The IPO Test: Why a Great First Day Can Be a Bad Investment

COMPANY RISK RATINGS

By InsidEntity Editorial Desk · Sep 26, 2026 · 21 min read

The average US IPO gained 19% on its first day. But over three years, IPOs listed through 2024 trailed the broad market by 20.5 percentage points. Across more than 9,000 listings, the lesson is less about the pop than the price paid.

The short version

The average US IPO has gained about 19% on its first day. Over three years, IPOs listed through 2024 trailed the market by about 20 percentage points. That gap, not the pop, is the story.

The pop is real, but uneven. Across 9,345 US IPOs from 1980 to 2025, the average first-day return was 19.0%. The median was 7.0%, and 16.5% closed below their offer price.
The aftermath is weaker. Measured from the first closing price, IPOs from 1980 to 2024 returned +19.1% over three years on average. The market returned +39.6% over the same windows.
The typical IPO lost money. Six in ten IPOs had a negative three-year return, and the median was -25.7%. A few very large winners hold the average up.
The year matters. The 2012 cohort beat the market by 31.8 points over three years. The 2020 cohort trailed it by 78.6 points.
Price matters most at the extreme. Large-company IPOs priced above 40 times sales returned -44.8% over three years on average.

What this evidence does not say: that IPOs are bad investments, or that any single IPO will follow the average. It describes what has happened across thousands of listings. It tells an investor what deserves scrutiny, not what to buy.

This is an InsidEntity research piece built on an evidence ledger. It is not investment advice, and it takes no position on any listing. How we built it, and the safeguards we applied, are set out at the end.

1. What happens after the pop?

IPOs have underperformed the broad US market over three years, even after a strong first day. The first-day return and the three-year return answer two different questions. One is about how the shares were priced at admission. The other is about what shareholders experienced afterwards.

Measure	Figure	Sample
Average first-day return (offer to first close)	18.9%	9,253 US IPOs, 1980-2024
Average 1-year return (from first close)	+5.6%	9,253 US IPOs, 1980-2024
Average 3-year buy-and-hold return (from first close)	+19.1%	9,253 US IPOs, 1980-2024
Market return over the same 3-year windows	+39.6%	Morningstar US Total Market Index
Market-adjusted 3-year return	-20.5 pp	9,253 US IPOs, 1980-2024
Median 3-year return	-25.7%	9,195 US IPOs, 1975-2021
IPOs with a negative 3-year return	60.0%	9,195 US IPOs, 1975-2021
IPOs that lost more than half their value in 3 years	38.5%	9,195 US IPOs, 1975-2021

Source: Ritter, Initial Public Offerings: Updated Statistics (14 September 2026), Tables 16a, 16e and 19. pp = percentage points.

The average hides the typical outcome. The mean three-year return is positive; the median is deeply negative. That gap comes from skew: a small number of extraordinary winners lift the average. Over five years the pattern holds, with a mean of +37.8%, a median of -32.0%, and 42.6% of IPOs down by more than half.

The gap has widened in the modern era. For the 1,724 IPOs from 2011 to 2024, the average first-day return rose to 23.0%. The average three-year return fell to +9.1%, trailing the market by 24.7 points. Bigger pops have not meant better outcomes.

A note on the starting point. These long-run returns start at the first closing price, not the offer price. That is the price most investors can actually buy at. An investor who received shares at the offer price also captured the first-day move.

2. Does the year you buy in matter?

It matters a great deal. Across the 18 complete cohorts from 2005 to 2022, the three-year outcome swung from 31.8 points ahead of the market to 78.6 points behind it. The pattern that stands out: the strongest listing years in this sample were followed by some of the weakest three-year results.

Source: Ritter, Initial Public Offerings: Updated Statistics (14 September 2026); InsidEntity IPO Evidence Ledger. 2023 and 2024 are partial horizons, measured to December 2025.

2020 and 2021: the strongest demand met the highest prices. The 2020 cohort of 165 IPOs gained 41.6% on day one, leaving $29.7 billion on the table. Three years later it was down 48.1%, 78.6 points behind the market. The 2021 cohort, a record 311 IPOs raising $119.4 billion, fell 49.1% over three years.

2012: a modest pop, the best modern outcome. The 93 IPOs of 2012 gained 17.7% on day one. Over three years they returned +81.9%, beating the market by 31.8 points.

2008: the crisis cohort. Only 21 companies listed in 2008, and 61.9% of them closed below their offer price on day one. That cohort then beat the market by 8.1 points over three years.

The lesson is not that crises are good times to buy. It is that the valuation environment a company lists into shapes what its shareholders experience afterwards. The 2023 and 2024 rows are early readings: their returns run only to 31 December 2025, so they cover roughly two to three years and one to two years respectively, depending on the listing date.

3. Does the first day predict the next three years?

Only loosely. A weak debut has been a warning sign across the whole group, but a strong debut has not been a promise. Every group, including the IPOs that rose on day one, trailed the market over three years.

Source: Ritter (2026); InsidEntity IPO Evidence Ledger. Three-year returns from the first closing price.

Broken debuts. Of the 654 IPOs from 2001 to 2024 that closed below their offer price, 442 (67.6%) still had a negative three-year return. The other third did not, which is why a bad first day is a flag and not a verdict.

The exceptions cut both ways. Academy Sports & Outdoors closed one cent below its offer price in 2020. It then returned +256.2% over three years. Yahoo! rose 153.8% on its first day in 1996 and returned +3,589.8% over three years. In both cases it was the business that delivered, not the debut.

The reading for an investor: the first-day price move tells you about demand for the shares on one day. It tells you little about whether the price paid was right for the business.

4. Does the price matter?

Yes, and most sharply at the top. Among established companies, the higher the price-to-sales ratio at listing, the bigger the first-day pop and the weaker the next three years.

Source: Ritter (2026); InsidEntity IPO Evidence Ledger. Large companies only: pre-listing sales of $100 million or more.

Read the ladder carefully. The average three-year return falls at every step up the valuation ladder: gradually through the middle, steeply at the top. IPOs priced at 20 to 40 times sales returned +2.9% over three years, 16.3 points behind the market. Those priced above 40 times sales returned -44.8%, 58.5 points behind. Even the cheapest group, below five times sales, trailed the market slightly, by 1.3 points.

The entry price changes the verdict. These returns start at the first closing price. Measured at the offer price instead, only 14 IPOs were priced above 40 times sales, and 12 of them underperformed. The exceptions were Mobileye (2014) and Astera Labs (2024, on a partial horizon). Separately, Nu Holdings (2021) beat the market for investors who bought at the offer price, but not for those who bought at the first close.

A low share price is not a low valuation. The figures in this article exclude IPOs offered below $5 a share, but Ritter tracks them separately. From 2001 to 2024, the 179 IPOs offered below $5 gained 67.1% on day one on average. Over three years they lost 46.3%, 82.6 points behind the market. IPOs offered above $5 returned +14.7%.

Listing through a SPAC produced worse results still. SPACs are also outside the main sample, and Ritter tracks the companies that merged with them separately. The 451 companies that listed this way between 2012 and 2022 returned -57.7% over three years on average. The 2021 vintage returned -73.0%.

5. Does the business underneath matter?

It does, but not in the way the usual rule of thumb suggests. “Never buy a loss-making IPO” does not survive the data on its own. Scale changes the picture.

Source: Ritter (2026); InsidEntity IPO Evidence Ledger. Sales in 2025 dollars; figures below the bars are returns against the market.

Small and loss-making was the weakest combination. The 2,880 IPOs that were unprofitable with sales below $100 million returned -10.3% over three years, 40.9 points behind the market. Across all loss-making IPOs the average was -0.5%, against +33.6% for profitable ones.

Larger loss-makers held up. Unprofitable companies with sales of $100 million or more returned +26.2%, just 2.7 points behind the market. Profitable companies of the same size trailed by 3.4 points. At that scale, being profitable at listing made little difference to the three-year result.

Size carries through. IPOs with pre-listing sales of $1 billion or more returned +32.4% over three years, 2.1 points behind the market. Smaller issuers returned +17.8%, 22.4 points behind.

Who backs the company matters less than it seems. Venture-backed IPOs trailed the market by 13.6 points, against 25.2 points for those without venture backing. Among larger companies, buyout-backed IPOs trailed by 0.9 points and others by 4.1 points. These are associations across groups, not proof that a sponsor improves outcomes.

The reading for an investor: the question is not “is it profitable?” alone. It is how large, how established and how proven the business is at the price being asked.

6. What happens to companies after they list?

Most of them leave the exchange within a few years. That is not the same as most of them failing, and the difference matters.

Staying listed is unusual. A 2026 study in Accounting & Finance by Gippel, Linnenluecke, Smith, Xu and Zhu found that five-sixths of IPOs delist within seven years of listing, so about one in six remains independently listed. A separate measure in the same study, limited to IPOs observed for at least eight years, found that more than 84% had delisted and that the median listed life was 6.4 years. The two figures come from different sample windows and are not combined.

Many exits are takeovers. Of 8,368 US IPOs from 1980 to 2017, 967 (11.6%) were acquired or bought out within three years of listing. An acquisition ends the listing, sometimes at a premium. For those shareholders, the end of the listing was an acquisition event, not necessarily a loss.

Conflating the two is the classic IPO statistics error. “Five in six IPOs delist” is a finding about how long companies stay public. It is not “five in six IPOs fail”. The study separates voluntary exits from involuntary ones, and so does this article. The exact split sits inside the paper’s paywalled body, so we do not quote it.

What seems to help companies last. A separate 2026 study in the Journal of Business Finance & Accounting found that newly listed firms led by CEOs with a financial background had a lower probability of involuntary delisting and survived longer. That is an association, not proof of cause. It is also a reminder that who runs the company is part of the investment case.

When it went wrong

The averages above say how often IPOs disappoint. These cases show how IPO risks can materialise. Each illustrates a different issue, and each maps to one of the five tests set out later in this article.

These cases were chosen to teach, not drawn as a statistical sample. Every fact comes from a filing, a regulator or a named report. Allegations of misconduct are described as allegations unless a court or regulator has made a finding.

IPO	Listed	What happened	How it failed	Test it maps to
Pets.com	Nasdaq, Feb 2000, $11	Sold 7.5 million shares, about $82.5 million. The board adopted a plan of complete liquidation nine months later, and the wind-down was announced on 7 November 2000.	Business model	5. Growth burden
Steinhoff Africa Retail (STAR)	JSE, Sep 2017, R20.50	Raised R15.4bn; the net proceeds went to Steinhoff subsidiaries, and Steinhoff kept 77%. Eleven weeks later the parent’s accounting scandal broke and STAR fell more than 30%.	Controlling shareholder	4. Governance
Libstar	JSE, May 2018, R12.50	Closed its first day at R12.30. Of R1.5bn raised, R700m cut debt and R800m paid a special dividend to existing shareholders. In 2026 it bought back shares at an average of R4.59, about 63% below the offer price.	Use of proceeds and price	2. Capital; 1. Valuation
Luckin Coffee	Nasdaq, May 2019, $17	The SEC charged that it intentionally fabricated more than $300 million of retail sales between April 2019 and January 2020, using related parties. Luckin settled for $180 million without admitting or denying the allegations. Its shares were suspended from Nasdaq at the end of June 2020.	Accounting	3. Financial quality
WeWork	IPO withdrawn Sep 2019	Its filing showed 20-vote founder shares and rent and loans involving the founder. It pulled the IPO, listed through a SPAC in 2021 at about $9bn, and filed for Chapter 11 on 6 November 2023.	Governance	4. Governance
Facebook	Nasdaq, May 2012, $38	Raised over $16bn and closed its first day at $38.23. By early September 2012 it had closed at $17.73, and it took more than a year to regain $38.	Pricing, then recovery	1. Valuation

Two patterns run through the list. In three of the six, a relevant feature was in the offering document before anyone bought: where the money went (STAR, Libstar) or who controlled the company (WeWork). And failure took different routes. Pets.com liquidated, WeWork went bankrupt after a later listing, and Luckin’s shares were suspended after the misconduct came to light. Facebook fell by more than half and then recovered, because the business underneath was profitable and growing.

That last case is here on purpose. A warning that ignores the recoveries overstates the risk. The point is not that IPOs fail. It is that some important IPO risks are visible in advance in the offering document: how the money will be used, who controls the company, related-party arrangements, the financial track record and the assumptions built into the valuation. Others, including undisclosed misconduct of the kind the SEC charged at Luckin, may not be visible at all.

The South African evidence

The JSE has its own record, but it is not a smaller version of the US one. It comes from a separate study, with a different period, benchmark and method. Its numbers cannot be set against the US figures or used to confirm their size. The two point in the same direction over three years, and that is all the comparison supports.

A study of 313 IPOs listed on the JSE between 1996 and 2007, published in the African Review of Economics and Finance (2014), found three-year underperformance of 65.59% using buy-and-hold abnormal returns (BHAR). Using cumulative abnormal returns (CAR), the figure was 59.77%. The study also found that the conclusion changes with the benchmark and the way returns are measured.

	US evidence	JSE evidence
Sample	9,253 operating-company IPOs	313 IPOs
Listing period	1980-2024	1996-2007
Measure	3-year return minus a total-market index	3-year abnormal return (BHAR and CAR)
Result	-20.5 percentage points	-65.59% (BHAR); -59.77% (CAR)
Caveat	Equal-weighted; from first close	Sensitive to benchmark and method

Read the table down each column, not across it. The two results are different measurements, and the gap between -20.5 points and -65.59% says nothing about whether the JSE was worse.

The 2026 rules put more weight on disclosure. The JSE’s simplified Listings Requirements took effect on 13 January 2026 for new applicants and on 16 February 2026 for existing issuers. The framework rests on disclosure, investor protection and orderly markets. Simpler rules make the prospectus more important, not less, because more of the judgement moves to the investor reading it.

The InsidEntity IPO Test: five questions before you subscribe

The evidence does not tell an investor what to buy. It tells them what deserves scrutiny before they decide. These five questions turn the record above into a pre-IPO checklist.

They are a framework for asking questions, not a prediction model or an investment rating. No combination of answers has been tested as a way to forecast returns, and an IPO does not become a good investment by answering all five.

1. Valuation: what must the company achieve to justify this price?

Work out the market value and enterprise value at the offer price. Compare price-to-sales, and price-to-earnings or free-cash-flow yield where they mean something, with listed peers. Use the fully diluted share count. In the evidence, the three-year return falls at every step up the valuation ladder and falls hardest at the top: large IPOs priced at 20 to 40 times sales returned +2.9% over three years, and those above 40 times returned -44.8%.

2. Capital: where does every rand and dollar raised actually go?

Read the use-of-proceeds section line by line. Separate new shares, where the money goes to the company, from existing shares, where it goes to the sellers. Note how much repays debt, pays out existing owners or funds an acquisition. The SEC’s investor bulletin is a good plain-language guide. A selling parent is not automatically a warning sign: Boxer’s 2024 JSE listing paid down Pick n Pay’s debt, and its shares were up about 48% including its first dividend by October 2025. What matters is whether the price leaves value for the new buyer.

3. Financial quality: do the earnings turn into cash?

Rebuild the path from revenue to free cash flow. Look for customer concentration, reliance on “adjusted” measures, share-based pay, related-party transactions and any change of auditor. Count the full years of audited results the prospectus contains. InsidEntity’s Financial Stability Rating (FSR) is scored only on full-year audited statements over a rolling three-year window, for the same reason: interim and pro-forma numbers are not a track record.

4. Governance: who controls the company, and what protects minority owners?

Check founder and parent ownership, voting rights by share class, board independence, the lock-up schedule and the size of the free float. Dual-class structures are now common: 10.3% of US IPOs on average since 1980, 33.1% in 2021 and 41.1% in 2025. The median public float fell from 31.5% in 2005 to 12.6% in 2023.

The return data does not show that dual-class IPOs did worse. Over three years they returned +29.5%, against +18.0% for single-class IPOs. So this is a governance question, not a proven return penalty: how much control stays with insiders, and what protects everyone else? It is the question InsidEntity’s Company Risk Rating (CRR) asks of every listed board, across four pillars: Director Independence, Director Capacity, Auditor Independence and Shareholder Influence. Before you subscribe, check the board’s record on InsidEntity.

5. Growth burden: how much future success is already in the price?

Reverse-engineer the revenue, margin and cash flow the offer price assumes. Then run a downside and a severe-downside case, and ask what still has to go right at each stage. The 2020 and 2021 cohorts show what happens when the price assumes everything goes right: first-day gains of 41.6% and 32.1%, then three-year losses of 48.1% and 49.1%.

What we know, what we don’t yet, and what the evidence cannot say

What we know

First-day returns for every US IPO cohort through 2025: 9,345 listings, averaging 19.0% on day one.
Complete three-year outcomes for every cohort through 2022, and early readings for 2023 and 2024.
The 2025 cohort’s first day: 90 IPOs, an average first-day gain of 29.3%, and 24.4% closing below their offer price.
That South Africa’s 1996-2007 listings trailed the market over three years, on the measures that study used.

What we don’t know yet

How the 2025 cohort does over time. One-year returns can only be measured after December 2026, and three-year returns after December 2028.
Anything about 2026 IPOs as a cohort. They enter the academic dataset with its 2027 edition.
How this year’s listings, from SpaceX and Cerebras to Dangote and Airtel Money, will perform. We do not estimate it; our follow-up, The 2026 IPO Test, applies the five questions to them.

What the evidence cannot establish

How any single IPO will turn out. The data describes distributions, not individual companies.
That any feature, such as a high valuation or a dual-class structure, causes poor returns. It shows associations.
That the US pattern holds on the JSE today. The South African study covers an earlier period with a different method.

The bottom line

The IPO itself is not the investment case. The first-day move measures demand for the shares on one day. The next three years measure the business, and whether the price paid for it left room for anything to go wrong.

Twenty years of US IPO cohorts, more than 9,000 listings and a separate South African study all document substantial three-year underperformance in their respective samples. The size of the pop tells you much less than the valuation you are buying at, the quality of the earnings, where the money goes and who controls the company afterwards. Those answers are in the offering document before the shares start trading.

Read the prospectus.

Coming next: The 2026 IPO Test. Our follow-up runs the five questions on this year’s biggest listings: SpaceX, Cerebras, China Resources New Energy, the Dangote refinery and Airtel Money.

Independent research. Not financial or investment advice. Statements about misconduct are limited to what the regulators cited have charged or found. No allegation of wrongdoing is made against any other individual or entity named.

Know more. Risk less. Decide better.

How we built this: approach, method and safeguards

The approach: ledger first, argument second. Before a word of this article was written, every statistic went into the InsidEntity IPO Evidence Ledger. It holds 20 US IPO cohort rows from 2005 to 2024 (complete three-year outcomes through 2022, early readings for 2023 and 2024), the long-run scoreboard, the cross-sectional splits, the survival research and the South African evidence. The argument follows the ledger, and every historical statistic in the article comes from it. This year’s listings are covered separately, in a follow-up, The 2026 IPO Test.

The method

Element	How it is measured
Population	US operating-company IPOs on the NYSE, NYSE American and Nasdaq. Excludes ADRs, unit offers, SPACs, REITs, closed-end funds, natural-resource partnerships, banks and savings institutions, small best-efforts offers, and penny stocks (offer price below $5). The low-price and SPAC comparisons in section 4 come from Ritter’s separate tables for those groups and are labelled as such.
First-day return	Offer price to first closing price.
Three-year return	Buy-and-hold from the first closing price to the earlier of the three-year anniversary or delisting, dividends included.
Market-adjusted return	The IPO’s three-year return minus the Morningstar US Total Market Index over the same window.
Averages	Equal-weighted: a $50 million listing counts the same as a $5 billion one, unless stated.
Data windows	First-day data runs through 2025. Three-year data runs through the 2024 cohort, measured to 31 December 2025.
South Africa	A separate study with its own sample, benchmark and method. It is never merged with the US figures.

The safeguards

A dataset this large makes it easy to say more than the evidence supports. These are the rules we applied to avoid that.

Primary source only. US figures come from Jay Ritter’s Initial Public Offerings: Updated Statistics, 14 September 2026 edition, read directly. No historical figure is taken from a news article or summary.
Every row carries a status. Each claim in the ledger is marked verified from the primary file, verified from the publisher’s page, or body-only. Figures that appear only inside paywalled papers are not quoted.
The numbers must reconcile. Seven cross-checks run inside the ledger, including the 20 cohort rows, the total of 2,352 IPOs and the full-sample three-year return. All seven pass.
Nothing is filled in. Three-year results for the 2025 cohort do not exist yet, and no 2026 cohort data exists at all. The article says so rather than estimating.
Each figure travels with its context. The population, period and method sit beside every major statistic, so a 1980-2024 US average is never read as a South African one.
Averages come with medians. IPO returns are highly skewed. Where the mean and median tell different stories, both are shown.
Delisting is not treated as failure. An acquisition ends a listing, often on good terms for shareholders. The article keeps voluntary and involuntary exits apart.
Patterns are not causes. When a group of IPOs did worse, the evidence shows an association. It does not prove the feature caused the result.
Known discrepancies are disclosed. Ritter’s tables differ by one IPO in a few years (2017: 106 vs 107; 2024: 72 vs 73) as records are refined. The ledger uses one table consistently.

What survivorship does not distort. Returns run to delisting, not just to the three-year mark. Companies that disappear are counted up to the day they leave, not dropped from the sample.

What remains a limitation. Most of the evidence is American. The benchmark choice changes the size of the gap, as the South African study shows. And a historical distribution cannot tell you how any single IPO will turn out.

Sources

Every historical statistic in this article traces to a row in the InsidEntity IPO Evidence Ledger, 2005-2024 (prepared 15 September 2026).

Datasets and research

Ritter, J. R., Initial Public Offerings: Updated Statistics, University of Florida, 14 September 2026 (Tables 1, 1a, 5, 13a, 15c, 16a, 16b, 16d, 16e, 16f, 17, 18a, 18b, 18d, 19, 19b, 19c, 19e, 20, 21, 22, 23, 24)
Gippel, J., Linnenluecke, M. K., Smith, T., Xu, Y. and Zhu, Y.-S., The Lifecycle of Typical IPOs: The Characteristics of Surviving Firms, Accounting & Finance (2026); findings taken from the abstract
Gounopoulos, D., Loukopoulos, G., Loukopoulos, P. and Wood, G., Do CEOs With a Financial Background Matter for the Success of Newly Public Firms?, Journal of Business Finance & Accounting (2026); findings taken from the abstract
Neneh, B. N. and Smit, Do IPOs underperform in the long run? Evidence from the Johannesburg Securities Exchange (JSE), African Review of Economics and Finance 6(2), 2014
Ritter, J. R. and Welch, I. (2002), A Review of IPO Activity, Pricing, and Allocations, Journal of Finance; and Gao, X., Ritter, J. R. and Zhu, Z. (2013), Where Have All the IPOs Gone?, Journal of Financial and Quantitative Analysis (both via Ritter 2026)
US SEC, Luckin Coffee Agrees to Pay $180 Million Penalty to Settle Accounting Fraud Charges, press release 2020-319, 16 December 2020
US SEC, Investor Bulletin: Investing in an IPO
JSE, FSCA greenlights JSE simplification project, January 2026