How we score IPOs: GMP, demand and your chances, explained

An IPO (initial public offering) is when a company sells its shares to the public for the first time. You apply, and if you’re picked, the shares land in your account on listing day: the day they start trading on the stock exchange.

On that first day the price can jump above what you paid, or drop below it. That difference is the first-day gain (or loss), also called the listing gain. Most of this site is about one question: which IPOs are worth applying for, and how likely are you to actually get shares?

When lots of people want shares, there aren’t enough to go around. The IPO is then called oversubscribed, and shares are handed out by lucky draw. The IPO subscription status you see on tracking sites is just a running count of how far over the line demand has gone.

GMP (grey market premium) is traders’ unofficial guess of an IPO’s first-day price. Say the IPO price is ₹100 and traders are willing to pay ₹130 for the shares before they list. The GMP is ₹30, or 30%, and people read that as “it will probably list near ₹130”. You’ll see it written as “IPO GMP” on most tracking sites.

But it’s a rumour with a number on it. No stock exchange records these deals and no regulator watches them. They happen over phone and WhatsApp, often in cash. Even the sites that publish GMP say the figures are only rough guides.

People still look at it because it’s the only “price” you can see before listing. Treat it as one clue, never a promise. And 2025 showed why: more than half of IPOs listed below what GMP suggested.

• Kostak is a fixed price a trader pays to buy your application before shares are handed out. You get paid whether or not you win shares.
• Subject-to-sauda is a price paid only if you actually get shares.

A backtest just means checking an idea against past IPOs to see whether it would have worked. We do this because “high GMP means profit” held up well in 2023 and 2024, then broke in 2025. A rule that worked for two years can fail in the third, and testing on past results shows you that before your money does.

What happened, 2023 to 2026

Demand turned out to be a better clue than GMP. A 2025 study of 107 big-company IPOs found the link between demand and first-day gain was stronger (0.62) than the link for GMP (0.50), on a scale where 1 is a perfect match. IPOs that were requested 50× or more averaged a +19.2% first-day gain. Those under 10× averaged −2.15%. The study is not peer-reviewed (opens in a new tab), so treat it as a hint, not a law.

The Edge Score is one number from 0 to 100 for each IPO. Higher means the IPO looks better compared with similar past IPOs. It blends four things, and each one is shown separately so you can see what’s pushing the score up or down.

• Expected first-day gain

  What traders predict (the GMP), as a percent of the top of the IPO’s price range. Useful, but the least reliable of the four.

• Demand

  How many times more shares people have requested than are on offer. Everyday-investor demand counts most, because unmet demand pushes prices up on day one.

• Price trend

  Whether the traders’ predicted price is rising or fading while the IPO is open. A prediction that keeps sliding is a warning sign.

• Market mood

  Whether the wider stock market is rising or falling, including between the IPO closing and listing day. A gloomy market drags down even popular IPOs.

You have to apply before the IPO closes, but the final demand numbers only arrive in the last few hours. If we scored past IPOs using those final numbers, we’d be cheating: judging choices with facts nobody had when they tapped “apply”. Experts call this look-ahead bias.

So every signal is built from only what was visible at the moment of decision. We test three moments.

This is also why we save dated snapshots of predicted prices. Other websites overwrite yesterday’s number with today’s, so what you could actually see at the time is easy to lose.

Before you check your IPO allotment status, it helps to know the odds going in. When more shares are requested than are available, the shares go to people by lucky draw. A rough guide: your chance of getting shares is 1 divided by how many times over the IPO is booked. If it’s 50× booked, that’s about 1 in 50, or 2%.

Each person (PAN card) can apply only once. So a family can raise its chances by having each member apply. Every extra person is another ticket in the draw, but it’s still a small chance each time.

Worked example: 50× demand, 10 family accounts

This is an estimate, not a promise. Real draws work slightly differently for different groups of investors, and we say so wherever the chances appear.

Picking winners isn’t enough if you rarely get shares. So we score each idea on the average profit per application: what you’d make on average if you applied many times.

We work it out like this: your chance of getting shares, times the amount you’d invest, times the first-day gain, minus the cost of having your money tied up.

An example, before any costs. You apply with ₹15,000, your chance of shares is 2%, and the first-day gain is +10%. Your gain if you win is ₹1,500, so on average 2% × ₹1,500 = ₹30 per application.

That’s small, which is why we show the family view. It also means an IPO with a huge predicted gain but a 1-in-100 draw can be worth less than a modest one you’re much likelier to get.

Any score is only as honest as the data behind it. Our IPO numbers are real. They come from three public websites, and past readings are stored so nothing gets overwritten.

• InvestorGain. Traders’ predictions and demand. A live collector saves fresh readings every 30 minutes on market days, and every few hours at other times.
• Chittorgarh. IPO dates, sizes, price ranges, banks, the company that runs the lucky draw (the registrar) and first-day prices. Stored as past data.
• IPO Watch. A second set of trader predictions. We use it to fill gaps where InvestorGain has none. Stored as past data.

Where our data falls short

• Predictions are unofficial. They come from grey-market traders, no regulator watches them, and different sites sometimes quote different numbers for the same IPO. For 8 IPOs no prediction was ever published. We show “Not published” for those, never a zero.

• Older demand is only a final total. For IPOs before 2026, we only know the final overall demand. The split between regular investors, wealthy investors and big institutions, and the day-by-day numbers, weren’t published publicly. So tests like “decide after day 1 or day 2” use predictions only on older IPOs. Day-by-day records start in September 2026 and grow every day.

• Older prediction history can be thin. About 4 in 10 IPOs that listed in 2023 have three or fewer prediction readings, and some readings only have a date, not a time of day. 2024 onwards is much fuller, with a typical IPO having 8 to 10 readings.

• What we cover. IPOs from 2023 to today, big companies and small companies. 1,185 have already listed, so we know their first-day price. By year of listing, 2023: 238 (59 big, 179 small). 2024: 338 (91 big, 247 small). 2025: 373 (104 big, 269 small). 2026: 236 (83 big, 153 small). Another 53 are open, closed or coming up. A few types are left out: REITs and InvITs (funds that own property or infrastructure) and follow-on offers (companies already listed selling more shares).

• We collect from public websites, not the exchanges. There is no free, official feed that has all of this in one place, so we read it from InvestorGain, Chittorgarh and IPO Watch. Those sites can change or go down. A paid, licensed feed from the stock exchanges is the plan before we charge anyone.

• Pulled IPOs go missing. IPOs that were withdrawn before they listed rarely appear on trackers, which can make past results look a little better than they were.

• Big and small companies differ. We treat big-company and small-company IPOs separately. Their first-day behaviour, ease of selling and risk are very different.

• Market mood still uses sample numbers. The Nifty and smallcap index levels behind the market-mood panel are not real yet, and the app labels them “Sample numbers”. Every other IPO number is real.

• Not investment advice. Past first-day results can’t tell you what the next IPO will do. Use the numbers as one clue, not a promise.

India’s market regulator is SEBI (the Securities and Exchange Board of India). Its rules decide what this product can and can’t do. This is our plain reading of the main ones, not legal advice. If you pay for a family member’s application from your own money, tax rules may apply, so check with a chartered accountant (CA).

The short versions of what people ask most.