Insider Transactions: Reading the Disclosure Carefully

An insider transaction is a dealing in a company's shares by someone connected to it, reported into the public record with a quantity, a price and a date. Reading one starts with its size. A director of Sarvani Coatings Limited acquired 5,000 shares against 24.00 crore in issue, about 0.0021 per cent, or one share in every 48,000. Size that small limits what the transaction can support before any question of meaning arises.

Everything below rests on one habit: a report states a fact and never a reason. The arithmetic is what decides the whole reading. The figures belong to Sarvani Coatings Limited, an invented issuer carried through this subject area so the sums have something to bite on.

What is an insider transaction, and what does the report actually contain?

A person close enough to a listed company that their dealings must be reported buys or sells some of its shares, and the fact of that dealing goes into the public record. A dealing and a record of the dealing are the whole of it. The report carries a name, a quantity, a price, a date, a mode of acquisition and the holding before and after. Six fields, all of them facts, none of them an explanation.

The missing reason deserves a second. The whole reading turns on it. The report was never designed to carry a reason, so no amount of careful reading will extract one, and a reader who believes they have found one has supplied it themselves. A report with no room for a reason is not a defective report. A form that asked a director to state why would be asking for something unverifiable, and a market that acted on unverifiable statements of intent would be a worse market than one that acted on quantities and dates.

Here is the household version, and it is closer than it sounds. A neighbour buys a second scooter. The certain part is that a scooter was bought, on a date, at a price. The unknown part is whether a child started college, whether the old one broke, whether a cousin needed one for six months, or whether the price was simply good that week. The purchase is a fact. Everything that might be said about the reason is the observer talking, not the scooter.

How big was it, and big against what?

Size comes first, before interpretation, and it comes first for an unglamorous reason: an unsized quantity reads far larger than it is. Five thousand of anything sounds like a lot. Five thousand shares out of 24,00,00,000 does not, and the second sentence is the true one.

So the quantity goes against three different things in turn. The choice of what it is divided by is not clerical. A denominatorThe quantity a figure is divided by. Choosing it is not a clerical step: it decides which question the resulting number is an answer to. decides which question the answer is an answer to, and three sensible ones are available here.

The three denominators answer three genuinely different questions, so a reader who computes only one has sized the transaction against a question they never asked. Shares in issue answers how much of the company changed hands. The free floatThe slice of a company's shares not held by the controlling group, and so the slice that is genuinely available to change hands. answers how much of the stock that is actually available to trade changed hands. One day's average daily traded volumeThe number of shares that go through in a typical session, averaged over a stretch, used here purely as something to divide by. answers how hard the transaction was to do at all.

Neither of the last two was handed over ready made, so both deserve a word on where they came from. The free float figure is 47.6 per cent of the share count, or 11,42,40,000 shares. The record puts the promoter group at 52.4 per cent, so the two halves of the register close on 100. The traded volume figure was not published as a quantity at all. The record publishes an average daily traded value of about Rs 42 crore, and at the illustrative Rs 486/- that is 8,64,198 shares. The division is where a rounding error would hide, so divide the value by the price yourself rather than accepting a share count somebody else derived.

One quiet check costs nothing and is worth doing. The free float reading can be reached a second way, through money: Rs 24,30,000/- of stock against a free float market capitalisation of Rs 5,552 crore also gives 0.0044 per cent. The price of Rs 486/- sits on the top and the bottom of the fraction and cancels, so the two routes agree exactly. The agreement is arithmetic rather than luck. Reaching the same answer twice is worth doing anyway: it confirms that the share count and the price in use belong to the same record.

Look at where the three marks sit. Against the whole company the transaction is barely on the scale at all. Against the free float the mark sits a shade to the right, and only because the free float is smaller than the company by a factor of 1 over 0.476, or 2.10. The free float reading is therefore exactly 2.10 times the first one: one share in every 22,848 of the stock that is free to trade, against one in every 48,000 of the company. Against a day of trading it has jumped two whole steps, and this is the one that changes how the transaction feels: 0.58 per cent of a session is a quantity that goes through without anyone noticing.

How long is that in practice? If a session runs for about 375 minutes and the day's volume arrived at an even rate through it, 0.58 per cent of a session is about 2 minutes and 10 seconds. The 375 minutes is an assumption stated here as a divisor and nothing more, and current trading hours are the exchange's to publish at nseindia.com rather than mine to write down. There is a second check that does not need the assumption at all. Market depthThe quantity showing on the order book close to the current price, which is a different measurement from a whole day's volume. in this record is about 18,000 shares showing on each side within half a per cent of the price, and 5,000 shares is 27.8 per cent of that. The entire transaction would have fitted inside the order book as it stood, without moving the price. Nothing states its smallness more sharply.

What does a small acquisition support?

Something, and it is worth being precise about what. A small acquisition supports the plain fact that a person connected to the company was willing to transact in its shares, at a stated price, on a stated date. The willingness is genuine information and should not be thrown away. Somebody with a closer view of the business than an outside reader's was, on that day, a buyer rather than a seller.

The limit sits inside the same sentence: willing to transact at that size, and for a size this small that is a weak statement about willingness and no statement at all about conviction. The word conviction is the one to watch. A motive arrives wearing the clothes of a reading through exactly that word. Conviction is a state of mind. A quantity is not a state of mind. Rs 24,30,000/- of stock, against a person whose total wealth, other holdings and commitments are all invisible from outside, says nothing about how strongly anybody believes anything.

What does it not support?

A purchase this small does not support a view about the business. The purchase does not support a forecast of earnings. The purchase does not support a claim about what the buyer expects to happen. The purchase does not support a change to any number in the model. And the purchase does not support the sentence people most want to write, that the director bought because of something.

The reason is structural rather than a matter of degree: the buyer's total position, their other assets, their tax position, their household commitments and their reasons are all outside the document, so any conclusion about the buyer requires facts the document does not contain. None of that is a call for caution. The inputs for that inference are absent, and no amount of care with the inputs that are present will manufacture them.

How many different stories fit this same purchase?

At least five, and they are not ranked here because ranking them would be the same error in a quieter voice. A purchase of 5,000 shares is fully accounted for by a routine allotmentShares handed to someone under an arrangement already agreed, rather than bought on the open market by a person deciding to buy. under a compensation arrangement. The purchase is fully accounted for by a decision to hold a qualifying number of shares. The purchase is fully accounted for by an ordinary adjustment inside a personal portfolio. The purchase is fully accounted for by a transfer within a household. The purchase is fully accounted for by simple interest in the shares at a moment when dealing happened to be open.

Each of those, on its own, completely explains the observation, so finding one of them plausible narrows nothing at all, and the report selects none of them. Readers find this part hardest because plausibility feels like evidence. It is not. When five accounts each fully explain a single observation, discovering that one of them is credible leaves the reader exactly where they started, holding a fact that five stories fit.

Consider a wedding in a nearby street. The house has taken delivery of forty chairs. Forty chairs is fully explained by a wedding, by a funeral, by a chair rental business, by a school event, by a relative storing stock. The known part is that a delivery happened. To tell people there is a wedding is not to read the delivery: it is to choose one of its explanations and drop the other four without saying so.

Nine insiders instead of one. Is that different evidence?

Suppose eleven purchases by nine different insiders arrive over four months. Eleven purchases are a genuinely different observation from a single purchase, and it would be silly to pretend otherwise. Two things change, and it is worth separating them.

The first is the aggregateThe combined quantity across several separate entries, which is the one thing a run of entries genuinely adds to a single entry. size. The record carries no quantities for those eleven purchases. The honest move is to suppose, out loud, that each was the same 5,000 shares, and then size the supposition. Eleven times 5,000 is 55,000 shares, or Rs 2,67,30,000/- at Rs 486/-. Against the shares in issue that is 0.0229 per cent, so the eleven together amount to a single share out of every 4,364. Against one day of trading it is 6.36 per cent, or under twenty four minutes of an ordinary session. Eleven times bigger, and still nowhere.

The second thing that changes is that a purely individual explanation becomes less economical. One person may well have been rebalancing a portfolio. Nine people rebalancing in the same four months is a stretch, so the accounts that apply to one person at a time start to look strained, and the accounts that could apply to a group at once do not.

The kind of evidence does not change. Eleven reports that each state no reason still state no reason between them, so a pattern is stronger evidence of the same weak kind rather than evidence of a different kind. Nothing in the run of filings forecasts anything. Nothing in it says a word about the margin question this record leaves open. The run of filings leaves a bigger fact and the same silence.

It happened three weeks before the results. What does the date establish?

A dated report establishes the date. A date is checkable, keeps, and can be set beside other dated facts a year from now, so the date is worth writing down. The date does not establish what the buyer knew on that date, and there is no route from a date to a state of knowledge that does not pass through an assumption somebody made up.

There is a second question sitting behind this one, and it is the question people actually want answered: was the date inside a window when a connected person was not permitted to deal? The conditions under which an insider may deal at all, and the length and trigger of any restricted periodA stretch during which a connected person may not deal. Its length and its trigger are the regulator's to define., belong to SEBI, and not one of them is written down anywhere above. A reader who wants to know whether a particular date fell inside or outside such a window opens sebi.gov.in and reads the wording standing there that day. Not a summary of it, and certainly not a number remembered from somewhere. Rules move, and only the wording in force on the day answers the question.

Work it all the way down, and watch the number shrink

Doing the whole thing in order, once, on the director's acquisition shows how far the number falls at each step.

Every figure in the table was recomputed from the absolutes rather than carried across, and the cross-checks run as follows. The record for Sarvani Coatings Limited puts the acquisition at 0.0021 per cent of the shares in issue, and recomputed from the absolutes, 5,000 over 24,00,00,000 is 0.00208333 per cent, rounding to the record's 0.0021, so the two agree. The free float route gives 5,000 over 11,42,40,000, or 0.00437675 per cent, printed as 0.0044, and the same reading reached through money, Rs 24,30,000/- against a free float market capitalisation of Rs 5,552 crore, gives the identical 0.00437675 because the price of Rs 486/- cancels top and bottom; that agreement is arithmetic and not a coincidence. The volume route derives its own denominator: Rs 42 crore of daily traded value at Rs 486/- is 8,64,197.53 shares, so 5,000 shares is 0.578571 per cent, printed as 0.58. Market capitalisation of 24,00,00,000 shares at Rs 486/- is Rs 11,664 crore. The figure ties to the record and is the check that the share count and the price belong together. The record publishes no share count at all, so the 24.00 crore figure is an assumption and every other figure is built forward from it. Profit after tax of Rs 278 crore over earnings per share of Rs 11.58/- would give 24.0069 crore, near but not equal, and rounded per share figures were never used to set the count. The record carries no quantities for those eleven purchases, so the 55,000 shares in the pattern section are a stated supposition of eleven equal ones. The 375 minute session is a divisor stated on screen rather than a rule taken from anywhere. Thresholds move, and only the regulator's wording in force answers for them.

Then the reading, stated in the order set out above. The reading supports one thing: a person connected to Sarvani Coatings Limited bought at Rs 486/- on a stated date, at a size that would have cleared inside the order book as it stood. The reading does not support anything about revenue of Rs 2,415 crore, anything about the gross margin question this record leaves open, anything about what the buyer expects. Then the five accounts, each of which fully explains the purchase and none of which the report chooses. Then the output, one line in a file.

A research file that records the transaction and its size, and changes nothing else, has handled this disclosure correctly and completely. If that reads as an anticlimax, the anticlimax is itself the point. The pull towards doing something with a scrap of information is exactly what turns a 5,000 share purchase into a paragraph about a director's confidence in the business.

What may actually be written down?

Three things, and no more. A sized fact. A short list of accounts that each explain it. And at most one question worth adding to a research fileThe written record a researcher keeps of what was observed and what, if anything, it changed. Its value is that it can be reread later against what happened., if a question genuinely arises. On a single small purchase a question usually does not.

Three recorded lines is a far smaller output than most readers expect from research, and the smallness is itself the finding rather than a shortfall in the work. Reading a disclosure well often produces almost nothing. A researcher who cannot tolerate producing almost nothing will produce something instead, and that is how a fact about one person's holdings becomes a paragraph about a company's prospects.

Back to the street for the last time. Seeing that the owner of the stall outside one office block bought two extra sacks of rice establishes that a purchase happened. The purchase says nothing about tomorrow's menu, whether trade is improving, or whether a supplier offered a discount. A note that two sacks went in records something true and reusable. A note that business must be picking up is fiction with a real number attached to it, and the number will make the fiction look researched.