Earnings Examiner

Measures: daily closing price with filing-derived markers — earnings (8-K 2.02, or the results 6-K for foreign private issuers), 10-Q/10-K filing dates, dilution (8-K 3.02, 424B5 takedowns, shelf registrations), debt events, and tier-1/2 distress signals; the legend filters by marker type. The dated stat block carries last close, market cap, % since last earnings / periodic, 52-week high/low and drawdown, 30- and 90-day realized volatility, beta, and debt / market cap. Multiples are filings-first: trailing P/E on LTM diluted EPS (via the shared LTM rule, “n/m” where EPS ≤ 0), plus EV / house EBITDA, EV / RPO and price / LTM sales, each basis-tagged. S&P 500 is an optional legend-toggled line, rebased to 100.

Source: daily closes from the Yahoo v8 chart API (keyless, unofficial) persisted to data/prices/ and shipped as prices/<ticker>.json; markers from the cached SEC submissions + the curated events log; EPS / sales / RPO / EBITDA from XBRL. “Close as of <date>” (weekly refresh; up to five sessions old), never live.

Limitations: market data is context — dated, and NEVER an input to any leverage, coverage, liquidity or runway figure. A price is a fact; a multiple is an opinion — the two are styled differently. Guidance-implied multiples are pending citation (curated). Vendor forward P/E is off by design. Unmatched 424B5 takedowns are flagged for a human to classify (debt vs equity), never silently placed. No composite score.

Measures: outstanding principal per facility, ordered senior-secured → convertible (→ preferred/mezzanine where present). The grouping toggle recolors the same bars by seniority, by recourse (who the lender can reach), or by rate type — the numbers never move. Book equity and market cap are shown as reference markers on their own scale in the strip beneath the chart (never as bars on the facility axis, which would crush the facilities). Market cap renders only when a dated <ticker>_market_cap series exists (script-refreshed weekly via reviewed PR; no page-load feeds). The recourse split groups principal by who the lender can reach — recourse (full + springing-to-parent) vs non-recourse (collateral-only); “limited” (capped guaranties) and unclassified principal are shown on their own lines, never folded into either total.

Source: registry (each entity’s debt footnote) + events (8-K/6-K). Dashed-border bars are secondary-sourced stubs. Recourse is curated from each facility’s guaranty/obligor disclosure (CRWV’s Q2-2026 recourse/non-recourse balance-sheet split is the template).

Limitations: book equity is not a recovery estimate; contractual spreads not restated in the 10-Q are shown as effective rates. Where a single filed line is a disclosed mix (CRWV’s OEM financing: recourse + non-recourse legs) it badges “mixed” and splits per its disclosed breakdown.

Measures: facility principal bucketed by maturity year from the hand-curated registry, split bullet vs amortizing/installment. Hatched = amortizing facilities with no disclosed schedule (full principal shown at maturity until two registry snapshots allow a paydown estimate). Diamonds = the company’s own filed aggregate principal-payment schedule, which allocates amortization to the years it is actually paid (suppressed when the filer’s schedule is fiscal-basis and would mislabel calendar buckets). Where the entity carries a curated contracted-inflows schedule, bars below the axis are the facing inflow leg on its own independent scale (a legend toggle, never sharing the debt axis): contracted receipts by calendar year (solid = a filed year-by-year schedule; hatched = coarse disclosed windows mapped to their majority calendar year — hover for the filing’s own wording).

Source: the entity’s latest periodic filing + post-quarter event filings; per-facility citations live in the registry (data/facilities/). Inflow rows are hand-curated from the revenue / lessor notes (data/overrides/) and reconciled against the XBRL contracted-revenue instant on every build.

Limitations: registry buckets show contractual maturities, not paydown timing; EUR notes excluded from USD totals until a company-reported USD figure exists; NBIS accreting converts show accreted-at-maturity amounts. The inflow leg is contracted revenue, not cash available for debt service — it carries delivery costs, definitions differ per entity (labeled inline), deposits already collected sit inside RPO, and fiscal-basis schedules are majority-mapped to calendar buckets. The reported/pro-forma toggle moves only the outflow side: there is no pro-forma fold for contracts. Announced amounts (not yet in RPO) render as chips only, never bars. Counterparty-dependence chips (⚑) name the credit the inflow leg leans on — obligor entity, guaranties, letters of credit — hand-curated per filing; they never haircut an amount. No net or “gap” figure is computed by design.

Measures: (1) pre-debt pipeline — liabilities for property & equipment additions (incl. OEM-financed) that typically get reclassified into debt once vendor financing is papered; a growing balance is debt the Note-10 table doesn’t show yet (watch flag: >25% growth and >$250M). (2) unconsolidated JV/VIE obligations — commitments to entities whose project debt is off the balance sheet. (3) financing obligations outside the debt table — e.g. failed-derecognition asset sales sitting in other liabilities.

Source: XBRL cash-flow supplement (automated) + hand-curated entries from the VIE, related-party and DCSP footnotes; all primary quality.

Limitations: the pre-debt comparative is the period the filer discloses (CRWV: year-ago quarter, not sequential). JV exposure components are maximums, not expected values, and exclude the JV’s own third-party debt, which is undisclosed. None of these amounts are added into gross debt or leverage — they are rendered alongside, per the no-blending rule.

Measures: the logged debt events after the balance-sheet date — date, type (issuance / repayment / amendment / draw / conversion / maturity), facility, signed Δ principal, source, and status (active folds into pro-forma; absorbed is already inside the new balance sheet).

Source: hand-curated from each event’s 8-K / 6-K / 424B (data/events); each row carries its filing citation.

Limitations: the pro-forma view = reported + the sum of active event Δs (a pure fold; the registry is never mutated). EUR-denominated events stay in EUR until a company-reported USD figure exists.

Measures: this filer’s own row of the counterparty matrix — the disclosed customer / tenant / lender exposures that name a counterparty: contract values (TCV, $), revenue and receivables concentration (%), lease tenancy (MW), equity stakes, and aggregate-RPO cells (customers unnamed — a disclosure-gap marker). The full cross-entity matrix and the systemic counterparty graph live in the Sector view; this tile is just the current name.

Source: the customer-concentration and contract-value footnotes for this entity (data/counterparties.yaml); every cell carries its own as-of date and source quality.

Limitations: “up to” TCVs are ceilings, not bookings, and definitions differ per contract (each cell’s hover note states its own basis). Filers rarely name their top customers — unnamed concentration is an aggregate-RPO cell, never allocated. A relationship disclosed without a magnitude reads “disclosed, magnitude not”. Nothing here feeds leverage or coverage math. ◎ = secondary-sourced.

Measures: per-facility financial covenants and springing conditions — covenant type (leverage, DSCR, min-liquidity, interest-coverage, springing), the disclosed threshold, its measurement basis, any springing/prepayment condition, the compliance-certificate cadence, and a status (compliant / not tested / breach / unknown). A trigger map, never scored or blended.

Source: each entity’s credit-agreement and debt footnotes (data/covenants/<ticker>.yaml); every row carries its own as-of date and filing citation.

Limitations: qualitative entries are expected — the register maps where a facility can trip, not a computed ratio. Status is “unknown” where the filing states the covenant but not an explicit compliance affirmation, and “not tested” where the facility is undrawn; neither is a pass. A disclosed breach also surfaces as a page-level flag.

Measures: one stacked bar per as-filed balance-sheet observation, gross funded debt split by seniority class (same palette and ordering as the capital stack) at the current observation; earlier points render single-source (the XBRL debt total, no seniority split — the curated registry exists only at the current quarter). Preferred/mezzanine renders as its own marker beside the bar, never inside it. The EBITDA overlay is a band between the LTM and annualized-latest-quarter house variants — the band’s width is the hypergrowth flattery, the current-quarter footnote flag made visible. Leverage is annotated per bar (both variants, compact; a dash where a variant is unavailable).

Source: reconstructed purely from the committed SEC XBRL companyfacts cache (one genuine observation per debt-dated balance sheet — no interpolation, no invented interior history) merged with the current curated registry observation. The reported/pro-forma toggle moves only the current bar (post-quarter events have no historical analogue).

Limitations: where a filer hasn’t tagged enough discrete quarters for an LTM (or a fiscal-year point has no discrete Q4), the band collapses to its available bound and the panel says so; EUR notes follow the existing rule (excluded until a USD translation is reported).

Measures: a single-axis bridge from funded debt up through the present values of operating- and finance-lease liabilities to lease-inclusive obligations, with the aggregate undiscounted not-yet-commenced lease payments drawn as a separate outlined marker (never a stacked step — it is undiscounted future cash, a different basis). A reference line marks the contracted-revenue floor. The folded table below carries contract coverage: gross debt per $1 of contracted revenue, and the funded-debt-to-lease-inclusive breakdown. Two ratios are shown side by side: gross-debt-per-contracted-dollar and lease-inclusive-obligations-per-contracted-dollar — the latter flagged mixed-basis numerator because it blends present-value and undiscounted layers.

Source: the entity’s balance-sheet lease liabilities (XBRL), the hand-entered not-yet-commenced aggregate (narrative-only in the filing, primary quality), and the contracted-revenue tag per the tag map.

Limitations: present-value and undiscounted layers are never merged into one clean number; the outlined marker is the visible basis separation. Contracted-revenue definitions are NOT comparable across entities — each entity labels its own numerator inline. Disclosed-but-excluded items (e.g. CRWV’s separately disclosed 393 MW site with a not-yet-commenced liability up to $14.7B) render as chips, never bars — the same announced≠contracted discipline.

Measures: per-period gross interest incurred, split expensed vs capitalized — the capitalized share sits in capex and property, not the income statement, so cash-interest coverage understates the true interest burden. The two-variant EBITDA band from the trajectory panel overlays it, and cash-interest coverage is annotated per period.

Source: reconstructed from the committed XBRL companyfacts cache (interest and capitalized-interest tags per the entity’s tag map, on its verified gross/net basis) — one discrete-period observation per filing, no interpolation.

Limitations: no projected series — the capitalization roll-off is conveyed by the historical trend only; the chart never draws a future bar. Bars are at true period length; a fresh-10-K or annual-cadence filer’s latest figure is a fiscal-year value labeled “(FY)” and kept off the quarterly axis.

Measures: a dumbbell per contract-linked facility — a dot at outstanding principal and a dot at the remaining contracted receipts disclosed as backing it, connected. Structurally unbacked debt (senior unsecured, converts, revolver) is grouped and subtotaled. The headline is the unmatched share of gross debt: principal not covered by a disclosed receipt figure, with its quarter-over-quarter delta once the snapshot archive holds a second observation.

Source: the curated facility field counterparty_backing — facility/SPV-level RPO allocations or disclosed contract values, cited per filing.

Limitations: if backing receipts are undisclosed for a facility, the receipts dot is absent and the row is marked “backing undisclosed” — never estimated. Contract-backed is kept distinct from the recourse class (a facility can be non-recourse yet unmatched, or full-recourse yet fully backed); both render as row badges. A contract shared across facilities is counted once in the aggregate backing total. When no facility discloses backing receipts, this tile collapses to a one-line status inside the capital stack.

Measures: a presentation bridge of disclosed components drawn as a true waterfall — house EBITDA, then stock-based comp (non-cash), the change in customer deposits / deferred revenue (an operating inflow that behaves like prepayment financing), any curated working-capital delta, and an explicit residual (taxes and other non-cash the itemized legs don’t capture) landing on the filed operating cash flow; then less disclosed capex to free cash flow. Every leg in the table is a step in the chart. Quality context: capitalized interest and accrued-but-unpaid capex.

Source: XBRL cash-flow statement (OCF, SBC, capex), the contract-liability balance for the deposit delta, income-statement tags for EBITDA — all read over the cash-flow statement’s own period window so no leg mixes a discrete quarter with a YTD figure; curated legs (working capital, accrued capex) from data/overrides.

Limitations: OCF and FCF are the filed anchors; the residual is disclosed-minus-itemized, never an estimate, and is never distributed back into the named legs. The bridge is never collapsed into one blended number. A leg absent for the period is omitted (the residual absorbs it), not guessed.

Measures: (1) automated direction screens from XBRL — trade-receivables growth vs revenue growth (AR compounding faster than revenue is the classic pulled-forward-sales flag), DSO and inventory days vs their year-ago period (direction over level, per the checklist), GAAP gross margin, latest-FY NRV write-downs, and customer-deposit contract liabilities. (2) hand-curated watch items — one row per checklist concern (reserve recycling, take-or-pay RPO vs bookings talk, customer circularity, rebates, concentration, SBC/useful lives), each carrying its own as-of and the exact filing language it reads.

Source: XBRL companyfacts for the screens; the receivables/inventory/revenue notes, MD&A NRV tables, and strategic-agreement disclosures for the watch items (data/overrides/<ticker>.yaml quality_watch). All display-only: nothing here enters leverage math, and there is no composite score by design.

Limitations: days metrics use quarterly flows × 91 days and distort when margins move violently (a COGS collapse shrinks inventory days without a unit change); year-ago comparisons use the filer’s 52/53-week calendar (±10-day match). Channel/distributor inventory — where a memory correction shows first — is outside SEC filings entirely; cross-check distributor prints (ARW/AVT/WPG/WT) per the checklist. NRV write-down detail is annual-only (10-K MD&A), so a mid-year reserve release will not show here until the 10-K.

Measures: AFUDC share of net income (equity vs debt AFUDC where split — the SCANA signature signal; rising = paper earnings during construction); CWIP and regulatory-asset growth vs rate base; FFO/debt against external agency reference bands; securitization / rate-recovery bonds segregated on their own row; holdco vs opco debt for the structural-subordination view.

Source: AFUDC / CWIP / regulatory-asset / rate-base / FFO figures from the utility’s filings, curated per entity; gross debt from the registry.

Limitations: the FFO/debt bands are external agency reference thresholds, clearly labeled — not a house score; securitization debt is never inside corporate leverage; equity units / hybrids stay in the mezzanine framework. This tile is gated on credit_shape: utility and stays hidden until a Cohort E name onboards with a registry.