How an Indian IPO is actually priced and allocated

How to read this. Nothing here assumes you have applied for an IPO before. Every term is explained the first time it appears. Boxes marked The arithmetic carry the numbers, so if you want to check a claim rather than take it, the working is there. Boxes marked If you already follow IPOs are for readers who know the basics and want the part that is not obvious.

The worked example throughout is the National Stock Exchange's own share sale, which closed on 21 September, allotted on the 22nd and listed on the 24th. It is the best available teaching case: enormous, fully documented, and it did something unusual on listing that most coverage read backwards.

01 · What an IPO actually is

An IPO is usually described as a company selling shares to the public. That is true and it leaves out everything that decides what happens to you.

What it actually is: a three-day auction, inside a narrow fixed price range, where the demand is split into separate pots that cannot help each other, and where nearly a third of the shares have already been handed out the day before the auction opens.

Each of those four features is set by regulation, not by the company. Each one changes your outcome.

The sequence, start to finish:

1. The company files a draft prospectus with SEBI. This is a long document describing the business, the risks and where the money will go. It carries no price. The public has 21 days to comment on it.
2. SEBI responds with observations. The company files the red herring prospectus, which is the same document plus a price band: a floor and a ceiling. The ceiling cannot be more than 20% above the floor.
3. One working day before bidding opens, the anchor book. Selected institutions are allocated shares at a price the bankers set. There is no auction here at all.
4. Three days of bidding. Everyone else applies, in fixed lots, at a price inside the band or at "cut-off".
5. The price is set.
6. Allotment, by rules that differ for each pot. Some proportionate, some by an actual computerised draw.
7. Listing, usually three working days after bidding closes.

The pots

Here is the structural fact that makes every headline number about an IPO misleading: the money is not in one pot.

The shares are divided into fixed buckets before anyone bids, and the buckets cannot borrow from each other except in narrow circumstances. There are three:

• Institutions. Mutual funds, insurers, foreign funds, banks. In the filings they are called Qualified Institutional Buyers, or QIBs.
• Wealthy individuals. Applications above ₹2 lakh. Called Non-Institutional Investors, or NIIs.
• Retail. Applications up to ₹2 lakh. That is almost certainly you.

And here is the part almost nobody tells a first-time applicant: which split you get depends on whether the company is profitable.

A company that passes SEBI's profitability tests can use the ordinary route: 50% institutions, 15% wealthy individuals, 35% retail.

A company that does not pass those tests must use a different one: at least 75% institutions, 15% wealthy individuals, and only 10% retail. And that route carries a condition. If the institutional portion is not fully covered, the entire issue is cancelled and everyone gets their money back.

So a retail allocation of 10% or less is a plain, factual statement, printed on the cover of the prospectus, that the company did not pass the profitability tests. It takes five seconds to check and it is one of the most informative things in the whole document.

If you already follow IPOs. You can recover which route an issuer took without opening the prospectus at all, from three published subscription numbers.

The headline subscription multiple excludes anchor shares, so it is a weighted average of the remaining pots, and the weights are fixed by the route. Assuming the anchor book uses its full allowance, the two routes give completely different weights, and only one reproduces the published headline.

Adroit Industries closed on 25 September at a reported 176.85 times. Institutions excluding anchor 195.04, wealthy individuals 332.35, retail 99.81. Under the profitable route those weights are 28.57 / 21.43 / 50.00, which gives 176.849 against 176.85 reported. Under the unprofitable route you get 215.17, off by 21.7%. Adroit took the profitable route.

Elevate Campuses, ₹2,100 crore, closed the same day at a reported 1.79 times. Institutions 2.52, wealthy individuals 0.84, retail 1.01. The profitable route gives 1.405, off by 21.5%. The unprofitable route, weights 54.55 / 27.27 / 18.18, gives 1.787 against 1.79 reported. Elevate took the unprofitable route.

Confirmed independently by the anchor book. Elevate's anchor was ₹945 crore on a ₹2,100 crore issue, exactly 45.0%. The anchor ceiling is 60% of the institutional portion: ₹630 crore under the profitable route, which ₹945 crore would breach, and ₹945 crore under the unprofitable one, met to the rupee. Two different methods, same answer.

Why it matters rather than being a party trick: Elevate's institutional book came in at 2.52 times, so the issue cleared. Its retail and wealthy-individual pots, the ones that cannot cancel an issue, came in at 1.01 and 0.84. Retail money went into a company that had not passed the profitability tests and that institutions had to carry, and the headline "1.79 times subscribed" conveys none of that.

02 · Does the company get the money?

This is the fastest useful read on any IPO and most people never make it.

There are two kinds of shares in an offering, and they are often mixed in one issue.

Fresh issue. The company creates new shares and sells them. The money goes to the company. It can build a factory, repay debt, hire.

Offer for sale. Existing shareholders sell shares they already own. The money goes to them. The company receives nothing. Not one rupee. The business has exactly the same cash on the day after listing as the day before.

Neither is wrong. A company that does not need capital has no reason to raise it, and existing investors are entitled to sell. But the two mean completely different things about what you are buying into, and the split is stated plainly on the cover page.

NSE's offering was 100% offer for sale. The entire ₹22,563 crore went to existing shareholders selling down. NSE itself received nothing, and did not need to.

That is worth holding next to what those shareholders were selling.

The arithmetic. At the issue price of ₹1,785, NSE was valued at about ₹4.42 lakh crore. Against FY26 profit of ₹9,228 crore, that is roughly 48 times earnings, and about 13.8 times book value.

FY25 profit was ₹10,978 crore. So FY26 earnings fell about 16% year on year.

Transaction charges are 78.65% of operating revenue, which makes the exchange overwhelmingly dependent on one line. Within that line, share of the equity-options market went from 96.9% in FY24 to 68.5% in Q1 FY27.

Forty-eight times last year's earnings, for a business whose earnings declined last year, whose dominant product is losing share at roughly ten percentage points a year, and where every rupee raised went to people leaving.

None of that is a scandal. NSE remains extraordinarily profitable and monopolies derating is normal. But it is a different picture from "India's exchange is listing", and all of it is in the prospectus.

Plain words. Earnings multiple, or P/E. The share price divided by the profit per share. Forty-eight times means you are paying forty-eight rupees for every one rupee of last year's profit. Whether that is expensive depends entirely on whether next year's profit is bigger.

03 · Who decides the price, and when

Short answer: the bankers, before bidding opens, and the three days mostly confirm it.

The band

The prospectus carries a floor and a ceiling. The ceiling may be up to 20% above the floor, and since January 2022 it must be at least 5% above.

In practice, issuers have settled at the bottom of that range. LIC's band was 5.21% wide, Tata Technologies 5.26%, Hyundai Motor India 5.09%, NSE exactly 5.00%. All four priced at the ceiling.

Before the 5% minimum existed, issuers went further: ICICI Securities filed a band of ₹519 to ₹520 in 2018. Nineteen hundredths of one percent. The minimum exists to stop exactly that.

The anchor book

Up to 60% of the institutional portion may be handed to anchor investors one working day before bidding opens, at a price the bankers set, to institutions the bankers choose. Sixty per cent of half the issue is thirty per cent of the whole thing, allocated before you can do anything.

NSE used nearly all of it.

The arithmetic. NSE's anchor book was 3,77,93,739 shares at ₹1,785, or ₹6,746 crore. Total offering 12,64,36,650 shares.

3,77,93,739 ÷ 12,64,36,650 = 29.89%. The ceiling is 30%.

Plain words. Anchor investor. A large institution invited to commit before public bidding, to give the issue a visible base of demand. Their shares are locked for a period, which matters later.

What the bidding actually showed

Same company, same price, same three days. Institutions asked for nearly thirteen times what was available to them. Retail asked for a little over one times.

The headline everywhere was 5.71 times subscribed. A single number that averages 12.68 and 1.39 is not summarising those two facts, it is hiding them. Institutions and retail have wildly different amounts of capital behind them, so a multiple measures the size of the pot at least as much as it measures enthusiasm.

If you already follow IPOs. Try to rebuild 5.71 from the parts and it does not work.

Add every share bid for, anchor block included, and divide by shares on offer: 4.30 times. The 5.71 appears only if you remove the anchor shares from the denominator, dividing non-anchor bids by the offer excluding the 3,77,93,739 anchor shares. Then it lands exactly on 5.71.

Neither is wrong. They answer different questions and both are defensible. But they are about a third apart, nobody states which one they are quoting, and different sites mix them freely. If you have ever compared a subscription figure across two websites and found they disagree, this is usually why.

And a flat listing is a good outcome

NSE listed at ₹1,818 against an issue price of ₹1,785. A rise of 1.85%. Much of the coverage called that disappointing.

Look at it from the seller's side.

The arithmetic. The gap between issue price and first-day close, times shares sold, is money the sellers could have charged and did not.

(₹1,818 − ₹1,785) × 12,64,36,650 = ₹417 crore.

Across 84 mainboard listings in 2026, the average first-day gain has been about 12.2%. On a ₹22,563 crore issue that would have been about ₹2,753 crore.

Pricing tightly was therefore worth roughly ₹2,336 crore to the sellers, far more than anyone involved was paid in fees.

A big first-day pop is usually described as the market rewarding a good company. More often it is a measure of how much the sellers under-charged, and the people most likely to be allotted a lot of a hot issue are not retail applicants in a draw.

04 · Why you probably won't get the one you applied for

When you apply you choose a number of lots. It is natural to assume more lots means more shares. In an issue that many people want, it does not.

Plain words. Lot. The smallest number of shares you may apply for. NSE's was 8 shares, ₹14,280 at the ceiling of the band.

SEBI's rule is that every successful retail applicant must receive at least one full lot. Once there are more applicants than there are lots to hand out, there is no way to honour that and give everybody something. So the registrar runs a draw. A literal computerised lottery. Some applicants get exactly one lot. Everyone else gets nothing and their money is unblocked.

Plain words. Registrar. The firm that runs the allotment and the draw, appointed by the company.

The arithmetic. NSE's retail pot held 4,42,52,828 shares, which at 8 shares a lot is 55,31,603 lots.

Retail was covered 1.39 times, so roughly 1 ÷ 1.39, about 72 in every 100 single-lot applicants, would have been allotted.

Adroit Industries, closing four days later, had a retail pot covered 99.81 times, from 31,47,253 applications. That is about 1 in 100.

Tata Technologies drew around 7.3 million applications in November 2023, roughly one in eight. Bajaj Housing Finance drew close to 9 million the following September, about one in five.

Now the part that feels unfair and is simply arithmetic: the more attractive an issue looks, the more people apply, and the smaller everyone's chance becomes. Your odds move against you exactly as your reason for wanting it grows. Your expected outcome is the listing gain multiplied by a probability that runs in the opposite direction.

Nobody designed that. It falls out of the guarantee that a winner gets at least one full lot, which is itself a decent rule.

One consequence worth knowing: the number that sets your odds is your own category's multiple, not the headline. Retail at 1.39 times and a headline of 5.71 times are different facts about the same issue, and only the first one is about you.

05 · The date your stock can fall

Anchor investors do not get their shares free of conditions. They are locked.

Plain words. Lock-in. A period during which shares that have been allotted cannot be sold. It ends on a fixed date calculated from allotment.

For anchor investors the lock-in comes in two parts: half the shares become sellable 30 days after allotment, and the other half at 90 days. These are not estimates. They are arithmetic on a date printed in a filing.

NSE was allotted on 22 September.

The arithmetic. Anchor allotment 3,77,93,739 shares.

First tranche, half of that: 1,88,96,869 shares, about ₹3,373 crore at the issue price, becomes sellable around 22 October 2026.

The balance becomes sellable around 21 December 2026.

That is a known quantity of shares becoming sellable on a known date, computable the moment the anchor allotment is filed, weeks or months in advance.

We do not tell you what will happen on those dates, and we will not. What we do is publish the calendar, for every live issue, with the filing page each date was read from. What has happened at past unlocks is on the record page, with the sample size and the condition that would show the pattern to be wrong.

Almost nobody publishes these dates in a usable form, which is the main reason this site exists.

06 · Grey market premium, and why we don't publish it

You will see a number called the grey market premium, or GMP, quoted everywhere during a live issue. It is presented as what the shares are "already trading at" before listing.

We do not publish it, and we never will. Three reasons.

It is unverifiable. There is no exchange, no clearing house, no settlement and no public record. The number is reported by a handful of intermediaries who also have positions. There is no document anyone can check it against, which means it fails the one test everything on this site has to pass.

It is unregulated. The trades behind it, where they exist, are outside the regulated market entirely.

It is the business model of nearly every site that competes with us. GMP generates a number that changes every few hours during a live issue, which generates traffic, which sells advertising. That incentive is precisely why it is everywhere and why it is never examined.

This site takes no money from anyone, so it does not need a number that refreshes hourly. Everything we publish is a figure from a filing, with the page it came from. GMP has no page.

We are testing, across a full set of recent issues, whether GMP has any relationship at all with what happens on listing day. When that work is finished we will publish the result whichever way it comes out, including if it makes us look wrong.

What to take from this

If you read nothing else:

The money may not go to the company. Check the cover page for offer for sale against fresh issue.

A retail allocation of 10% or less means the company did not pass SEBI's profitability tests. Also the cover page.

Your category's subscription multiple is the one that sets your odds. The headline is a weighted average that mostly measures pot size.

Allotment above one times is a draw, not a queue. More lots does not mean more shares.

A flat listing means the issue was priced well. A large pop means the sellers under-charged, and you were probably not the one who collected it.

Every one of those is a fact from a document, not a view about a company. We publish all of them for every issue, with the filing page each came from, so you can check us rather than trust us.