Finding · computed August 5, 2026 from 517 transitions
A lettered round is a markup. A hyphenated sub-round is a bridge.
Across 130 companies' charter histories, the label a round carries predicts its price move. Series D behaves differently from Series C-3, and the gap is large enough to be useful.
| Destination series | n | Median multiple | Mean | Down rounds |
|---|---|---|---|---|
| Lettered (A, B, … D) | 349 | 1.58× | 2.09× | 9.2% (32) |
| Hyphenated (C-1, D-2, E-1) | 167 | 1.08× | 1.44× | 25.1% (42) |
Multiple = price per share of the destination series divided by the price per share of the series immediately before it, both taken from the Original Issue Price definitions in the company's own certificate of incorporation.
The sharpest case: a letter stepping into its own sub-round
Restrict the measurement to the 65 transitions where a plain letter is followed by that same letter's first sub-round (C → C-1, E → E-1). Median 1.00×, and 30.8% are down rounds. This is the shape the market reads as "they could not clear the next letter".
What it means, and what it does not
A sub-round is priced roughly at the prior round: a median of 1.08× is a flat extension of existing terms, which is what a bridge is. A new letter is priced as a raise: 1.58×. So a company's own naming carries information about the transaction before any term is disclosed, and it is available to anyone reading a charter.
It does not mean a sub-round is bad news about the business. Extensions are routinely used to add an investor at agreed terms, to bring in a strategic partner, or to close a tranche. The measurement is about the price move, and nothing here identifies why any individual round was priced as it was.
The limit on this finding, stated here rather than in a footnote
Every company measured here reached a registration statement with the SEC. That population is the survivors. A company that priced a bridge round and then wound down never files a charter exhibit, so it never enters this denominator. The bias therefore runs against the finding: the real private market almost certainly carries more sub-rounds, and more of them down, than a sample of eventual registrants can show. Read every figure below as a floor, not an estimate of the whole market.
Two smaller limits, for completeness. A charter states the Original Issue Price of each series, which is a price per share and not a valuation: it is comparable across that company's own rounds, which is what is measured, and it is not comparable across companies without a share count. And where a company filed more than once, the filing carrying the most priced rounds is used and the others discarded, so no company is counted twice.
The rows that were excluded, and why the finding does not depend on them
16 of 533 transitions are excluded as unit errors: a move outside 0.05×–20.0× in a single round is not a price move, it is a par value or a share-class mismatch read as a price. One charter yields an A-1 of $0.0000125 followed by a B at $1.15, which is arithmetically a 91,796× markup and is obviously nothing of the kind. The affected companies are ACV Auctions Inc., Angion Biomedica Corp., Braze, Inc., Castle Creek Biosciences, Inc., CoreWeave, Inc., Fervo Energy Co, Sonendo, Inc., UiPath, Inc., Weave Communications, Inc., and their other rounds are unaffected.
The finding is unchanged either way. Including those rows the lettered median is 1.59× and the hyphenated 1.08×; excluding them it is 1.58× and 1.08×. A result that survives the removal of its own suspect rows is the only kind worth publishing, which is why the count is stated here rather than buried.
Reproducing it
The round histories are in the published corpus, one row per company, each carrying the accession number and URL of the charter exhibit it was read from. The classification rule is exact: a destination series matching a plain letter is lettered, one matching a letter followed by a hyphen and a number is hyphenated. Every transition where either price is absent is dropped rather than treated as zero.