Financial Shenanigans

The Forensic Verdict

TW's accounting reads cleanly. Seven straight years of operating cash flow exceeding GAAP net income, accounts receivable growing in line with revenue, capex/depreciation well-behaved, and no restatement, material weakness, auditor change, or SEC inquiry in primary filings or external research. Forensic risk score: 22 — Watch, not because anything is broken but because three structural features deserve underwriting — (1) LSEG is a 89.9%-voting controlling parent with a $93M related-party market-data contract, (2) FY2025 GAAP net income was inflated by a one-shot $270.9M Canton Coin mark-to-market gain that the company correctly excludes from Adjusted EBITDA but leaves embedded in headline earnings, and (3) the $860M ICD acquisition (Aug 2024) drove FY2024 free cash flow after M&A to roughly zero and added $335M of goodwill not yet tested under operating stress. What would change the grade: any disclosure that ICD goodwill is being tested for impairment, or any qualified / emphasis-of-matter language from Deloitte in the FY2026 10-K.

The 13-Shenanigan Scorecard

Five of thirteen tests register as Yellow, none as Red. The yellow flags are all structural — controlled-company status, capitalized software treatment, lumpy acquisition cash flows, balance-sheet line aggregation, and crypto mark-to-market — none of which suggest hidden deterioration.

Breeding Ground

The accounting context is shaped by one dominant feature: LSEG is the parent. Tradeweb is a controlled company with 89.9% of combined voting power held by LSEG/Refinitiv via a four-class share structure (Class B and D each carry 10 votes per share). Three of eleven directors are LSEG-designated. The audit committee, however, is fully independent and chaired by Steven Berns (former Shutterstock CFO), and Deloitte is the auditor with no resignation, no qualified opinion, and no material-weakness disclosure in the FY2025 10-K.

The breeding ground is moderately elevated by controlled-company status and the LSEG related-party contract, but dampened by a clean audit relationship, an independent audit committee, and conservative non-GAAP hygiene. Net read: structural risks worth knowing, none that would amplify a hidden accounting problem if one existed.

Earnings Quality

Reported earnings look earned and sustainable, with one large caveat: in FY2025 Other Income (Loss), Net of +$263.4M (driven by a $270.9M Canton Coin crypto mark-to-market) is responsible for the entire delta between +20% Adjusted EBITDA growth and +62% GAAP net income growth. The company's own Adjusted EBITDA correctly strips this out.

Revenue vs receivables — clean

The FY2023 receivables spike to $549.6M is a reader-trap, not an earnings issue: it is a wholesale clearing/settlement balance with a matching $409M payable to brokers and dealers — the two offset to near zero in cash effect. The MD&A's working-capital table separates "accounts receivable" from "receivable from brokers and dealers and clearing organizations"; on the customer-only line, FY2025 accounts receivable was $257.8M vs $222.3M in FY2024 (+16% vs revenue +19%).

The FY2023 DSO of about 70 days reflects timing of a year-end settlement spike. Excluding the broker/dealer clearing balance, customer DSO has been remarkably stable at 43-58 days for seven years. No revenue-pull signal.

Earnings vs cash earnings — the Canton Coin overhang

Canton Coin gains, which Tradeweb earns by running validator nodes and then marks to fair value through Other Income, contributed $270.9M to FY2025 GAAP net income. The Tax Receivable Agreement liability adjustment swung from a $9.5M expense in FY2023 to a $9.8M income credit in FY2025. Both items are properly classified below operating income, properly disclosed, and properly excluded from Adjusted EBITDA — but a reader staring at the headline 44.9% net income margin in FY2025 versus 33.0% in FY2024 will overstate the underlying business unless they back the items out. The right operating-margin reference is the Adjusted EBITDA margin: 54.0% in FY2025 vs 53.3% in FY2024, a much more modest 70-basis-point expansion.

Capex / D&A — well-behaved

Total cash spent on capex plus capitalized software ($103M in FY2025) is below D&A ($250M), but the gap is explained by around $176M of acquisition-related and Refinitiv pushdown amortization. Stripping that out, organic D&A is roughly $74M and capex+software ($103M) is reasonably around 1.4x — within normal range for a software-intensive trading-platform operator and consistent across the seven-year window.

Cash Flow Quality

Operating cash flow exceeds net income every year shown, FCF tracks CFO closely (low capex intensity), and working-capital changes are not the engine of CFO strength.

CFO/NI compressed from 2.12x (FY2021) to 1.27x (FY2025) — that compression is not a sign of weakening cash quality, it is the mechanical effect of GAAP net income being inflated by the $270.9M Canton Coin gain (which has zero cash impact). On a Canton-adjusted basis (excluding the non-cash gain and the corresponding tax), FY2025 CFO/NI is still firmly above 1.6x.

Acquisition-adjusted FCF — the ICD year was funded by accumulated cash

The chart is the test that fails for acquisitive compounders: FY2024 FCF after acquisitions was approximately zero because the $860M paid for ICD plus r8fin nearly absorbed the year's $898M of CFO. Tradeweb funded the deal from accumulated cash (the balance fell from $1.7B at YE23 to $1.3B at YE24) without drawing the revolver. FY2025 reset to $1.13B with no closed M&A. Multi-year acquisition-adjusted FCF is positive, but lumpy — the right cash-generation reference is the seven-year sum (around $3.5B), not any single year in isolation.

Working capital is not the engine

The bridge from net income to CFO is dominated by non-cash items (depreciation, amortization, stock-based comp, and in 2025 the Canton Coin gain reversal). Working-capital movements are a modest contribution in FY2025 (mostly accrued comp and income taxes payable timing). There is no payable-stretching, customer-prepayment, or receivable-sale lifeline in evidence.

Metric Hygiene

The non-GAAP architecture is conservative on two fronts that compounders often abuse: the assumed effective tax rate (25.0%) is higher than the actual rate (21.6%), and Adjusted EBITDA actually excludes the large Canton Coin gain rather than counting it. The two hygiene issues to flag are the $176M acquisition-amortization add-back inside Adjusted EBIT (recurring as long as M&A continues) and the gap between headline GAAP net income margin (44.9%) and operating economics (Adjusted EBITDA margin 54.0%) created by the Canton Coin gain.

The widening gap between the two lines in FY2025 is the Canton Coin gain. Adjusted EBITDA margin moved only 70 bps; GAAP net income margin moved 1,190 bps. Underwrite the green line.

What to Underwrite Next

This is not a checklist of generic diligence — these are the specific signals that would move the forensic grade in the next two reporting cycles.

What would downgrade the forensic grade: an ICD goodwill impairment, a Deloitte resignation, an SEC enforcement disclosure, a restatement, a material weakness, or evidence that the LSEG related-party contract was re-priced to Tradeweb's detriment on terms not reflective of arm's-length.

What would upgrade it: another year of clean cash conversion with the Canton Coin gain rolling off in the comparative base, ICD passing a goodwill impairment test, and Deloitte issuing a clean audit opinion with no critical-audit-matter additions.

Does the forensic work affect valuation, sizing, or margin of safety? Not the underwritten valuation multiple. It does justify two overlays: (a) when comparing 2025 GAAP earnings growth to peers, normalize for the $270.9M Canton Coin gain (~30% net income overstatement at the headline); and (b) for any DCF, use acquisition-adjusted FCF averaged over three years, not single-year reported FCF, because the ICD-year distortion is large enough to throw a single-year multiple off by an order of magnitude. Accounting risk is a footnote, not a thesis breaker.