Perella Weinberg (PWP): the screen says broken, the cash flow disagrees

Run Perella Weinberg through any standard quality screen and it fails on arrival: negative book equity, negative GAAP EPS in three of the last five years, and a trailing multiple that looks expensive on the one year that was positive. A screener has no reason to look twice. That mechanical rejection — not obscurity, the firm's name is on the door of one of the better-known advisory boutiques — is the specific form the coverage gap takes here.

Coverage Gap

PWP is a $1.55B market cap

in a sector where attention concentrates on banks fifty times its size. But the more interesting barrier is structural: the company reports negative stockholders' equity (-$127.4M at FY2025 year-end)

which zeroes out any ROE-based screen, and its GAAP earnings line swings from loss to profit in ways that make trailing P/E either undefined or alarming-looking. A quantitative filter doesn't ask why equity is negative. It just moves on. If the reason turns out to be benign — a holding-company structure that routes most income and equity to partnership unit holders rather than to the listed share class — then the screens are discarding the name on an accounting artifact, and almost nobody downstream of those screens ever re-examines it.

That "if" is genuine: the cache used for this analysis carries consolidated XBRL figures only, and does not itemize non-controlling interests or the equity-award amortization that a definitive read would require. The 10-K's footnotes are the test this thesis has to pass.

Business

Perella Weinberg sells advice — M&A, restructuring, capital solutions — and its revenue moves with the deal cycle: $801.7M in FY2021's M&A boom, down to $631.5M in FY2022, $648.7M in FY2023, then a record $878.0M in FY2024, and $750.9M in FY2025

There is no recurring-revenue base here and no pretense of one; roughly 700 employees [FMP] company profile generate fees deal by deal. Anyone buying this name is accepting cycle exposure — the question this analysis cares about is whether the market is charging a second, unwarranted discount on top of the cyclical one.

Moat / Mispricing

The mispricing candidate is the persistent, violent disagreement between the GAAP earnings line and the cash flow statement. FY2024 is the clearest case: record revenue of $878.0M produced a reported net loss of $64.7M — while operating cash flow was positive $223.4M

A business that loses money operationally does not usually throw off nine figures of operating cash in the same year.

The pattern is consistent with an Up-C-style partnership conversion: substantial income attributed to non-controlling unit holders, heavy non-cash equity-award amortization from the go-public transaction, and a share count that changes as units exchange into the listed class — weighted diluted shares jumped from 53.2M (FY2024) to 100.8M (FY2025) in this dataset, which is not what organic dilution looks like. But "consistent with" is the strongest claim the cache supports. The chart above is the shape of the hypothesis, not its proof; the 10-K's NCI and compensation footnotes are where it gets confirmed or killed. Note also what the chart does not say: operating cash flow is lumpy (negative in FY2022, only $34.8M in FY2025), so this is not a smooth cash machine — the claim is narrower, that the GAAP line alone systematically misstates the economics in both directions.

Capital Allocation

Book equity went from +$271.4M (FY2021) to -$127.4M (FY2025) — and the path was not monotonic: it bottomed at -$421.4M in FY2024 before recovering roughly $294M in FY2025

For most companies that trajectory means losses are eating the balance sheet. Here, the five years of GAAP results net out to roughly -$38M (three loss years partly offset by the FY2022 and FY2025 profits) — nowhere near enough to explain a ~$399M equity decline. The remainder has to be distributions, buybacks, or reclassifications between the partnership and the listed entity, which would make the negative equity a choice about where capital sits rather than evidence of destruction. The cache cannot distinguish among those explanations — flagging that squarely: this is the section of the eventual 10-K reading with the most riding on it. The company also pays a dividend (last declared $0.28/share)

which loss-making businesses in genuine distress typically cannot sustain.

Risks

Checking the desk's standing blind spots against this name:

• Key-person risk (the big one here): an advisory boutique is its senior bankers. Revenue walks out the door in a way that never shows up in a balance-sheet screen. This desk has historically caught governance and key-person risk late, and nothing in the quantitative cache addresses retention, comp structure, or founder succession. This risk is elevated, not merely present, for this business model.
• Technology-speed risk: lower for an advisory franchise than for a product company, but not zero — deal sourcing and analysis are exactly the white-collar tasks being automated fastest.
• TAM optimism: the temptation is to model a permanent step-up from the FY2024 record. The five-year revenue series argues otherwise — FY2024 looks like a cycle peak until proven structural, and FY2025's 14% decline already argues for the cycle reading.
• Beta caveat: at 1.65 [FMP] profile data , the market already trades this as a high-beta cyclical. The thesis disputes the accounting discount, not the cyclical one.

Leading signals for the scheduled reviews: at 3 months, the next 10-Q's non-controlling-interest split — does it confirm the decomposition this thesis assumes; at 6 months, senior-banker headcount and any announced partner departures (the key-person risk has no balance-sheet signature, so it has to be watched in filings and press); at 12 months, whether revenue holds above the FY2022-FY2023 trough (~$630-650M) and FY2026 stays GAAP-profitable — if either breaks, the cycle reading beat the accounting-artifact reading.

Valuation

At $16.47 against FY2025 diluted EPS of $0.47, trailing P/E is roughly 35x — which looks expensive, and illustrates the trap in both directions: GAAP EPS here is suppressed by the same structural items that produce the losses in other years, so the 35x is likely overstated as a measure of what a Class A share actually earns. But quantifying the "true" earnings power requires allocating income between unit holders and shareholders, which the consolidated cache cannot do. The honest statement is that no multiple computed from these consolidated GAAP figures — neither the scary 35x nor some adjusted alternative — is decision-grade until the NCI split is read out of the 10-K. What can be said: a 7% gap between reported market cap and price-times-diluted-shares in this dataset is itself a symptom of the dual-class structure that creates the whole setup.

Re-rating Catalyst

Two mechanical paths. First, unit exchanges: as partnership units convert into Class A shares (the 53.2M → 100.8M diluted-share move suggests this is already well underway), more of the income statement lands on the listed share class, and the reported EPS line starts resembling economic reality — at which point screens stop auto-rejecting the name. Second, the cycle: if advisory revenue re-approaches the FY2024 level while GAAP optics are normalizing, the market would be repricing two discounts at once. Neither path requires anyone to "discover" PWP; both just require the accounting fog to clear on schedule.