Sector vs Thematic Research: What Defines the Set
Sector research works on a set of companies defined by what they do now. The definition changes slowly, and it is settled outside the analyst. Thematic research works on a set defined by a claim. The analyst settles the claim and can rewrite it whenever the argument moves. The single difference in how the set gets defined produces every other difference between the two.
Build the basket both ways over the same four companies
The calculator takes the figures read off each filing, the level at which the sector is being read, and the threshold the repainting rule uses. The calculator cuts the same four companies twice, names what each cut returns, and shows how much of the theme basket is already sitting inside the sector basket.
At the settings the panel opens on, the sector cut returns three names carrying Rs 18,354 crore and the theme cut returns three names carrying Rs 19,705 crore. Two names sit in both, Nandivarman Paints Limited and Sarvani Coatings Limited, carrying Rs 16,905 crore between them. The shared weight is 92.1 per cent of the sector basket and 85.8 per cent of the theme basket, so moving from one cut to the other changes roughly one rupee in seven. Read the sector at the wider level instead, where the chemicals made for coatings sit inside it, and the overlap reaches 100 per cent: every rupee the theme rule admits is already a sector holding, and the second basket has stopped being a second basket. The collapse into a single basket is the state worth noticing, and the panel holds it for as long as it is wanted.
The panel above builds two lists and nothing else. The list is where the whole argument sits. Neither approach can say a word until somebody has decided which companies are being talked about. Deciding membership looks administrative, the sort of thing settled in five minutes and never revisited. It is not. Almost every difference set out below is that one decision working itself out.
What is sector research, on its own terms?
Sector research reads the conditions faced by a group of companies that do similar things. Understanding what is happening to all of them comes before reading any one of them. Sector research asks what the input costWhat a business pays for the materials that physically go into what it sells. Move that price and every maker using the same material feels it at once. is doing, what demand is doing, how crowded the field is, and whether the money being made is being made by everybody or by somebody in particular. Then, and only then, it looks at an individual company and asks how much of that company's record is the field it sits in and how much is the company itself.
The defining move of sector research is that it fixes the group first and reads the company second. Consider a street of ten sweet shops in the same market. Before deciding whether the third one is doing well, the question is whether the whole street had a good festive season. If sugar and ghee got cheaper for all ten of them, the third shop's better margin is a fact about sugar, not about the shopkeeper. Sector research is that instinct, written down and made checkable.
The set here is a set of makers of decorative and industrial coatings: Nandivarman Paints Limited, Sarvani Coatings Limited, and Kesaria Surface Solutions Limited. Sarvani Coatings sells both decorative and industrial, Nandivarman Paints is almost entirely decorative, and Kesaria Surface Solutions is industrial heavy, so its margin behaves differently through a cyclicalRising and falling with the wider economy instead of heading steadily one way. Such a business does well while construction and capital spending are strong, and badly once they are not. stretch. The coatings list was not assembled by anybody's argument. The list is just what those three companies make.
What is thematic research, on its own terms?
Thematic research tests a claim that some change will matter, across whatever companies the claim applies to, wherever those companies happen to sit. The claim comes first and the list comes out of it. The claim is written, then the rule that decides who is touched by it, then the rule is applied, and the set follows. The search then turns to whether the claim is showing up in the numbers, and, just as important, to what would have to be true if the claim were wrong.
The claim tested below is a repainting claim: that households are repainting sooner than they used to, and that the shortening of that gap will show up first in volumes and then in what suppliers to those makers ship. Notice that the claim reaches across a boundary before it reaches any company at all. The claim touches the people who make paint and the people who make what goes into paint. The claim is about a flow of demand, and demand does not stop at a classification.
So the theme set is Sarvani Coatings Limited and Nandivarman Paints Limited, both selling into households, plus Thottam Chemicals Limited, a supplier of resins and additives sitting one step up the chain from both of them. Kesaria Surface Solutions is not in it. Industrial coatings are sold to manufacturers on contract, and a household repainting decision does not reach them.
Thottam Chemicals Limited supplies resins and additives to coatings makers. Is Thottam Chemicals in the coatings sector set?
What actually defines the set in each?
One sentence carries the rest of the argument. A sector set is defined by activity: what a company itself makes or sells, read off its own operations, and classified by somebody other than the analyst. A theme set is defined by a claim: a written rule about who is touched by a change, composed by the analyst doing the work. Every other difference between sector research and thematic research, including the way each one fails, follows from who gets to decide membership.
Who settles the set is easy to read as a small procedural point. When somebody else settles the set, the analyst inherits a boundary they did not draw and cannot move to suit the argument. An inherited boundary is a constraint, and constraints are inconvenient. The same boundary is also a protection, and the two failures set out below are exactly what it protects against.
Who settles which companies are in a sector, and who settles which are in a theme?
A theme's membership was revised twice over three years. Can its performance be compared across those three years?
Why can one set be compared with itself over years and the other not?
A company changing what it makes is a slow, expensive, publicly visible event, so a sector set changes slowly. Two years of the same set can therefore be placed side by side and the comparison means something. The set measured did not move between the two readings. A theme set can change the moment its author refines the argument. Two years of that placed side by side may be comparing two different collections of companies while calling them by the same name.
A frozen membership is why sector data can be read as a series and theme data usually cannot, and why any theme series should be treated with suspicion until its membership is known to have been frozen. Frozen means the rule was written at the start of the period, the list it produced was recorded then, and nobody has touched it since. If nobody can confirm that, the series is not a record. The series is a description tidied up with hindsight.
Work it on the numbers. Sector revenue was Rs 43,500 crore in the earlier of the two years and Rs 48,300 crore in the later one, a rise of 11.0 per cent. Sarvani Coatings Limited sat inside that set both times, at Rs 2,120 crore then and Rs 2,415 crore now, growing 13.9 per cent while the sector managed 11.0 per cent, so it ran 2.9 points ahead. Its market shareOne company's sales as a fraction of everything sold by everybody in the same line of work. The share climbs only if that company outgrows the rest of them. therefore moved up, reading 4.87 per cent in the earlier year and 5.00 per cent in the later one, worth 0.13 of a percentage pointThe plain gap between two percentages. A share reading 4.87 per cent one year and 5.00 per cent the next has moved 0.13 of a percentage point, which is a different measure from growing 0.13 per cent..
The comparison is legitimate for one reason and one reason only: the denominator and the members are the same in both readings. Nobody rewrote the definition of the sector between the two years, so the two totals were built the same way. The record used here carries the two year share path for Sarvani Coatings and not for the other two makers.
The three makers in the sector set were the same in both years. What does that allow that a rewritten theme set does not?
How does each one fail, and why are the failures opposite?
Sector research fails by being too narrow. Sector research draws a boundary around companies doing similar things, and then a change comes along that does not respect the boundary. An input price move starts at a supplier, travels through the makers and lands on a dealer's shelf. The cause was one step outside the line, so a researcher reading only the makers sees the effect arrive with no visible cause. The set was well behaved and the world was not.
Thematic research fails in the opposite direction. Because the analyst holds the pen, the set absorbs whatever seems to fit, and a set that can absorb anything can never be contradicted. Once every company that does well is judged to have been touched by the change, and every company that does badly is judged not to have been, the claim has stopped being a claim. The claim has become a description of the past with a rule attached afterwards.
The two failures are opposite, so a researcher who guards hard against one has moved closer to the other. Tightening the sector boundary to keep the reading clean means seeing less of the chain. Loosening the theme rule so that it catches everything the change might touch costs the ability to be wrong. No setting escapes both. The honest move is to name which one the work is currently more exposed to.
A researcher guards hard against a theme becoming too loose. What are they now more exposed to?
Which question is each one good for?
Sector research answers what is happening right now to a group of companies doing similar work. Input costs, demand, crowding, who is taking share from whom. The sector question is a present tense question about conditions, and the answer is built from records that already exist. Thematic research answers whether a claimed change is real and where it would show up first. The thematic question is a question about a claim, and the answer is built by writing down what would be expected to show up and then going to look.
The two are different questions rather than two competing answers to one question. Swap them and both go quiet. Ask the sector reading whether household repainting cycles are shortening and it has no way to answer. The record it works from was not built to test a claim about the future. Ask the thematic test what is happening to coatings margins this year and it has no stable base to answer from. Its set was assembled to suit an argument.
Do sector research and thematic research answer the same question?
What does the worked set look like when it is drawn?
Both approaches are now set loose on exactly the same material. The two sets can then be watched cutting across each other. The sector is worth Rs 48,300 crore, and the table below carries each company's share of it. The three makers together hold Rs 18,354 crore, or 38.0 per cent. The other 62.0 per cent belongs to a long tail of smaller makers, none of them large enough to be named here.
Now apply the repainting rule. The repainting rule pulls in Sarvani Coatings and Nandivarman Paints, both of them selling into households. The rule also pulls in Thottam Chemicals Limited. Thottam Chemicals is not a coatings maker at all and holds none of that Rs 48,300 crore, but a change in repainting frequency reaches the resins that go into the paint. The rule drops Kesaria Surface Solutions. Industrial contract work does not move when a household decides to repaint a bedroom.
| Company | In the sector set | In the theme set | Share of the sector |
|---|---|---|---|
| Nandivarman Paints Limited | In | In | 30.0 per cent |
| Sarvani Coatings Limited | In | In | 5.00 per cent |
| Kesaria Surface Solutions Limited | In | Out | 3.0 per cent |
| Thottam Chemicals Limited | Out | In | none of it |
| In both sets, Rs 16,905 crore | 2 | 2 | 35.0 per cent |
Say the contrast plainly: two of the three sector members are in the theme, one of them is out, and a company from outside the sector entirely is in. The two members shared by both sets carry Rs 16,905 crore between them, or 35.0 per cent of the sector, and 92.1 per cent of the three makers taken together. The maker the theme drops carries 3.0 per cent of the sector and 7.9 per cent of those three. Adding 35.0 and 3.0 returns the 38.0 per cent the three makers hold. The addition is the check that the split is clean.
Which company in the worked set above is in the sector and out of the theme, and which is the reverse?
Can the two be used together, and in which order?
Yes, and the order matters. The sector reading goes first and gives a stable base measured the same way in both periods. A theme cannot supply that base for itself. The theme goes second and supplies the reach along the chain that the sector boundary cuts off. Then the useful move: running the theme's predictions against the sector record rather than against the evidence the theme itself went and collected.
Testing the theme against the sector record is the single most useful arrangement of the two, and the only version in which the theme could have been wrong. If the repainting claim is real, it predicts something specific about volumes in a record that was not assembled by anybody who wanted the claim to be true. The sector record can say no. A file of supportive evidence gathered while building the theme cannot say no. Nothing unsupportive was ever put into it.
What does this change about how the work is actually done?
An analyst covering coatings keeps two files and never merges them. The first is the sector file: the same set, the same definitions, updated every quarter, and deliberately boring. Its value is that nothing in it moves except the numbers. The second is the theme file. The theme file holds the current claim, the membership rule in the exact words it was written in, the date it was written, and a list of predictions that could fail.
A peer setThe small group of companies an analyst puts alongside the one being studied, chosen so that the comparison is fair rather than flattering. assembled for the sector file gets left alone once it is set. Leaving it alone feels wrong the first few times. The discomfort goes the first time a favoured theme is checked against that fixed set and loses. An investor reading somebody else's thematic note applies the same discipline from the outside: asking when the membership rule was written, whether it has been touched since, and what the claim predicted that has not happened yet. If none of those three has an answer, the note is a description of the past.
The discipline is simply this: whoever wrote the set does not get to grade it. The sector file, built by nobody in particular for no argument at all, is therefore the better record to hold a theme against.
The error that gets made, and what it costs
An analyst reports that a theme has performed well over three years. The membership of that theme was revised twice during those three years as the argument got refined, so the three year comparison is setting three different collections of companies beside each other and calling the sequence a track record. Nobody lied. Each revision looked reasonable on the day it was made.
The direction of the error is predictable, and that is what makes it dangerous rather than merely untidy. Revisions made after the outcomes are visible tend to remove what did badly and admit what did well. Refining an argument feels exactly like that from the inside. The mechanism is the same as survivorshipA record that only counts what is still standing at the end. The failures were dropped along the way, so the survivors make the whole group look stronger than it was., arriving through the membership rule rather than through a company disappearing.
Watch it happen on the record here. The two year gross marginWhat is left of revenue after the cost of the materials that went into the product, before salaries, selling costs and everything else. Usually written as a percentage of revenue. gain was 3.0 points at Sarvani Coatings, 2.4 points at Nandivarman Paints and 3.6 points at Kesaria Surface Solutions. The theme as written held the first two of those plus a supplier with no published margin record, so its members with a record averaged 2.7 points. Revise the rule once the outcomes are on the table and Kesaria Surface Solutions, the best gain in the whole record at 3.6 points, walks in. The average becomes 3.0 points.
Not one company's own record changed, and the reported figure still improved by 0.3 of a point, bought entirely by editing the membership. Worse, a theme revised until it holds every good outcome has quietly rebuilt the sector set, whose own average across the three makers is that same 3.0 points. The claim has stopped adding anything.
The fix is mechanical rather than moral. A theme compared across time must have its membership frozen at the start of the period and reported as it stood then, additions and removals logged with their dates and reasons. A theme series with no frozen rule behind it should be read as a description of what happened, never as a record of what the claim achieved.
Publishing a theme record in India
Reporting a set's past performance to somebody else is conduct, not just arithmetic, and conduct for a research analyst in India sits with the Securities and Exchange Board of India. Requirements, thresholds and periods move, so the current position at sebi.gov.in is what settles the matter, and it should be confirmed there before anything describing a set's track record is published.
What should a theme be tested against?
When does the distinction stop mattering?
The difference between the two decides a great deal, so it is worth being just as exact about the conditions under which it decides nothing. A reader who has been told a distinction matters tends to carry it everywhere, including into the places where the answer comes out the same either way. There are four such places here, and the calculator above produces the first one on demand.
The first is overlap. At the settings the panel opens on, the two cuts share two names carrying Rs 16,905 crore, or 85.8 per cent of the theme basket. Read the sector at the wider level, where the chemicals made for coatings sit inside it, and that figure goes to 100 per cent: every name the theme rule admits is already a sector holding. Past roughly that point the choice between the two framings is a choice of filename. The holding has not changed, and the theme's apparent reach along the chain turns out to have been an artefact of reading the sector narrowly.
The second is a theme narrow enough to sit inside one sector anyway. A repainting claim confined to companies that make decorative paint has no member the coatings sector does not already hold, so it cannot report anything the sector reading could not. A theme earns its separate existence by reaching somewhere the sector boundary cuts off. Here the reach is one supplier, Thottam Chemicals Limited, carrying Rs 2,800 crore. Removing that single name from the rule collapses the two exercises into each other.
The third is a holding period shorter than the time either thesis needs to express itself. The sector reading here took two years to show Rs 43,500 crore becoming Rs 48,300 crore, and the repainting claim expected its first evidence in volumes and only then in what suppliers ship. Hold anything for a fortnight and neither of those has had a chance to say a word, so whichever framing picked the names, whatever happened to the money came from somewhere else.
The fourth is a mandate that permits only one of the two framings. A fund that may hold only companies classified inside one sector cannot buy Thottam Chemicals Limited however good the repainting argument is. The theme then hands back the two makers the sector cut already held, with extra paperwork attached, and the question was settled before the analysis started, by a document nobody was reading as an analytical instrument.
None of the four announces its own expiry. Overlap narrows the day a member changes what it makes, or the day somebody nudges the threshold in the rule; a narrow theme widens the first time its claim is extended one step up the chain; a holding period lengthens quietly, usually after the decision that assumed a short one was already taken; and a mandate gets rewritten between one committee meeting and the next. The reading that made the distinction safe to ignore was a reading of one particular day, so the working habit is to re-run the overlap rather than to remember a verdict about it.
A theme rule is written narrowly enough that every company it admits already sits inside one sector. What has the choice between the two framings decided?
Where can the parts of this that are not invented be checked?
| Who holds it | What to look for | Site |
|---|---|---|
| Securities and Exchange Board of India | The conduct and disclosure position a research analyst works under, including what may be said about a past record | sebi.gov.in |
| National Stock Exchange of India | The filed quarterly and annual numbers for any listed member of a set being assembled | nseindia.com |
| BSE Limited, formerly the Bombay Stock Exchange | The same filings for an issuer listed there, useful for checking one already pulled elsewhere | bseindia.com |
Nandivarman Paints Limited, Sarvani Coatings Limited, Kesaria Surface Solutions Limited and Thottam Chemicals Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.