Market Capitalisation, Free Float and the Size Bands

Take every share a company has in issue, price them all at the last trade, and the total is its market capitalisation: a measure of scale at one instant. Free float capitalisation repeats that multiplication over only the shares genuinely able to change hands. In India the large, mid and small grouping is set by rule against a ranking, not by any rupee figure worth memorising.

A quoted price is a record of one agreement at one instant and carries the date it was struck. A share count is what the listing documents state, and a corporate action moves it. And a figure is meaningless until it is stated what it is a figure of. Multiply the price by the count and the result is a size at one instant, a number with two places it belongs and one place it must never be used at all.

What is market capitalisation actually measuring?

On one street there are twelve identical shops in a row. Eleven of the twelve have been held by the same households for thirty years and nobody there has any intention of selling. The twelfth changes hands one Tuesday for Rs 40 lakh. Now price the whole street. The reflex answer is Rs 4.80 crore, twelve times Rs 40 lakh, and the reflex answer is not stupid. Twelve times Rs 40 lakh is the only arithmetic available and does say something real about the scale of the row. The multiplication took the price agreed between one seller who wanted out and one buyer who wanted in, though, and applied that price to eleven properties whose holders were never at the table.

Market capitalisation does exactly this, and it does it with no apology. For Sarvani Coatings Limited, a maker of decorative paints and industrial coatings, the illustrative figures as at 28 August 2026 are 24,00,00,000 equity shares in issue and a price of Rs 486/-. Multiplying gives Rs 11,66,40,00,000. Everyone writes that as Rs 11,664 crore. The Rs 11,664 crore applies the price of one small transaction to every share in issue, including the 12,57,60,000 shares held by the promoter and promoter groupWhoever a company identifies in its filings as having founded or controlled it, together with the persons and entities connected to them. Membership is fixed by definition in the rules, never estimated by a reader. that were never for sale and could not have been bought at Rs 486/- or at any other price that day.

Market capitalisation is therefore a measure of scale rather than a price anybody could pay. The figure answers the question, roughly how big is this thing, in a unit that allows one company to be set beside another without knowing anything about either. Capitalisation does not answer the question, what would this cost, and it does not come close.

Float-Adjusted Market Cap: which shares could actually be bought?

Back to the street of twelve shops. Suppose the question is not the abstract worth of the row but the cost of buying into it. Only one shop is going, so the number that matters is not twelve shops times Rs 40 lakh but one shop times Rs 40 lakh. The one shop figure is not a worse version of the twelve shop figure but the answer to a different question, and kept apart the two cause no confusion.

Free float capitalisation is that second number. The same multiplication, share count times price, runs only over the shares genuinely available to change hands. The excluded holding is the closely held stake, most commonly the promoter and promoter group, though the exact carve out follows the rules of whoever is doing the counting. The split is not a matter of estimation. Every listed company files a shareholding patternA periodic filing in which a listed company sets out who holds its shares, grouped into categories. The pattern is published by the company through the exchanges and is a document to be read rather than a figure to be worked out. with the exchanges, and that filing is where the proportion comes from.

For Sarvani Coatings the illustrative pattern as at 28 August 2026 splits four ways: promoter and promoter group 52.4 per cent, foreign portfolio investorsOverseas investors registered to buy listed Indian securities as a financial holding rather than to run the business. A category in the shareholding filing, not a judgement about the investor. 18.2 per cent, domestic institutionsInsurers, pension money, mutual funds and banks based in India, holding for people who are not themselves on the register. Another reporting category, kept apart from individuals in the filing. 14.6 per cent, and retail and others 14.8 per cent. The last three add to 47.6 per cent, and 47.6 per cent is the free float. Run the arithmetic and Sarvani Coatings is at Rs 5,552 crore on a free float basis, against Rs 11,664 crore on a total basis, and the difference is not a rounding matter but a different answer to a different question.

Market Cap vs Free-Float Market Cap: which figure belongs where?

Here is the part that trips people, and it trips them precisely because both numbers are correct. Neither is an estimate. Neither is stale. Both come out of the same two inputs. And yet for Sarvani Coatings they are Rs 11,664 crore and Rs 5,552 crore. The gap of Rs 6,112 crore is more than half the headline. Nothing about a figure that is off by more than half can be waved through as a detail.

The rule for keeping them apart is a question, not a formula: does this use care about how big the company is, or about how much of it can be bought? Size questions take the total. The classification into large, mid and small takes the total. Almost every casual comparison in print, and almost every screen an analyst runs, takes the total. Availability questions take the float. An indexA published, rule based list of companies with a stated weight on each, used as a reference point. How one is built and rebalanced is a separate subject from what a weight is calculated on. weight takes the float. Any question of the form, how much of this could I actually accumulate, takes the float.

Quote one where the other belongs and the error is not slight, it is exactly the closely held proportion. For Sarvani Coatings the mistake means overstating availability by Rs 6,112 crore, or understating scale by the same amount, depending on which way it ran. And notice the uncomfortable detail in this case: the closely held block at Rs 6,112 crore is larger than the entire free float at Rs 5,552 crore. The part of the company that cannot be bought is bigger than the part that can.

Where does an index weight come from, worked through?

The availability side can be followed all the way to a rupee. Take an unnamed fifty company list weighted on free float capitalisation, with an aggregate free float of Rs 27,76,000 crore. Sarvani Coatings brings Rs 5,552 crore of float to that total. Its weight is Rs 5,552 crore divided by Rs 27,76,000 crore, or 0.20 per cent.

A percentage of an index is an abstraction until money moves, so make it concrete. A tracker fundA fund that sets out to hold a published list in the published proportions rather than to select among them. The fund buys what the rule tells it to buy, in the amount the rule tells it to buy. holding Rs 10,000 crore against this list would need Rs 20 crore of Sarvani Coatings. Set that against the float and it is 0.36 per cent of every share that is genuinely available. Set it instead against the headline Rs 11,664 crore and it looks like 0.17 per cent, flattering the position by pretending the closely held block is somehow in play. Free float weighting exists precisely so that a fund is never directed to buy shares that no buyer can obtain, and the discipline is worth more the more closely held a company is.

Suppose the same list weighted on total capitalisation instead. Sarvani Coatings would be brought into the weighting on Rs 11,664 crore rather than Rs 5,552 crore, or 2.10 times as much company. The resulting percentage depends on every other constituent's float as well as this one's, and a percentage reconstructed from a single ratio would look derived and be wrong. The ratio is determinate and the percentage is not.

How is the large, mid and small classification actually decided?

Now the question everybody arrives with, and the answer that disappoints. The wish is for a number: above such and such a figure a company is one thing, below it another, and Rs 11,664 crore would settle the matter on the spot. The number exists. The number belongs at the body that publishes it rather than in a reader's memory, and the reason is not caution but accuracy.

In India the size grouping is not a matter of opinion and is not derived from any single company's figures. The grouping works by rankingPutting every eligible company in order on one measure. A position is then held relative to the others rather than against a fixed line. Position twenty means nineteen are ahead, nothing more.: eligible listed companies are ordered by market capitalisation, and a published rule cuts that order into the groupings. The framework sits with two bodies. The Association of Mutual Funds in India maintains and publishes the ranked list on a stated cycle, and the Securities and Exchange Board of India runs the scheme categorisation that makes the grouping bite. Go to amfiindia.com for the first of those and to sebi.gov.in for the second.

The cut points move with the list, and the list is redrawn on a cycle. So a reader who memorises a rupee boundary will eventually be wrong without noticing, the memorised figure going stale silently rather than loudly. The whole argument for reading the rule at the body that issues it rests on that. A stale price carries its date, so it announces itself. A stale threshold looks like a definition, and so announces nothing.

Why is the classification a rule rather than a judgement?

Readers usually find this the surprising part. Everywhere else in research the training is to exercise judgement. Here the instruction runs the other way, and firmly. Why?

A ration queue served by printed token numbers carries the same logic. Nobody at the counter decides who looks more deserving. The token decides. The token feels cold until the alternative is considered. The alternative is not fairness but a fight. Two people at the counter would rank the same queue differently, and both would be certain. The token is not there because the counter staff lack judgement. The queue needs one answer rather than two defensible ones, and the token supplies it.

A size classification carries the same load. The classification is not a description written for readers, it is a constraint written for portfolios. If a scheme is permitted to hold companies of one grouping and not another, then the boundary decides what may be bought, and every holding either complies or does not. Suppose the grouping were a matter of judgement: one analyst puts a company in one grouping and another puts it in the next, and the same portfolio is then simultaneously compliant and non compliant, a failure of the rule itself rather than a difference of opinion. Mechanical classification is therefore not a limitation to be apologised for but the point of the arrangement. One consequence is worth holding on to. The rule attaches to a position in a ranking rather than to anything about the company, so a company can be reclassified without anything about it having changed.

How does a company move between the classifications?

There are exactly three routes, and one of them is the one nobody expects. The first is obvious: the company's own price moves, its capitalisation moves with it, and it climbs or slides in the ranking. The second is administrative: the list is redrawn on its stated cycle and positions settle where they settle. The third is the interesting one. Every other company's price moves while this one sits perfectly still, and the ranking rearranges itself around a company that did nothing at all.

The experience is a familiar one. A student's marks in a test are the same as last term, to the point. Everyone around that student improves, and the student drops from seventh in the class to twelfth. Nothing about the paper changed. The position did. A company can be reclassified with no movement whatever in its own price, purely because the companies around it moved, and that single fact is the clearest demonstration that the grouping is a relative position and not an absolute property of the business.

What does market capitalisation not measure?

Now the negative space, and most of the damage gets done there. Market capitalisation is built entirely out of equity and takes no account of what the company owes or what it holds in cash, so it is not the value of the business. Capitalisation is not what a buyer would pay for the whole thing either, for the reason the street of twelve shops gave: it prices every share at a level struck for a handful of them. And it is not size in any operational sense at all.

The last of those three produces a specific and common error, so it deserves a moment. A company with very large revenue can carry a modest capitalisation, and a company with modest revenue can carry a very large one, and neither situation is unusual or perverse. Capitalisation reflects what buyers and sellers currently assume about future profits; revenue reflects what has already been billed. Using market capitalisation as a proxy for how big a business is in operational terms is a substitution the figure cannot support, and it is entirely avoidable because operational size has its own measures sitting in the statements.

Set out concretely: a buyer taking the whole of Sarvani Coatings would inherit Rs 240 crore of borrowings and would acquire Rs 312 crore of cash and investments. Neither of those two amounts appears anywhere inside the Rs 11,664 crore. How they are combined with an equity figure to describe what a whole company costs is a separate calculation, taught separately.

How does an analyst actually use these two figures?

Meghna Iyer covers coatings makers. Both figures are printed everywhere, so she does not spend her day computing capitalisations. Her work is to decide, several times a day, which of the two a sentence needs. She has a habit worth stealing: she writes the basis into the sentence itself rather than leaving it to be inferred. Not Rs 11,664 crore, but Rs 11,664 crore on a total basis at Rs 486/- as at 28 August 2026. The longer form reads heavier. The longer form also survives being copied into somebody else's note. The shorter version does not.

The two figures then do different jobs on her desk. The total tells her which peer set a company belongs in and which comparisons will be read as sensible. The float tells her something about the position she can realistically take a view on, and something about how much index driven money is likely to be attached to the name. When the two diverge sharply, as they do here, she treats the divergence itself as information about the company rather than as an inconvenience: a business with more than half its register closely held behaves differently from one with a wide float, and that difference will show up long before anything appears in the statements.

The same discipline works at household scale, and this is the version worth keeping. A neighbour puts a flat at Rs 90 lakh because the identical one downstairs sold for that. Fine. The second question is the other one: how many people in this building want to sell this month, and how many want to buy? The first question gives a scale and the second gives a market, and confusing the two is the single most expensive habit in reading any price at all.