SPACs and real IPOs don't belong in the same statistic
Most "average IPO return" numbers blend blank-check shells with operating companies. The two cohorts behave nothing alike — and the blended number describes neither.
Look up "average first-day IPO pop" and you will find a confident number. What you will rarely find is what that number is averaging — and in most recent windows, a large share of it is not operating companies going public at all. It is blank-check shells.
Two things wearing one label
A special purpose acquisition company (SPAC) raises money first and finds a business later. Its units price at $10 by construction, and until a merger target appears, the stock is essentially a box of cash plus a warrant. There is no revenue, no product, and no reason for the price to move much in either direction. Most don't: they drift within a few percent of trust value while the sponsor hunts for a deal.
An operating-company IPO is the thing people picture when they hear "IPO": a real business with revenue, filings, roadshows, and a price discovered by actual demand. These listings can gap 40% on the open, run for weeks, collapse after lockup expiry — the entire dramatic repertoire IPOs are famous for.
Both show up in the same exchange listing feeds, on the same days, under the same "IPO" tag. If you pool them into one average, you get a number that describes neither.
What the split looks like in practice
When we rebuilt the peak model behind the Recent IPOs page, the recent US cohort broke down roughly like this: about half of newly priced listings were SPAC units sitting at exactly $10.00. The median SPAC peaked about 1.3% above issue. The median operating company peaked about 43% above issue.
A blended median across both cohorts landed in the single digits — a number that overstates what a SPAC does by several multiples and understates what a real listing does by a factor of five or more. Splitting the cohorts wasn't a refinement; the split explained more of the variance in peak outcomes than any sector-based grouping we tested. Sector, the dimension most IPO commentary leans on, explained less than a random partition of the same data.
Three quieter traps in IPO statistics
The SPAC blend is the loudest problem, but the same rebuild surfaced others worth knowing about whenever you read an IPO stat:
- Survivorship filters. Dropping listings with negative outcomes before computing "typical gain" — even accidentally, via a filter that looks innocent — inflates the estimate badly. In our data, one such filter overstated the typical result by more than half.
- Means instead of medians. One +148% biotech drags an average a long way. Medians with a p25–p75 band tell you what the typical listing did.
- Drawdown measured backwards. "High minus low over the window" is not the drawdown an investor experienced if the low happened before the high — which, for recent listings, it usually did. The honest measure walks forward from the peak.
What to do with this
If you evaluate a new listing, first ask which cohort it belongs to. A SPAC at $10.02 is not "underperforming the average IPO" — it is doing exactly what the SPAC cohort does. An operating company that opened flat is genuinely unusual, and worth a closer look.
The Recent IPOs tracker labels every listing as a SPAC or an operating company and models the two separately, with each figure carrying the number of listings behind it and a confidence score. And when a new listing graduates into the broader market, the screener picks it up with the same fundamentals as everything else — which is where the real question ("is this a business worth owning?") eventually gets answered.
None of this is a trading edge. It is a description of what the recent cohort has done, measured carefully. But most of the bad IPO decisions we see start with a statistic that was measured carelessly — and the SPAC blend is the most common way that happens.