Playbook › Playbook
A non-GAAP definition change leaves a fingerprint — the same prior period filed twice
Claim
Non-GAAP metrics are not defined by GAAP, so a company can change one. What it cannot do silently is present the comparative — the prior-year column sits right next to the current one and must be on the same basis. That obligation is the audit trail.
The two-step method
Step 1 — date the change from the WORDING, not the numbers
Pull the definitional paragraph of the non-GAAP metric from consecutive earnings releases and diff the exclusion list. A definition change shows up in language before it shows up in a number a reader would notice.
HLNE's Fee-Related Earnings, isolated to one filing:
| Release | Definition |
|---|---|
| Q3 FY25 (2025-02-04) | "earnings from recurring management fees… excluding (a) incentive fees and related compensation, (b) interest, (c) tax, (d) equity in income of investees, (e) non-operating gain/loss, (f) other" |
| Q4 FY25 (2025-05-29) onward | "earnings from revenues that are measured and received on a recurring basis… excluding (a) incentive fees, net of fee related performance revenues, and related compensation, (b) equity-based compensation…" |
Two changes, made simultaneously, on one date: a new revenue category folded in and stock-based compensation folded out. Stable since.
Step 2 — size it from the same period filed twice
| HLNE, 9M ended 31 Dec 2024 | Reported FRE |
|---|---|
| As originally filed (10-Q, 2025-02-04, old definition) | $165,789k |
| As restated comparative (10-Q, 2026-02-03, new definition) | $186,115k |
| Difference | +$20,326k (+12.3%) |
And it reconciles to the dollar against exactly the changed items:
``` +19,681 SBC added back +1,228 FRPR folded in −583 incentive-fee-comp reclassification
+20,326 ```
Same company, same nine months, two numbers 12.3% apart. No modelling, no assumptions — two filings and a subtraction.
Why it matters more than it looks
At HLNE the change converted a declining metric into a fast-growing one:
| 9M window | Reported | Restated to one conservative basis |
|---|---|---|
| Growth | +36.8% | −16.0% |
| Margin | 59.3% | 32.3% |
The short seller's headline claim — "FRE would have declined ~16% instead of growing 37%" — checked out to the decimal. The company never rebutted the arithmetic; it attacked the messenger's motives.
⚠️ Two guards against overstating the finding
- A conservative restatement is usually asymmetric — say so. Stripping the revenue a company added while keeping the compensation earned on it understates the true figure. At HLNE, fee-related comp rose 44% on the back of the new revenue line. −16% is the most punitive defensible construction, not the only one, and the honest core figure sat between the two. Report the range, not the worst case as fact.
- Separate the metric from the cash. The definition change touched a supplemental metric. HLNE's GAAP diluted EPS and its $419.1M of FY26 FCF were unaffected and verified. The damage is to the growth-and-margin-expansion narrative — which is what a premium multiple pays for — not to solvency or the dividend. Conflating the two overstates the bear case.
The denominator trick that hides it in plain sight
HLNE's reported FRE margin series read 48% → 50% → 53% — nothing like the alleged 32% → 59% jump, which is why the allegation initially looked refuted. They were the same period on two denominators: the headline series divides by total fee-related revenue including the new category; the allegation divides by management fees only. Before concluding a margin series contradicts a claim, confirm both are on the same denominator.
Where to look
- The non-GAAP reconciliation table in each 8-K Exhibit 99.1, quarter by quarter.
- The definitional footnote beneath it — this is the one that changes.
- The prior-period comparative column in the first filing after a suspected change.
- Note that exhibit presentation pages are frequently embedded images, so a text search of the filing can return nothing while the figure sits in the deck. Absence from the text layer is not absence from the record — but it does mean unaudited and unverifiable, which is worth stating.
Rule
Whenever a company leans on a non-GAAP metric in its own narrative, diff the definitional footnote across two years of releases before modelling the metric. If the wording moved, find the restated comparative and size it. This is ~20 minutes of filing reads and it can invert a growth rate.
Related
[[pitfall-up-c-structure-corrupts-six-vendor-fields]] — the other half of the HLNE analysis · [[pitfall-unrealized-equity-marks-break-headline-pe]] — the GAAP twin: same instinct, different line · [[pitfall-gate-cleared-by-adjective-not-arithmetic]] · [[principle-primary-source-beats-vendor]].
History
- 2026-08-12 — filed during
/analyze HLNE, testing the 2026-04-27 Hunterbrook short thesis.staticbecause it follows from the comparative-presentation obligation, not from any vendor's behaviour or any one company.