinking. Anything traceable to a movement is done. Nobody measured it at all, so there is no need to ask who did, or under what method, or whether they were being careful. It happened. The question goes first because it is nearly free, and on most fund documents it disposes of the majority of the column before anything harder has to be considered.

Question 2: if no cash moved, is it a computation the contract fixes?

Most readers carry a two-box model into a fund document. Either a number is money, or it is somebody's opinion. The two-box model is wrong, and it is wrong in a way that costs real understanding: a large and important set of figures is neither.

Here is the everyday version. Money sits in a fixed deposit that has eleven months left to run. The bank has not paid the interest. The interest cannot be spent. But the amount that has accrued so far is not anybody's opinion either: the rate is in the contract, the convention for compounding it is in the contract, and the dates are known. Two careful people with the same document will produce the same figure to the rupee. The accrued amount is unpaid and exact at the same time, and nothing in a two-box model has room for that.

A contractual computationA figure fixed by applying a contracted rate and convention to known dates. is a figure produced by applying a contracted rate and convention to known dates, so it is exact without being money. Nobody exercised judgement to arrive at it and nobody has received it. The test for whether a figure belongs here is to ask what rate, what convention and what dates produced it. If those three things answer the question completely, and nothing about the world outside the contract had to be assessed, the figure sits in this box.

Why are two kinds not enough?

Put the three kinds side by side on one fund on one day and the gap in the two-box model becomes obvious. Nilgiri Growth Partners Fund II, invented, at its record date, reports Rs 4,38,00,00,000 that has been paid to its investors in four distributions. The same fund carries an accrued preferred return of Rs 2,29,98,27,451, of which not one rupee has been paid to anybody. And it carries Rs 2,82,00,00,000 of residual valueThe total of the carrying values of everything not yet sold. across five positions, being the sum of what those positions are currently measured at.

Try to sort those three into two boxes and the middle one will not go. The accrued figure is plainly not cash: the fund has paid none of it, and because every rupee it has ever distributed went to returning capital first, none of it has ever been reduced by a payment either. But it is equally plainly not an estimate of valueA figure produced by judgement about what something would fetch.. Nobody looked at a business and formed a view. The rate is 8.0 per cent a year compounded annually, the convention for what it accrues on is written into the fund's own documents, and the dates of every call and every distribution are known. Given those three inputs the figure is arithmetic, and two careful people would produce the same rupees.

The practical consequence is a mistake in each direction, and both are common. A reader who has only two boxes will treat that accrued figure as soft on the ground that nobody has paid it, and quietly discount it. Nothing about the figure is uncertain, so the discount is wrong. The same reader may instead treat the figure as money the fund is holding, wrong for the opposite reason: the accrual is a claim the contract has built up, not a balance sitting anywhere.

Question 3: what is the denominator, and who chose it?

The first two questions ask what a figure is made of. The third asks what it is being compared against, and it only applies to ratios. The denominator is the end that gets left unnamed more often than any other, and the reason is simple: a ratio is a division, and a division has two ends, but people quote only one.

Two analysts describe the same invented fund on the same day. One says it is at 1.80 times. The other says 1.50 times. Neither is careless and neither is wrong. Nilgiri Growth Partners Fund II, invented, has total value of Rs 7,20,00,00,000 at its record date. Divided by the Rs 4,00,00,00,000 that went into the nine companies, it gives 1.80 times. Divided by the Rs 4,80,00,00,000 that investors actually paid into the fund, it gives 1.50 times. Divided by the Rs 5,00,00,00,000 that was promised at the start, it gives 1.44 times. Three true figures, one fund, one day, and the entire difference is which denominatorThe figure a ratio divides by, which decides what the ratio is answering. somebody chose.

The gap between the first two is Rs 80,00,00,000, being the management fee of Rs 70,20,00,000 and fund expenses of Rs 9,80,00,000 that were drawn on top of the money that bought companies. The fee and the expenses are not a small thing to leave unnamed. So the second half of question 3 matters as much as the first: somebody chose that denominator, and a denominator is a choice about which question the ratio answers. How each of these multiples is built, and what each one is for, is a separate subject covered on its own. Only the discipline of asking belongs at this stage.

Question 4: what date does the figure accrete or apply to?

Every figure on a fund document is true as at a date, and about half of them stop being true the moment that date passes. A printed figure looks permanent, so the date is easy to say and surprisingly hard to hold on to.

Both of the other kinds carry a date, and they carry it differently. A computed figure has an accretion dateThe date to which a computed balance has been rolled forward., meaning the date it has been rolled forward to. The accrued preferred return of Nilgiri Growth Partners Fund II, invented, is Rs 2,29,98,27,451 as at its record date at the end of Year 9 Q2, and at the fund's last capital call a quarter earlier it was a different number. Nothing happened in between except time passing, and time passing is exactly what moves a computed figure. A measured figure has a measurement date instead, and between measurement dates it does not move at all, however much the world does.

A figure that has not been remeasured looks exactly like a figure that has just been confirmed, and the measurement date is where the reading error lives. The strongest demonstration in this invented fund's own record runs in both directions at once, and it is worth walking through slowly.

Holding 3 is Tungabhadra Logistics Private Limited, invented, and it cost the fund Rs 60,00,00,000. At the end of the fund's Year 7 it was carried at Rs 1,20,00,00,000, or 2.00 times cost. In Year 8 Q1 it was sold for total proceeds of Rs 1,50,00,00,000, or 2.50 times cost. So the reported value of that position moved half a turn, being Rs 30,00,00,000, on the day the position was sold. Nothing about the business changed on that day. The kind of number changed: an estimate stopped being an estimate and became a receipt.

Holding 6 is Vaigai Edutech Private Limited, invented, and it cost Rs 30,00,00,000 and ran the other way. The holding was carried at cost, at 1.00 times, all the way through the fund's Year 5, and was written to 0.70 times, being Rs 21,00,00,000, at Year 6. The record of this invented fund notes that the deterioration in that business had been visible for three quarters before the reported figure moved. Neither case shows whether a measurement was right. The cause of a gap between a measurement date and the event it should reflect is a separate subject, covered on its own.

Question 5: what would change this figure, and who would have to act?

The last question turns a classification into understanding, and it does so by forcing a mechanism to be named rather than a category. Of any figure: what would have to happen for it to be a different number next quarter, and who would have to do it?

For recorded cash the answer is always the same and always short: another payment. The Rs 4,38,00,00,000 that Nilgiri Growth Partners Fund II, invented, has distributed does not become Rs 4,50,00,00,000 because anybody thinks differently about anything. The distributed total becomes larger when another rupee is actually paid, and until that happens it sits exactly where it is. For a contractual computation the answer is time and cash: the accrued figure grows because the calendar moves, and it falls when a payment reaches it under the contract's own order.

For an estimate the answer is that somebody has to measure it again. The figure changes when a person acts and not when the world does. Holding 6 of this invented fund has a carrying valueThe estimate at which an unrealised holding sits on the fund's books. of Rs 21,00,00,000 at the record date. Only two things would move that figure: a new measurement by whoever performs it, or a sale to a buyer. The business could have a very good quarter or a very bad one and the printed figure would not move until one of those two things happened. Who performs that measurement, on what timetable and under what method is a separate subject with its own treatment. Question 5 stops at naming that somebody has to act.

What happens when one figure mixes the kinds?

Everything so far has assumed a figure is one kind or another. The most misleading figures on a fund document are not like that. The misleading ones are ratios whose top and bottom are different kinds of number, and they carry no mark saying so.

Here is the important one. Nilgiri Growth Partners Fund II, invented, reports total value to paid in of 1.50 times at its record date. Look at what is inside that. The numerator is Rs 7,20,00,00,000, and it is made of Rs 4,38,00,00,000 of cash actually distributed plus Rs 2,82,00,00,000 of measured value across five positions. The denominator is Rs 4,80,00,00,000, and every rupee of it is cash that investors really sent. So the top of that ratio is part receipt and part measurement, the bottom is entirely receipt, and the printed figure of 1.50 times says nothing about the mixture at all. That is a mixed figureA ratio whose numerator and denominator are different kinds of number..

The arithmetic that takes it apart is one line and it is worth committing to memory. The cash distributed divided by the capital paid in gives 0.9125. The measured value divided by the same capital paid in gives 0.5875. The two parts add to 1.5000 exactly, and that is the check: two parts failing to sum to the headline mean one of them is built on the wrong denominator. And of the Rs 7,20,00,00,000 numerator, Rs 2,82,00,00,000 is measured rather than received, a share of 39.2 per cent. Nearly two fifths of this fund's reported worth rests on measurement.

Notice what the red border in that figure is doing and what it is not. The border marks a sentence a reader might say, not a figure on the fund's books. There is nothing wrong with the Rs 2,82,00,00,000. The measurement of five positions was properly produced, made by somebody, on a date, under a method. An estimate is not a failure and a measurement is not a lie, and a reader who ends up distrusting every measured figure has learned the wrong thing. The wrong sentence is only the one saying a fund has returned something it has not yet paid.

Which figures on one statement are which?

Procedures are easy to agree with and hard to apply, so run this one over a real object. Investor 1 of Nilgiri Growth Partners Fund II, invented, is a domestic life insurance company that promised Rs 1,00,00,00,000, being 20.0 per cent of the fund. Because every investor in this fund is drawn strictly in proportion, its own figures are simply the fund's figures at one fifth. Six lines arrive on its statement each quarter, they print in one column, and they are three different kinds of number plus one mixture.

Work the six out for yourself before reading the sort. Commitment Rs 1,00,00,00,000. Capital contributed Rs 96,00,00,000, being 20.0 per cent of the Rs 4,80,00,00,000 the fund has drawn. Distributions received Rs 87,60,00,000, being 20.0 per cent of Rs 4,38,00,00,000. Unfunded commitment Rs 4,00,00,000, being Rs 1,00,00,00,000 less Rs 96,00,00,000 and this investor's own figure rather than the fund's. Share of residual value Rs 56,40,00,000. And total value Rs 1,44,00,00,000, being 1.50 times the Rs 96,00,00,000 paid in, the same multiple as the fund's own.

One line on that statement deserves a note of its own, and it is the line most readers skip. Unfunded commitment of Rs 4,00,00,000 is this investor's own figure, being its Rs 1,00,00,00,000 promise less the Rs 96,00,00,000 it has actually sent. The fund's unfunded commitment is a different number entirely, being Rs 20,00,00,000, and confusing an investor-level figure with a fund-level one is the most persistent arithmetic fault in this whole subject. Both are correct at their own level, which is exactly why the mistake survives so many readings. Question 3 catches it: name the denominator, and here name whose figure it is.

Where does every figure of this fund go when it is sorted?

Run the sort across the whole invented fund and something useful falls out that no single statement line shows. Nothing is left over, and one entire column sits outside the fund's reported value altogether.

Look at what the middle column does. Rs 2,29,98,27,451 of accrued preferred return is a real, exact, contracted figure, and it appears in none of the multiples this fund reports. The accrual is not in the Rs 7,20,00,00,000 of total value and not in the Rs 4,80,00,00,000 paid in. A figure can be both perfectly precise and entirely absent from the headline, and a reader who thinks a fund report is a complete list of its numbers will never go looking for it. How that accrual is computed, and what it does when the fund eventually pays out, is covered under the preferred return, and it appears here only as a specimen of the second kind.

What does each kind license, and what does it not?

Classification without consequence is just tidiness, and this is where the procedure pays for itself. Each kind licenses a different set of sentences, and the licence is narrower than most readers assume.

The estimate row is the one to sit with: its two refusals pull against each other and readers usually only notice one of them. A measured figure may not be called money. Nor may it be called wrong, or inflated, or unreliable, because the reader has measured nothing, and the fact that a figure is an estimate says nothing whatever about whether it is a good one. Both of those are conclusions, and the only thing the classification licenses is a description of what kind of statement is in front of the reader.

One hour with a private fund's report: what is worth doing?

This is the practical end, and it matters because more people read these documents than ever put money into one. Somebody at an insurance company checks a quarterly statement against its own books. Somebody in a consulting team is handed four reports and asked what they say. Somebody drafting a committee note has to write two paragraphs that will be read by people who will not open the appendix. All three are doing exactly this procedure, whether or not they have a name for it.

The first pass takes a pencil and nothing else. A mark goes beside every figure in the numeric column that corresponds to a movement: on most reports that means contributed capital, distributions and any realised proceeds. The first pass is fast and it settles most of the report. The second pass takes the unmarked figures and asks what would have produced each: a contracted rate applied to known dates, or somebody's measurement. By the end of two passes the parts of the report that describe the past can be told from the parts that describe a current opinion, and almost nobody handed that report will have done it.

Then comes the one piece of arithmetic that this whole procedure exists to enable. The headline multiple splits in two: what has been distributed over what was paid in, and what is still carried over what was paid in. For Nilgiri Growth Partners Fund II, invented, that is 0.9125 and 0.5875 against a headline of 1.50 times, and the two parts must add to the headline exactly or something is wrong with the denominators. Both numbers belong in the committee note rather than the one. A sentence saying a fund has returned 0.91 times and estimates the rest at 0.59 times is longer, less satisfying and completely defensible. Most people making the decision would take that trade if anybody offered it to them.

One caution about how far this travels. The procedure classifies figures; it does not rank funds, it does not say which report is better prepared, and it says nothing at all about what any position will eventually be sold for. Two reports can carry very different estimated shares for reasons that have nothing to do with either manager, starting with how far each fund is through its own life. The procedure stops at naming what is being read.