How to Trace a Fund Transaction to Its Statement Line

To trace a transaction is to walk one instruction forward until it lands as a line in the holding record. The folio, scheme and plan are settled first. Then the instruction itself and the note confirming it reached the scheme are retrieved, along with the value per unit and the date attached to it. The trace itself is a division, a comparison, and a test of every amount taken out against the quantity it was struck on.

Somebody gave an instruction. Money left a bank account, or units left a holding, and some while later a line appeared in a statement. The line looks plausible. Whether it is right is another matter, and the holder cannot say why. The gap between plausible and checked is closed by a fixed run of checks that anybody can do at a kitchen table with two documents and a calculator. A traceWalking a single instruction forward through the papers it generated, testing each figure against the one before it. is not a feeling that the number looks about right; it is a short sequence of comparisons, each one of which can only be made after the one before it.

Girnar Asset Management Limited, an invented asset manager, runs the Girnar Large Cap Equity Fund, an open ended equity scheme with net assets of Rs 4,200 crore against 120.00 crore units in issue. The division is worth doing rather than accepting: Rs 4,200 crore divided by 120.00 crore units is Rs 35.00 a unit exactly. Kalyani Bhagat manages the portfolio and Sohail Merchant heads operations. A trace runs on records rather than on people, so neither of them appears in one.

Three things are settled elsewhere. The nature of a unit, and what a subscription and a redemption do to one, is covered at the start of this sequence. How the value per unit is struck out of the scheme books is covered under pricing. Which day of value attaches to an instruction is decided by conditions the Securities and Exchange Board of India (SEBI) lays down rather than by anybody at the asset manager. And the three decimal convention on unit counts, with the tiny residue it leaves behind, is worked through in the transaction calculator. All three are used in what follows.

What does it actually mean to trace a fund transaction?

Tracing means taking one instruction and following it forward, artefact by artefact, until it reaches the line it produced in the holder record, checking at each handover that the figure carried forward is the figure that was carried in. A trace is not an audit of the scheme. A trace forms no view on whether the investment was sensible. A trace answers one narrow question: does the line in the record follow from the instruction given, at the value per unit that instruction was considered against?

A courier parcel has the identical shape, and most people have already done this trace without calling it one. A parcel is handed over and a slip comes back. The slip says what was handed over, not what arrived. At the far end somebody signed for a box. Checking the delivery means matching three things: what was handed over, what the receiving record says arrived, and whether the two describe the same object. If they do not, the useful question is never a vague one about the parcel being wrong; it is which of the three records disagrees with the other two. A trace is exactly that discipline applied to money and units, and its whole value is that it turns an argument about a difference into a statement about which record is out of step.

One warning decides what a successful trace is allowed to claim, so it belongs at the front rather than the back. If every figure ties, what has been established is that the arithmetic is internally consistent. Whether the right day of value attached to the instruction is a separate question. SEBI settles it, through conditions published at sebi.gov.in, and no quantity of division on the holder's side can confirm or deny it. The two questions feel like one and are not, and keeping them apart is what makes the rest of the method work.

What has to be pinned down before a single document is opened?

Four things, and all four settled before a single document is opened. One instruction. One folio. One plan. One scheme. The moment any of the four is allowed to drift, every comparison further down becomes meaningless without announcing itself as meaningless, so all four belong at the top of the working sheet.

Here is what drifting looks like in practice, and it is embarrassingly easy to do. Somebody holds the Girnar Large Cap Equity Fund in two places: a holding built up over years in one folio, and a newer holding in a second folio opened through a different route. The holder adds up units across both, compares the total with one instruction they remember giving, and finds a difference. There is no error anywhere in that picture. There are simply two folios and one memory. The same trap has a second shape, the two plans of one scheme. The Girnar Large Cap Equity Fund runs at an expense ratio of 0.85 per cent on one plan and 1.65 per cent on the other. The holdings behind the two plans are identical. A different charge runs against each, so their values per unit are not. A unit count checked against the wrong plan value per unit will not tie, and nothing is wrong.

A comparison that spans two folios, two plans, two schemes or two instructions is not a trace at all, and the great majority of reconciliations that go nowhere went nowhere because two of something were quietly being added together. The fix costs nothing. Name the four, in writing, before the first division.

What do the instruction and the arrival note each record?

Two papers now come out of the drawer. Each carries less than it looks as though it carries, so what each is for has to be stated precisely. The first is the instruction itself: the request submitted, a record of what was asked for. The second is the acknowledgementA note issued when a request reaches the scheme. The note carries a reference, confirms arrival, and is silent on units.. The acknowledgement records that the request reached the scheme and gives a reference to quote later.

Set side by side, the two papers have a definite shape. The instruction says: this folio, this scheme, this plan, this amount of Rs 1,00,000/-. The acknowledgement says: that instruction arrived, and here is a reference for it. Between them the two papers record everything about the transaction except the one figure under check, the number of units. A trace therefore needs a third input before it can begin.

The missing unit count is not an oversight in the paperwork, and the reason is worth understanding rather than resenting. At the moment the instruction is made and acknowledged, the value per unit it will be considered against has not yet been struck, so there is no unit count anywhere to record. A parcel slip cannot state the weight the receiving warehouse will book, for the same reason: the weighing has not happened. So the acknowledgement is not a weak document, it is a document about arrival, and it is authoritative about arrival and about nothing else.

Why is the value per unit two facts rather than one?

The value per unit step needs a number and a date, and a trace that fetches only the number stalls here. The number is the value per unit that the record says this transaction was considered against. The date is the day that particular value belongs to. The Girnar Large Cap Equity Fund carries net assets of Rs 4,200 crore against 120.00 crore units. The division gives Rs 35.00 a unit, and Rs 35.00 is the figure used from here on.

The question a trace asks is which struck value the instruction was considered against, so a value per unit with no date attached to it can neither confirm nor contradict anything. A number without a day cannot answer a question about days. A value fetched from a scheme record is half a fact. A value together with the day it belongs to is something that can be compared.

The step does not ask whether that value was the correct one to attach to the instruction. Which struck value applies depends on conditions laid down by SEBI, published at sebi.gov.in and repeated in the documents for the scheme. The step itself is narrow and honest: the value the record says was used, and the day it belongs to, are both written down and carried into the next check.

How is the unit count rebuilt and set beside the record?

Now the arithmetic, and it is one division. Take the amount and divide it by the value per unit. Seven does not divide a hundred thousand cleanly, so Rs 1,00,000/- divided by Rs 35.00 gives 2,857.142857 and the digits run on without ending. Carried to three decimals that is 2,857.143 units. Set that beside the recorded unitsWhatever unit figure the holder record carries for the folio. The rebuilt figure is tested against the recorded figure, never the other way round. in the statement line, which on this traced transaction read 2,857.143. The two agree, so the check passes.

The agreement came at a cost worth looking at hard. The recorded 2,857.143 units is not the exact answer to the division; it is the exact answer rounded up by precisely one seven thousandth of a unit. Multiplying the rounded figure back by Rs 35.00 gives Rs 1,00,000.005, not Rs 1,00,000/-. Half a paisa has appeared out of nowhere, and it appeared because a number that never terminates was written to three places. The half paisa is not an error, it has no owner, and it follows this transaction through every later check.

A mismatch in the third decimal and a mismatch in the whole number are two entirely different findings, and separating them is most of the skill in tracing. A third decimal difference is a rounding convention: the holder's calculator and the register rounded a non terminating figure at different moments or in different directions, and nothing is wrong with either. A difference of tens of units is a different animal altogether and belongs in the diagnostic step further down. Treating the first like the second wastes a query. Treating the second like the first loses a real question.

What has to be written beside a deduction before it can be checked?

Subscriptions rarely exercise this step, so the other direction serves better: a redemption out of the same folio. The holder cancels 1,000.000 units. At Rs 35.00 a unit that is a redemption value of Rs 35,000.00. An exit load applies. A load rate belongs to a scheme's own terms and is never a figure to carry across from a worked example, so the rate used here is an assumption, fixed at 1.00 per cent and flagged as an assumption everywhere it appears. One per cent of Rs 35,000.00 is Rs 350.00, so the proceeds are Rs 34,650.00. The Rs 350.00 is a deductionMoney removed on the way out, so that what reaches the holder is smaller than what the cancelled units were worth. and it is the kind of line a trace has to test rather than accept.

Now make the check real. Write the baseWhatever quantity a percentage was measured on. Swap the quantity and the identical rupees produce a different percentage. beside the rupees, and then write it a second time on the other quantity. Both readings are true, and somebody who has only seen one of them will one day meet the other and think an error has been made. Rs 350.00 works out at 1.00 per cent when measured on the redemption value of Rs 35,000.00. The identical Rs 350.00 works out at 1.0101 per cent when measured on the Rs 34,650.00 that actually arrived. Dividing 350 by 34,650 gives one hundred over ninety nine as a ratio. The ratio runs on as 1.010101 and is written here as 1.0101 per cent.

A deduction quoted as a percentage is untestable until the quantity behind the percentage is written down, and the same rupees carry two different correct percentages depending on whether they are struck on the redemption value or on the money received. So the working rule for this step is mechanical: a percentage never goes on the trace sheet without the rupee amount and the quantity it was struck on, all three together, in one line.

Which paisa is right when two people round in a different order?

One part of tracing is never warned about, and once it has been seen tiny differences stop reading as accusations. On the same folio, suppose the whole holding is redeemed rather than a round thousand units. The holding is 2,857.143 units, and at Rs 35.00 a unit its redemption value is Rs 1,00,000.005, carrying the half paisa that the three decimal convention created back when the units were first written down. Applied to that, the ASSUMED load of 1.00 per cent splits the arithmetic in two, depending on nothing more than when each person reaches for the rounding.

One operator subtracts exactly and rounds at the end. One per cent of Rs 1,00,000.005 is Rs 1,000.00005. Rs 1,00,000.005 less Rs 1,000.00005 is Rs 99,000.00495, and to the paisa that is Rs 99,000.00. A load is an amount of money and money is held in paise, so the other operator rounds the load to the paisa first, giving Rs 1,000.00. Rs 1,00,000.005 less Rs 1,000.00 is Rs 99,000.005, and that figure sits exactly halfway between two paise. It cannot be stated to the paisa without somebody quietly choosing a rule. Two people have followed the same instruction with the same figures and reached answers half a paisa apart, and neither of them has made an error.

Why this matters far more than half a paisa does: it fixes what a small residue means. A residue in the last place is a statement about rounding order, not about honesty. Resolving it means knowing which order the register used, and that is a question about convention rather than about arithmetic. So a trace that throws up a difference in the smallest place the record carries has found something real, and what it has found is that the difference cannot be settled from the holder's side. The difference is noted, and the trace moves to a check that can actually fail.

Does the closing unit balance follow from the opening one?

Every check so far has looked inside one transaction. The balance check looks at where the transaction landed, and it is the only check on the list that can genuinely catch something the arithmetic cannot. The rule is simple: the opening balance, plus units created or minus units cancelled, must give the closing balanceWhere a folio stands in units once a transaction has been written into it. The closing balance has to follow from the row above it. written on the same row. On the traced folio the opening balance is 2,857.143 units, 1,000.000 units were cancelled, and the closing balance should therefore read 1,857.143 units, worth Rs 65,000.005 at Rs 35.00 with the same stubborn half paisa still attached.

Now picture the failure this catches, in a form everybody has met outside finance. A household pays the electricity bill for the flat they live in and the money leaves the account correctly, the right amount to the right utility on the right connection. The connection just happens to be the one for their old flat, and they still have a login for it. Nothing about the payment is wrong. Everything about where it landed is. A transaction can be arithmetically perfect in every figure and still have been written against the wrong folio, and the balance check is the only step in this whole run that will find it.

The backwards division set out below always agrees with itself, and the balance check is worth holding up against it. A test that cannot fail says nothing when it passes. The balance check is the opposite: it draws on rows the transaction did not produce, so it has something to disagree with, and when it passes it has actually said something.

What does dividing the amount by the recorded units show?

Here is the move that makes the difference, and it is taught almost nowhere. The recomputation says 2,857.143 units. The record says 2,828.571 units. The natural response is to describe the difference of 28.572 units and report it. Describing it is the wrong move. The division runs backwards instead: the amount divided by the unit count that was actually written down. Rs 1,00,000/- divided by 2,828.571 units gives Rs 35.35354 a unit, or Rs 35.3535 to four decimals. Rs 35.3535 is the implied value per unitWhat comes out when the amount is divided by whatever unit figure the record carries. The implied value is the one that would have produced that record., and it is the difference between having a complaint and having a diagnosis.

Look at what came out. Rs 35.3535 is not a strange number. Rs 35.3535 sits about 1.0101 per cent above Rs 35.00, which is precisely what dividing a figure by 0.99 does to it. One hundred divided by ninety nine is the same ratio that turned up in the deduction step. Rs 35.3535 is an ordinary, wholly believable value per unit for a scheme sitting in that region. A clean plausible implied value points at a question about which day of value attached, and an implied value that lands on nothing recognisable points instead at the record or at the inputs fed into the check. The gap of 28.572 units said nothing. The Rs 35.3535 says which conversation to have.

One honest refinement. Multiplying 2,857.142857 by 0.99 gives 2,828.5714285. Written to three decimals that is the 2,828.571 in the record. Because that recorded figure is itself rounded, the implied value works out at Rs 35.353541 rather than the exact hundred over ninety nine point of Rs 35.353535. The two agree to four decimals and differ in the millionths, and that residue is the same three decimal convention showing up for the third time rather than a new phenomenon.

Now the caution that keeps this step honest, and it is a caution about what the step can and cannot do. Write the relationship out: amount equals units multiplied by value per unit. The forward check divides the amount by the value per unit and compares with the recorded units. The backwards check divides the amount by the recorded units and compares with the value per unit. The two checks are the same equation rearranged, so the ratio by which the implied value exceeds the struck value is always exactly the ratio by which the recomputed units exceed the recorded units, and the two comparisons can never disagree with each other. Run the numbers and see it: 2,857.143 divided by 2,828.571 is 1.0101163, and Rs 35.353541 divided by Rs 35.00 is 1.0101163 as well. The backwards step is not a second test. The backwards step restates the same finding in the units of its cause, and that is why it names a cause where the forward version measured only a symptom. The check that can genuinely fail on its own evidence is the closing balance, and that is why it stays on the list.

Where does each kind of problem actually go?

Three kinds of problem can read the same way in a statement line and yet need three different destinations, and sorting a given one into the right destination is the last step of the trace. An input problem means the check was fed something wrong: the other plan value per unit, the wrong folio, a mis-keyed amount, a value from a day that was never in question. Nobody else is involved. The holder corrects it and runs the trace again. A record problem means the register carries something that does not follow from its own rows, and the closing balance check is what usually surfaces it. An attachment problem means the arithmetic is fine and the implied value points at a different day.

Each has one destination. An input problem stays with the holder. Anything wrong in the register itself belongs with the registrar and transfer agentThe party keeping the holder record for a scheme. The agent writes the transactions in and holds the unit balance for every folio., which maintains the holder record and is the only party able to change what it says. An attachment problem asks which struck value applied, and that is settled by conditions coming from SEBI, whose current wording sits at sebi.gov.in and is set out again in the documents for the scheme, so it is put as a question about those conditions rather than as a complaint about a number. Putting a name to which of the three is in hand decides whether the question gets an answer or merely a receipt. The route and the interval for raising any of them are SEBI matters.

One more thing decides who the question goes to, and it has to be said plainly. When two artefacts disagree, they are not equal witnesses, and each is authoritative about exactly one thing. The holder's own note of the instruction is authoritative about what was asked for. The acknowledgement is authoritative about arrival and about nothing else. The scheme published value per unit is authoritative about the value and the day it belongs to. The holder record maintained by the registrar and transfer agent is the authoritative record of units held. Where the holder's working and that record disagree, the record stands until whoever maintains it changes it, and a distributor cannot change a record. A distributor can help assemble the papers and pass a question along, which is useful and is not the same thing as holding the register. Where the units sit in a depository account rather than in a statement of account, the record being traced against sits with the depository instead, and the two in India are Central Depository Services (India) Limited (CDSL), at cdslindia.com, and National Securities Depository Limited (NSDL), at nsdl.co.in.

What does the whole trace look like run end to end?

Here is the whole run in one place, worked on the transaction carried from the opening figures onward, with the broken variant sitting in the row where it belongs. The third column, read down, is the trace itself rather than a description of one.

Notice how little of that run is arithmetic. Two divisions, one subtraction and a comparison. The work of a trace is not calculation, it is deciding what each figure is allowed to mean, and every step above exists to stop somebody drawing a conclusion the record cannot support. The order matters for that reason, and a step skipped is worse than a step done badly: a skipped step leaves a belief standing unchecked.

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.