Sentinel Credit

Basic Workflow

• Step 1: Enter company details (name, ticker, sector)
• Step 2: Fill in financial data from the most recent fiscal year or LTM (all values in $millions)
• Step 3: Choose qualitative mode — Auto-Estimate for quick analysis, or Manual for analyst-level precision
• Step 4: Click "Run Credit Rating Analysis"
• Step 5: Review the composite rating and drill into sub-agency scores

Key Data Sources

• Income Statement: Revenue, EBITDA, EBIT, Interest Expense, Net Income, D&A
• Cash Flow: Cash from Operations (CFO), Capital Expenditures (CapEx), Dividends
• Balance Sheet: Total Debt, Cash, Total Assets, Total Equity, Goodwill + Intangibles

Through-the-Cycle Considerations

Rating agencies use normalized (through-the-cycle) financials, not peak or trough numbers. For cyclical companies at earnings peaks or troughs, consider:

• Using 3-year average EBITDA for commodity or cyclical companies
• Adjusting for one-time items (restructuring, litigation, asset sales)
• Forward-looking EBITDA for companies undergoing transformation

Auto-Estimate Mode

Derives all 18 qualitative factors from financial metrics using calibrated heuristics. Best for rapid screening or when detailed company knowledge is unavailable. The algorithm considers sector classification, margin profiles, leverage levels, and balance sheet characteristics.

Manual Mode

Allows direct input of each qualitative factor. Recommended for final credit committee analysis where analyst judgment is critical. Key factors include:

• Industry Risk (1-6): Cyclicality, barriers to entry, competitive dynamics
• Competitive Advantage (1-6): Market position, brand strength, IP, switching costs
• Financial Policy (-1 to +3): Management's approach to leverage, dividends, M&A
• Comparable Rating Analysis (-2 to +3): Final analyst adjustment to align with peer ratings

Summary View

• Composite Rating: The final blended rating (S&P scale) — this is the primary output
• Composite Score: The underlying numeric score (lower = better, 1=AAA, 21=D)
• Agency Ratings: Individual S&P, Moody's, and Fitch model outputs
• Key Metrics: Leverage, coverage, and profitability ratios the model used

Drill-Down Detail

Click any section header to expand detailed scoring breakdowns:

• S&P Methodology: Business Risk Profile (BRP), Financial Risk Profile (FRP), Anchor, Modifiers
• Moody's Methodology: Business Profile, Financial Profile, Factor scores
• Fitch Methodology: Operating Profile, Financial Profile, Notching
• Model Adjustments: Scale ceilings, cyclical adjustments, goodwill penalties

Investment Grade

High Yield (Speculative)

• Data quality matters: Ensure EBITDA, EBIT, and D&A are internally consistent (EBITDA = EBIT + D&A)
• Negative equity: Companies with negative equity from share buybacks (e.g. McDonald's, Starbucks) are handled automatically with equity-deficit penalties
• Scale ceilings: Companies under $5B revenue face automatic rating ceilings that mirror agency methodology
• Goodwill drag: Companies with goodwill above 35% of assets receive a penalty reflecting acquisition risk
• Cyclical industries: At peak earnings, the model applies through-the-cycle adjustments to prevent inflated ratings

The Predictive Intelligence tab runs forward-looking distress and earnings-quality models on top of the rating composite: Altman Z-Score (financial-distress / bankruptcy probability), Ohlson O-Score, Beneish M-Score (earnings-manipulation), Piotroski F-Score (financial-strength composite), Merton structural distance-to-default, and a multi-year-trend overlay across DSSI / AAS / EEV / CFQD / IBA / Working-Capital Deterioration models. The output is a forward-looking "what does this company look like in 12-24 months" view that complements the point-in-time rating.

Use Predictive Intelligence when: (a) you are evaluating a credit you'll hold for >12 months and need trajectory signals, (b) you suspect earnings-quality issues that the rating composite may not fully reflect, (c) you need a single-screen distress-probability summary for an internal credit-committee.

The Credit Committee surfaces an 11-agent + Chair Agent verdict on every rating: Financial Statement, S&P Methodology, Moody's Methodology, Fitch Methodology, Working Capital, OBS-Forensic, Liquidity & Refinancing, Sector Specialist, Macro / News, Forward-Path, Predictive-Validity. The Chair Agent synthesizes the 11 verdicts into a consensus_confidence (HIGH / MEDIUM / LOW / VERY_LOW), confidence-band (direction + magnitude), divergence panel, and recommended action (auto-publish / flag-for-analyst / escalate-to-CCO).

The committee NEVER modifies the engine rating. It characterizes the engine's output for the IC memo's customer-facing committee section. Adverse-action notices (Reg B principal-reasons disclosure) are generated from Layer-1 deterministic kernel only — never from committee narrative.

The OBS-Forensic Agent reads disclosed footnotes for "supply-chain financing programs," "receivables financing arrangements," "trade-finance arrangements," and related language. Vague disclosure (existence-of-programs without quantitative scope) is itself flagged. Where vague SCF disclosure is detected, the engine applies a partial-credit haircut to the disclosed-debt baseline using peer-set median percentage-of-revenue calibration.

The WCF Analysis tab quantifies working-capital efficiency: CCC, DSO, DIO, DPO components, the Sentinel WCF Score (0–100), peer benchmarking, and the WCF Pro-Forma module that emits three implementable scenarios (Conservative / Target / Aggressive) with full credit-committee-relevant projections — adjusted leverage (reported and full-basis), coverage, liquidity cushion, projected rating impact. See the WCC and Apex case studies for end-to-end examples.

Standard non-financial-corporate ratios (Debt/EBITDA, EBITDA/IE, FFO/Debt) are nonsensical on banks. Sentinel implements a separate banks methodology anchored on the BICRA framework: a US country-anchor (currently A) plus a 6-dimension Bank Risk Score (BRS) modifier covering capital, deposits, duration, liquidity, asset quality, and profitability. The BRS-to-rating mapper (sentinel-banks-rater.js) translates the 0–100 composite to the S&P 21-notch scale.

The banks v0 methodology has a documented systematic bias of approximately −1.25 notches versus S&P consensus on US large banks. v1 calibration is in progress. Bank ratings should be treated as analytical illustrations only at this stage.

The Outcome Ledger logs every rating run with full provenance: input hash, output rating, exception trail, committee verdict, model card, calibration provenance. Used for audit-trail compliance under SR 11-7 model-risk-management and for internal back-test validation. The ledger is append-only and is exportable for external model-validation review.

The Drift Monitor compares the current model run against a baseline calibration period and surfaces any drift in input-distribution, rating-distribution, or default-rate-implied PD. Used for ongoing model-monitoring under SR 11-7. The monitor flags drift exceeding documented thresholds and recommends recalibration when warranted.

Customers with proprietary historical loan-performance or counterparty-default data can upload it (through the BYO Calibration tab) to fit a tenant-specific calibration overlay (BYOLH-1). The overlay adjusts the standard PD/LGD curves to the customer's empirical default experience without disturbing the underlying methodology. Used by lenders with sector-concentrated portfolios to improve the accuracy of CECL/IFRS9 PD estimates.

Upload a borrower's supplier list to surface supply-chain concentration risk: geographic concentration, single-name concentration, single-source dependency, OFAC-adjacent exposure. Used to evaluate working-capital risk amplifiers — the obligor may look healthy on a stand-alone basis but be exposed to supplier-side disruption that materially affects their continuing operations.

SR 11-7 (Federal Reserve / OCC) is the model-risk-management framework regulated banks use to govern internal and vendor-provided models. The SR 11-7 Evidence Pack tab documents the evidence Sentinel provides to institutional customers so their MRM groups can discharge their SR 11-7 obligations when they deploy Sentinel ratings: methodology documentation, calibration provenance, sample-adequacy gating, model card, governance-contract framework, audit-trail. The pack is updated on every model release.

The Sample Reports tab links to downloadable customer-facing artifacts: BA Committee Memo (sample IC memo with the v1.1 LLM-enriched 11-agent committee output), Sentinel Executive Briefing (PDF + PPTX). The Case Study tabs cover: Hertz (bankruptcy alert), Valeant (fraud alert), WCC (working-capital optimization for a public mid-cap distributor), Apex Industrial Supply (synthetic mid-market private-company buyer credit decision — the RapidRatings-customer-targeting case). Internal-only case studies on First Brands Group are available under separate NDA.

Agency Weights

• S&P Global: 35% weight — Business Risk Profile + Financial Risk Profile + Modifiers
• Moody's: 35% weight — Business Profile + Financial Profile factor grid
• Fitch: 30% weight — Operating Profile + Financial Profile + Notching

The composite numeric score = (S&P × 0.35) + (Moody's × 0.35) + (Fitch × 0.30), subject to post-blend adjustments.

Through-the-Cycle Cyclical Trough (v2.3)

When EBITDA margin < 5% and revenue > $5B in cyclical industries (IR ≥ 4), a −2 notch uplift is applied to approximate mid-cycle ratings. This prevents trough-year metrics from overstating credit risk.

Revenue-Based Scale Ceilings (v2.4, refined v2.7-v3.0.0)

• < $1B revenue: Ceiling BB- (13), or B+ (14) if D/EBITDA > 1.5
• $1-2B revenue: Ceiling BB- (13), or B+ (14) if margin < 5% or peak-cycle commodity (IR ≥ 4, D/EBITDA < 1.0)
• $2-3B revenue: Ceiling BB+ (11), or B+ (14) if leveraged cyclical (D/EBITDA > 4.0, IR ≥ 4)
• $3-5B revenue: Ceiling BBB- (10)

Balance Sheet Quality (v2.2 – v2.6)

• Goodwill drag: Goodwill ≥ 35% of assets adds +0.20 to composite
• Double jeopardy: Scale/scope = 6 AND goodwill ≥ 40% floors composite at B+
• Financial policy + leverage: FP ≥ 3 AND D/EBITDA > 3.0 adds +0.10

Backtest Results — Launch Baseline (2026-04-25)

The model has been validated against a 50-name proof-of-concept across U.S. corporate IG and HY issuers, fixed seed=42, against refreshed S&P 2026-Q2 agency ratings:

• Within 1 notch: 52% — raw three-agency blend before post-composite overlays. Wilson 95% CI: 38–66%.
• Within 3 notches: 76% — raw blend.
• MAE: 2.36 notches; Spearman ρ = 0.51.
• Adjusted (post-overlay) tier: within-1 36%, within-3 70%, MAE 2.90. Overlays are net-negative on the current sample (helped 10 / hurt 22 / neutral 18) and are disclosed; the customer-facing claim is the raw blend, not the adjusted tier.
• Coverage: 50 S&P 500 names spanning Industrials, Energy, Communications, Consumer Discretionary, Consumer Staples, Healthcare, Financials, Materials, Tech, and Utilities.
• Documented divergence clusters: on a documented set of legacy/cyclical mega-caps (telecom, post-COVID airlines, junk-rated auto OEMs, sub-mega-cap E&P, live-events platforms), the engine runs 5–10 notches less conservative than NRSRO consensus; on franchise/scale mega-caps (consumer staples, pharma, managed care), it runs 1–3 notches more conservative. Both clusters are disclosed in the rating payload's limitations[] array and in the Known Limitations page.
• Acceptance thresholds (production target): within-1 ≥ 65%, within-2 ≥ 85%, |μ±| ≤ 0.5. Current within-1 of 52% is below this target; the path to 65% requires structural changes to the qualitative-side scoring of legacy/cyclical mega-caps (industry_risk grid expansion), not additional overlays.
• Roadmap: 125-ticker stratified S&P 500 expansion (Q3 2026) tightens CI to ±5 pp; Russell 3000 10% sample (Q4 2026) extends coverage into small-cap and mid-cap.

The NEWS-1 adapter (sentinel-news-adapter.js) processes structured news events with a 12-event taxonomy: regulatory_action, regulatory_investigation, litigation_loss, litigation_pending, credit_event, covenant_waiver, debt_restructuring, m_a_target, m_a_acquirer_levered, commodity_shock, sovereign_event, strategic_alternatives. Each event has a documented credit-risk magnitude. Per-event cap −1.5 notches; aggregate cap −1.5 notches; one-sided downward only. Co-exists with the existing keyword classifier in sentinel-news-signal.js; the adapter provides a clean integration surface for production news feeds (Bloomberg / Reuters / structured-event APIs) bypassing keyword brittleness.

The REC-1 overlay (sentinel-recovery-overlay.js) addresses travel/leisure/cruise/airline names that are mid-recovery. Five-conjunction gate: sector ∈ travel/leisure/cruise/airline AND leverage > 5.0x AND CFO/revenue > 0.15 AND coverage 1.5x–4.0x AND NI/revenue < 0.01. Magnitude: −1 notch (rating UP). X-12b cap-aware. Empirically fires on CCL only; AAL and RCL excluded by gates. Panel S&P MAE 1.10 → 1.07; within-1 70.0% → 73.3%.

The MCS-1 overlay (sentinel-megacap-stability-overlay.js) addresses mega-scale retail / consumer-staples names whose AA rating reflects scale-driven cash-flow stability rather than high margins. Five-conjunction gate: sector ∈ retail/consumer-staples + revenue > $300B + market_cap > $200B + leverage < 3.0x + coverage > 6.0x. Magnitude: −2 notches; AA floor; X-12b cap-aware. Fires on WMT only. Panel S&P MAE 1.07 → 1.00 (broke through 1.0 floor); within-2 86.7% → 90.0%.

The SUS-1 overlay (sentinel-utility-stability-overlay.js) addresses regulated electric/gas utilities where the engine over-penalizes sector-typical 5–7x leverage. Discriminator: capex/revenue > 0.30 cleanly separates regulated utilities from oil & gas / energy services. Five-conjunction gate: sector contains energy/utility/electric/gas/power + capex/rev > 0.30 + EBITDA margin > 25% + coverage > 2.5x + leverage 4.0x–9.0x. Magnitude: −3 notches; A floor; X-12b cap-aware. Fires on NEE, DUK, SO, AEP, EXC. 100-panel S&P MAE 1.67 → 1.56.

The CCO-2 overlay (sentinel-concentration-conservatism-overlay.js) addresses engine-too-lenient names where agencies haircut for qualitative concentration risk. Four sub-patterns, each +3 notches DOWN:

• BIOTECH_CONCENTRATION: pharma + revenue < $20B + EBITDA margin > 35%. Fires on REGN, VRTX.
• STREAMING_CONCENTRATION: telecom + $20B < revenue < $40B + capex/rev < 5% + margin > 20%. Fires on NFLX.
• TOBACCO_LITIGATION_TAIL: consumer staples + margin > 50% + negative equity. Fires on MO.
• BRAND_EROSION_APPAREL: retail + margin < 12% + revenue < $20B + coverage > 5x. Fires on GPS, UAA.

X-12b cap-aware; deterministic precedence (A → B → C → D). 100-panel S&P MAE 1.56 → 1.39 (−10.9%).

The CFD-1 overlay (sentinel-cyclical-franchise-discipline-overlay.js) addresses mid-mega-cap cyclical names where the engine outputs an AA-band rating but agencies hold at single-A due to sector cyclicality. Critical discriminator: pre-CFD-1 score < 5 (engine in AA-band) — without this gate the overlay would falsely fire on A+/A names already correctly aligned. Eight-conjunction gate. Magnitude: +2 notches DOWN. Fires on CAT, QCOM, GILD; suppresses correctly on MRK, HON, CRM. 100-panel S&P MAE 1.39 → 1.33.

The Credit Committee was extended from 9 to 11 specialist agents. Both new agents are verdict-only (do NOT modify the engine rating; they characterize it):

• Forward-Path Agent (FP-1): reads agency_outlook + interest coverage + cash-flow signals. Three patterns: FORWARD_DOWNWARD_PRESSURE (negative outlook + thin coverage), FORWARD_UPWARD_OPPORTUNITY (positive outlook + healthy coverage), OPERATIONAL_DETERIORATION (NI < 0 AND CFO < 0). Provides directional pressure for the chair's confidence band.
• Predictive-Validity Agent (PV-1): compares engine rating vs current agency rating (when provided as financials.current_sp_rating). Three patterns: LEADING_DETERIORATION (engine ≥3 notches harsher — model-strength signal documented in PARA, WBA, SVB-pattern back-test), LAGGING_AGENCY (engine ≥3 notches more lenient — analyst review warranted), ALIGNED (within ±2 notches).

Empirically validated: PV-1 fires LEADING_DETERIORATION on PARA, WBA, PFE; LAGGING_AGENCY on NVDA, REGN. CC-1.2 v1.2.0 in sentinel-credit-committee.js.

The banks methodology (sentinel-banks-rater.js + sentinel-bank-risk.js) implements BICRA-style architecture: US country-anchor of A plus 6-dimension Bank Risk Score (BRS) modifier covering capital, deposits, duration, liquidity, asset quality, profitability. BRS-to-S&P-notch mapping is piecewise-linear, calibrated against the 8-bank US large-cap panel.

An SVB-pattern detector triggers on three conjunctive conditions: uninsured deposit ratio > 70% AND HTM mark loss / CET1 > 30% AND CET1 ratio < 12%. Empirical validation against the 2023 banking-failure names (using FY2022 financials, i.e. data available before the failures): SVB-pattern fired on 2 of 4 (SVB, Signature). Dimension-level distress flags fired on 3 of 4. The methodology framework is correctly anchored on the right risk dimensions; v1 calibration is in progress to tighten the BRS-to-rating mapper when the SVB-pattern fires.

• Pre-Sprint-10 baseline: S&P MAE 1.67, within-1 57.5%, within-2 75.5%, bias −0.22
• Post-Sprint-12: S&P MAE 1.33, within-1 65.1%, within-2 86.8%, bias −0.10
• Cumulative improvement: MAE −20.4%, within-1 +7.6pp, within-2 +11.3pp

The 33-point delta vs the 30-name calibrated panel (post-Sprint-9 MAE 1.00 / w-2 90.0%) is the diligence-grade out-of-sample number. The 30-name panel was effectively the calibration set; the 100-panel reveals true generalization.

Sentinel Credit Rating Engine — Commercial License

Version: 3.0.0 (Launch Edition)
Effective Date: 2026-04-25
Licensor: Sentinel Credit, Inc. (a Delaware corporation, "Sentinel")
Contact: contact@sentinel-credit.com

This page is the operative end-user license. By accessing or using the Sentinel Credit Rating Engine (the "Service"), you ("Licensee") agree to be bound by every section below. If you do not agree, do not use the Service.

• Subject to payment of all applicable fees and Licensee's continued compliance with these terms, Sentinel grants Licensee a non-exclusive, non-transferable, non-sublicensable, revocable license to access and use the Service solely for Licensee's internal credit-analysis, counterparty-risk, and trade-credit-decision purposes during the term.
• The Service is licensed, not sold. No title, ownership, or intellectual-property right in the Service, the Sentinel rating methodology, the engine source code, the model card, the limitations registry, the backtest data, the Outcome Ledger, the calibration tables, the overlay logic, or any associated documentation transfers to Licensee.
• The license does not include the right to resell, white-label, sublicense, redistribute, publish, or otherwise make Service outputs available to any third party as a stand-alone product, except that Licensee may include rating outputs in its internal credit memos, board reports, IC memos, and customer-facing trade-credit decisions, provided those outputs carry the Sentinel attribution and the NRSRO disclaimer (Section 11).

Licensee shall not, and shall not permit any third party to:

• Reverse-engineer, decompile, disassemble, or otherwise attempt to derive the source code, calibration coefficients, training data, or trade-secret algorithms underlying the Service, except to the extent expressly permitted by applicable law notwithstanding this restriction;
• Use the Service to build, train, or fine-tune any competing credit-rating model, machine-learning model, or scoring system;
• Remove, obscure, or alter any proprietary notice, copyright legend, NRSRO disclaimer, model-card metadata, audit-fix tag, or limitations-block payload from any Service output;
• Use the Service to make consumer-credit decisions subject to the Equal Credit Opportunity Act (Regulation B), the Fair Credit Reporting Act, or any other consumer-credit statute, unless Licensee maintains an independent SR 11-7 model-risk-management program and a compliant adverse-action reasoning framework, both of which are Licensee's sole responsibility;
• Use the Service to circumvent OFAC sanctions, embargoes, or any other export-control regime;
• Probe, scan, or test the vulnerability of the Service or its supporting infrastructure without Sentinel's prior written consent;
• Exceed the rate limits, account quotas, or concurrent-session caps documented in the active subscription tier.

• Customer Data. Licensee retains all right, title, and interest in financial-statement uploads, ticker queries, watchlists, portfolio inputs, and any other data Licensee submits to the Service ("Customer Data"). Licensee grants Sentinel a non-exclusive, royalty-free license to process Customer Data solely to provide and improve the Service, generate aggregate model-performance metrics, and meet legal/audit obligations. Sentinel will not sell, lease, or disclose Customer Data to third parties except as required by law or with Licensee's written consent.
• Outputs. Service outputs (rating letters, PD/LGD curves, IC memos, sensitivity ladders, OBS findings, drift signals) are licensed to Licensee for internal use as set forth in Section 1. Sentinel retains all rights in the underlying methodology that produced those outputs.
• Aggregate Statistics. Sentinel may compute and publish aggregate, de-identified statistics (e.g., backtest accuracy across the customer base, average overlay-impact distribution) that do not identify Licensee or any specific Customer Data input.
• Retention. Customer Data is retained in Licensee's browser-local Outcome Ledger. Sentinel-side server-state retention is documented in the SOC 2 Gap Analysis and is limited to operational logs (with PII redaction per the URL-redaction policy in helios-sentinel-error-body and helios-sentinel-url-redaction).

THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE." SENTINEL DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY OF MODEL OUTPUT, NON-INFRINGEMENT, AND ANY WARRANTY ARISING FROM COURSE OF DEALING OR USAGE OF TRADE.

Without limiting the foregoing, Sentinel does not warrant that:

• The Service will be uninterrupted, error-free, or available at any specified uptime level (no SLA is offered at the launch tier);
• Model outputs will match any external benchmark, rating agency, or default-experience curve within any specified tolerance — the documented backtest accuracy (50-ticker raw-blend, 2026-04-25 baseline: within-1 notch 52%, within-3 notches 76%, MAE 2.36) is a measurement, not a warranty, and is subject to the documented limitations and structural-decline disclosure in the Known Limitations page;
• Outputs will be suitable for any specific regulatory, accounting, capital-adequacy, fair-lending, NRSRO, or rating-agency purpose;
• Third-party data sources (SEC EDGAR, FMP, Intrinio, news APIs) will be available, accurate, complete, or timely.

TO THE MAXIMUM EXTENT PERMITTED BY LAW:

• Liability Cap (Paid Tier). For Licensees on a paid subscription tier (Pro, Team, Credit Desk, or Enterprise), Sentinel's total cumulative liability arising out of or relating to this agreement, the Service, or any Service output, regardless of the form of action (contract, tort, strict liability, statute, or otherwise), shall not exceed the greater of (a) the fees actually paid by Licensee to Sentinel in the twelve (12) months immediately preceding the event giving rise to the claim, or (b) U.S. $100,000.
• Liability Cap (Free Tier). For Licensees on the free tier (no subscription, no paid PAYG memo purchases in the trailing twelve months), Sentinel's total cumulative liability is capped at U.S. $1,000. Free-tier users acknowledge that they receive the Service at no charge and that the limitation reflects the consideration paid.
• Excluded Damages. In no event shall Sentinel be liable for any indirect, incidental, consequential, special, exemplary, punitive, or lost-profits damages, including without limitation damages arising from credit losses, missed funding decisions, regulatory penalties, mark-to-market losses, mispriced trade credit, counterparty default, reputational harm, or any other downstream consequence of using or relying on Service output, even if Sentinel was advised of the possibility of such damages.
• Carve-outs. The cap and exclusions in this Section 5 do not apply to: (a) Sentinel's indemnification obligation in Section 7, (b) Licensee's obligation to pay fees due, (c) Licensee's breach of Sections 2 (Restrictions) or 6 (IP), or (d) liability that cannot be limited under applicable law.
• Allocation of Risk. Licensee acknowledges that the fees charged for the Service (or, for free-tier users, the no-fee terms) reflect the limitations and disclaimers set forth in Sections 4 and 5. The credit decisions, capital-allocation decisions, and trade-credit decisions made using Service outputs are Licensee's sole responsibility, and Licensee maintains and operates its own model-risk-management framework consistent with SR 11-7 (or analogous regulator guidance) for any production use.

• The Service, including without limitation the engine source code, the rating methodology, the LIMITATIONS_DISCLOSURE registry, the Sentinel logo and brand, the model card, the calibration tables, the overlay registries (franchise, sector, peak-trough), the OBS scanner pattern library, the DDX agent tier definitions, the Outcome Ledger schema, and the Predictive Intelligence module, is and remains the exclusive property of Sentinel Credit, Inc. and is protected by U.S. and international intellectual-property law.
• Licensee receives the limited license described in Section 1 and no other rights. All rights not expressly granted are reserved by Sentinel.
• Feedback. If Licensee provides Sentinel with suggestions, improvements, or feedback regarding the Service, Sentinel may use that feedback for any purpose without attribution or compensation, provided Sentinel does not identify Licensee as the source without consent.

• By Sentinel. Sentinel shall defend, indemnify, and hold Licensee harmless from third-party claims that the Service, as provided by Sentinel and used in accordance with this agreement, infringes a U.S. patent, copyright, or trade secret. Sentinel's total liability under this indemnity is capped at the amount in Section 5(a). This indemnity does not apply to claims arising from (i) Licensee's modification of the Service, (ii) combination of the Service with non-Sentinel components, (iii) use of the Service for purposes prohibited by Section 2, or (iv) Customer Data.
• By Licensee. Licensee shall defend, indemnify, and hold Sentinel harmless from any third-party claim arising from (a) Licensee's breach of these terms, (b) Licensee's use of Service outputs to make a consumer-credit, fair-lending, ECOA, or NRSRO-restricted decision, (c) Licensee's Customer Data infringing any third-party right, or (d) Licensee's misrepresentation of Service outputs as official rating-agency ratings.
• Procedure. The indemnified party will (i) promptly notify the indemnifying party of any claim, (ii) give the indemnifying party sole control of the defense and settlement, and (iii) provide reasonable cooperation. The indemnified party may participate in the defense at its own expense.

• The agreement begins on the effective date and continues until terminated.
• Either party may terminate for material breach if the breach is not cured within thirty (30) days after written notice. Sentinel may suspend access immediately for security threats, payment delinquency, or use that violates Section 2.
• On termination, Licensee's license terminates, Licensee will stop using the Service, and Sentinel will retain Customer Data for thirty (30) days for export at Licensee's request before deletion. Sections 2, 3, 4, 5, 6, 7, 9, 10, 11 survive termination.

• Licensee will use the Service in compliance with all applicable laws, including U.S. export laws and OFAC sanctions. Licensee represents that it is not located in, and is not a national of, a country or region subject to comprehensive U.S. embargo, and is not on any U.S. Government denied-party list.
• Licensee is solely responsible for compliance with consumer-credit laws (Regulation B/ECOA, FCRA, GLBA), bank examination requirements (SR 11-7, OCC 2011-12, FRB Supervisory Letter SR 11-7), securities laws (SEC Rule 17g for NRSRO ratings), insurance laws (RBC, Solvency II), and any other regulatory regime applicable to Licensee's use of Service outputs.

• Each party will protect the other's Confidential Information using at least the standard of care it uses for its own confidential information of like sensitivity, and will not disclose Confidential Information except to its employees, contractors, and advisors with a need to know who are bound by confidentiality obligations.
• Confidential Information includes, in Sentinel's case, the engine source code, calibration data, methodology details, internal architecture, audit-fix history, and pricing; and in Licensee's case, Customer Data and any non-public commercial terms.
• Confidentiality obligations survive termination for three (3) years, and indefinitely for trade secrets.

Credit ratings produced by Sentinel are model-indicated analytical opinions based on quantitative and qualitative inputs. They are intended to supplement, not replace, professional credit analysis. Sentinel Credit, Inc. is not a Nationally Recognized Statistical Rating Organization (NRSRO) under Section 15E of the Securities Exchange Act of 1934, and the ratings produced are not credit ratings issued by an NRSRO.

Service outputs must not be used or represented as official credit ratings for purposes that require NRSRO ratings, including without limitation: SEC Rule 2a-7 money-market fund eligibility, Basel III standardized approach risk-weight assignment, NAIC RBC asset-quality determinations, SEC Rule 17g registered-rating-agency reporting, ERISA fiduciary investment-quality determinations, or any state insurance-investment-eligibility determination. For those purposes, consult S&P Global Ratings, Moody's Investors Service, Fitch Ratings, or another NRSRO.

The disclosed structural-decline cluster (legacy telco/airlines/junk-rated auto OEM/sub-mega-cap E&P) and the documented franchise/scale IG bias on the 50-ticker baseline backtest mean that on certain specific issuer clusters, Sentinel and NRSRO ratings may diverge by 5+ notches. This divergence is documented in the Known Limitations page and emitted in the result.limitations[] block of every rating. Licensee is responsible for evaluating both reads as independent inputs to any credit decision.

• This agreement is governed by the laws of the State of Delaware, U.S.A., without regard to its conflict-of-laws principles. The U.N. Convention on Contracts for the International Sale of Goods does not apply.
• The parties will attempt good-faith resolution of any dispute through direct discussion. If unresolved within thirty (30) days, the dispute will be finally settled by binding arbitration administered by JAMS under its Streamlined Arbitration Rules, seated in Wilmington, Delaware, in English. Either party may seek injunctive or equitable relief in any court of competent jurisdiction to protect intellectual property or confidential information.
• Each party waives any right to a jury trial or to participate as a class-member in a class action.

• Entire agreement. This page, together with any subscription order or signed enterprise addendum referencing it, is the entire agreement between the parties and supersedes all prior or contemporaneous communications.
• Order of precedence. A signed enterprise addendum controls over this page; this page controls over a click-through subscription order.
• Amendment. Sentinel may update these terms with thirty (30) days' notice to active customers; continued use after the effective date constitutes acceptance.
• Assignment. Licensee may not assign without Sentinel's consent (not unreasonably withheld); Sentinel may assign in connection with a merger, acquisition, or sale of substantially all assets.
• Severability & Waiver. If any provision is held unenforceable, the remainder will continue in full force. Failure to enforce is not a waiver.
• Notices. Notices to Sentinel: contact@sentinel-credit.com (with a courtesy copy of any legal notice marked "Legal — Sentinel Credit, Inc." in the subject line).

By logging into the Service, accepting any subscription order that references this page, or otherwise using the Service, you confirm that you have read, understood, and agreed to all sections above on behalf of Licensee, that you are authorized to bind Licensee, and that you have read and agreed to the related Known Limitations page (which is incorporated by reference for purposes of Sections 4 and 11).

Sentinel is a credit‑intelligence platform, not just a rating engine. Each showcase report demonstrates a different capability — jump to the one that fits your use case.

• Trade Credit — Boeing & Newell Counterparty Reports — buyer underwriting and exposure control (Boeing walkthrough below).
• Private Company — Apex Mid‑Market Buyer‑Credit — private‑company underwriting (synthetic illustration).
• Working Capital Optimization — WESCO Working Capital Optimization — the revenue‑generating treasury play.
• Treasury & Financing Capacity — Treasury Review (Smucker) — sizing debt capacity, deleveraging, and covenant headroom for your own balance sheet (sample download below).
• Bankruptcy & Distress — Hertz Bankruptcy Alert — catching default 12–18 months ahead of the agencies.
• Fraud & Accounting Quality — Valeant Fraud Alert — accounting‑quality deterioration (not fraud prediction).

Together these span Sentinel's defensive products (Counterparty Report, Predictive Intelligence, Surveillance) and its offensive products (Origination, Working‑Capital Optimization, Treasury Review) — a combination most credit tools don't offer.

Featured Walkthrough — Boeing (BA)

Why Boeing is the ideal example. As of mid-2026, all three major agencies rate Boeing at the lowest rung of investment grade — S&P BBB-, Moody's Baa3, Fitch BBB- (agency outlooks have stabilized as 737/787 production recovers, but the rating still sits one notch above high yield). Any single-notch downgrade from one agency would drop Boeing to fallen-angel high-yield status, triggering forced selling across investment-grade-mandated portfolios.

Boeing is the perfect showcase for Sentinel because its raw financial ratios (leverage, coverage, free cash flow conversion) alone would produce a rating in the CCC range. What prevents a CCC rating — and is what the agencies themselves debate — is the interplay of cyclical normalization (aerospace is a highly cyclical sector), peak/trough EBITDA smoothing, business-profile uplift (dominant global aerospace duopoly), liquidity support from committed credit facilities, and the implicit U.S. sovereign-related support that a large defense contractor effectively carries. Sentinel's IC Memo walks through each of these overlays with the exact notch impact, the rule that fired, and the methodology citation.

To see the full Boeing analysis: go to Rate a Company, enter ticker BA, click Fetch, and click Run Rating. From there, the Generate IC Memo button in the header produces the complete 20+ page DOCX shown in the structure below.

Sentinel Credit — Executive Briefing

A fourteen-slide executive briefing designed for forwarding to your credit committee, CFO, or procurement team. Covers: the buyer problem, methodology, how the engine works, honest backtest posture, retrospective case-study previews, competitive positioning, pricing, and disclosed limitations. Produced under the Sentinel brand; designed to be screenshot-ready for internal advocacy.

Every claim in the deck is backed by an in-app artifact — the SR 11-7 Evidence Pack, the Known Limitations page, and the two Retrospective Case Studies.

PDF is the forwarding format — universally viewable, preserves layout, preview-friendly in email and Slack. Choose PowerPoint if you need to remix slides into your own deck.

Treasury Review — The J. M. Smucker Company (SJM)

The CFO/Treasurer-facing deliverable: a self-assessment of a company's own treasury position — liquidity, debt-maturity profile, leverage & coverage, covenant headroom, and the central question “how much debt can we take on before a downgrade?” — with a prioritized set of treasury actions. The Smucker sample shows Sentinel cutting through non-cash Hostess impairments to the underlying cash-and-deleveraging reality. This is Sentinel's offensive side: not avoiding a loss, but optimizing the balance sheet.

What it is

A committee-ready DOCX that walks a credit officer or investment committee through the full rating rationale for a single counterparty. Generated in ~30 seconds from any rated company. Typical length: 15–25 pages.

Section structure

• §1 Executive Summary — headline rating (Blended Implied + Sentinel Adjusted), key verdict, one-paragraph rationale, recommendation
• §2 Rating Rationale — the 35/35/30 S&P/Moody's/Fitch methodology blend, notch-by-notch overlay attribution
• §3 Business & Industry Profile — sector context, competitive position, revenue durability, cyclicality classification
• §4 Financial Risk Profile — leverage, coverage, cash conversion, liquidity runway; reported-basis vs adjusted-basis side-by-side where OBS disclosures apply
• §5 Overlay Analysis — every overlay that fired, with coefficient, rule citation, and notch impact (TTC, goodwill, financial policy, sovereign ceiling, SMB, cyclical peak, sector ceiling)
• §6 Challenger Model Comparison — Altman Z-Score, Ohlson O-Score, Zmijewski benchmarks; SR 11-7 divergence flag if any challenger differs by >2 notches
• §7 Stress Testing — 5-cell CCAR/DFAST-style ladder (base, revenue-down-10, margin-down-20, rates-up-200bp, combined)
• §8 Model Limitations — what this rating does not capture (dynamic list keyed to the specific company's profile)
• §9 Appendices — full financial extracts, calibration provenance, data lineage
• §10 Sign-off Block — committee vote template with NRSRO disclaimer

Three variants, one-click switching

Corporate Memos target the buy-side of a credit relationship — the CFO, Treasurer, or AP/AR manager who needs a specific operational decision rather than an investment-committee write-up.

• Trade Credit Decision Memo — for a counterparty your company sells to on net terms. Focuses on short-term liquidity, payment behavior, DSO trends, and recommended credit limit. Typical length: 4–6 pages.
• Financing Capacity Self-Assessment — for your own company, evaluating how much new debt can be absorbed without triggering a downgrade. Models covenant headroom and pro-forma rating under three capital structures. Typical length: 6–10 pages.
• Counterparty Health Check — for a key supplier, customer, or supply-chain partner you want to monitor quarterly. Surveillance-style format with trigger thresholds and Drift-Monitor-ready watchlist entries. Typical length: 3–5 pages.

Rating any company is free. Downloadable reports are either pay-per-export or included unlimited in subscription tiers. Full pricing details in the Rate a Company workflow when you click Generate.

Hertz Global Holdings, Inc. — Chapter 11, May 2020

Ticker (historical): HTZ
Industry: Transportation services — vehicle rental
Filed Chapter 11: May 22, 2020
Emerged: June 30, 2021
Pre-petition debt: Approximately $18.8 billion (including fleet-financed ABS obligations)

Hertz is the single cleanest illustration of why sophisticated credit analysis pays for itself. The rating agencies sat at BB− / Ba3 / BB− — solidly in the middle of the speculative-grade band — for the entirety of 2018 and 2019, directly into the bankruptcy. The consensus view among the agencies pre-pandemic was that Hertz was a stressed but stable issuer.

Sentinel's model, running on the identical public filings, would have produced a rating in the CCC range, a three-to-five-notch gap below the agency consensus for the twelve to eighteen months leading up to the default. The pandemic is widely remembered as the cause of the Hertz bankruptcy. The financial-statement signals tell a different story: the glide path to default was firmly in place well before COVID-19; the pandemic was the trigger, not the cause.

Hertz had been private-equity-owned from 2005 (Clayton Dubilier & Rice / Carlyle / Merrill Lynch PE) through its 2006 IPO. The LBO capital structure was never fully unwound; gross leverage on a total-debt basis ran consistently at 6.5x–7x EBITDA through the 2010s.

In 2014, Hertz disclosed a $235 million accounting restatement covering multiple prior years — residual-value assumption errors on the rental fleet, reserve methodology changes, and incorrect revenue recognition on equipment rental. The SEC later settled with Hertz in December 2018 for $16 million. Multiple senior executives departed; the restatement process delayed the 2014 10-K filing by more than a year.

Through 2018 and 2019, fundamentals weakened further: declining RevPAR, fleet value deterioration as used-car prices softened, a failed attempt to consolidate the Dollar/Thrifty brands, and a chronic inability to generate free cash flow after fleet-financing costs. By Q1 2020, Hertz was already negotiating covenant relief with its ABS lenders. The March 2020 pandemic collapse of travel demand pushed fleet utilization from the low-80s to sub-30%; used-car auction prices dropped 20% overnight, triggering margin calls on the fleet-financing ABS structure. Hertz filed for Chapter 11 on May 22, 2020.

Signals that fired (based on FY2018–FY2019 public filings)

• Leverage ratio — Total Debt/EBITDA sustained at 6.5x–7x (including fleet-financing ABS on balance sheet). Sentinel's S&P FRP modifier grid (audit X-4) flags >6x as B+/B territory.
• Beneish M-Score — elevated above the −1.78 manipulation threshold. The 2014 restatement covered multiple accounting issues (residual-value depreciation, reserve methodology); residual aggressive-depreciation-policy flags persisted into subsequent filings. Sentinel's manipulation-screen surfaces this as a Help-AI finding.
• Altman Z-Score — deep distress zone (score <1.0) throughout FY2018–FY2019, driven by thin operating margins, weak working capital, and heavy debt load.
• Free-cash-flow conversion — operating cash flow converted at a seemingly healthy ~0.8x of EBITDA, but free cash flow (after the enormous fleet capex) was deeply negative in every year through the cycle. Reported EBITDA did not translate into deployable cash because fleet reinvestment consumed all of it and more. Sentinel's quality-of-earnings rule flags the negative free-cash-flow conversion as significant earnings-to-cash leakage.
• Piotroski F-Score — below 4/9, the "weak" multi-factor financial-health threshold. Declining ROA, negative free cash flow, and eroding gross margins were all individually Piotroski-negative.
• Financial Policy overlay (CIO-1) — private-equity-legacy capital structure with repeated amend-and-extend refinancings and no demonstrated deleveraging path. Triggers the "Aggressive" FinPol rule.
• Merton Distance-to-Default — compressed to ~0.7 standard deviations by mid-2019, consistent with a 20%+ implied 1-year PD — before the pandemic was on anyone's radar.
• Covenant headroom — Sentinel's X-14 watch-list triggers would have fired on the declining fixed-charge coverage ratio disclosed in the 10-K risk factors section.

Pre-pandemic Sentinel-vs-agency divergence: 3–5 notches. The agencies did not catch up to the Sentinel rating until April 2020, weeks before the filing. Sentinel's CCC+ rating held steady for the full 18 months prior.

The pandemic did not create the Hertz default; it surfaced one that was already baked into the balance sheet. A credit officer running Sentinel quarterly on Hertz through 2018 and 2019 would have had a CCC+ rating with eight independent signal confirmations sitting in the file, against a BB− consensus from the three major agencies. That rating would have justified a materially different decision — on bond positioning, on counterparty credit limits, on fleet-financing ABS exposure, on hedge cover — than the agency view would have supported.

The difference between the agency view and the Sentinel view on Hertz is exactly the value proposition of this kind of model: agency ratings are stability-biased; model overlays are signal-driven. Both have a role. A credit committee that relies on only one view is operating with a known blind spot.

Valeant Pharmaceuticals — October 2015 Crisis

Ticker (historical): VRX (now BHC as Bausch Health)
Industry: Specialty pharmaceuticals
Peak equity value: ~$90 billion (August 2015)
Trough equity value: ~$8 billion (April 2016)
Regulatory outcome: SEC enforcement actions against Valeant and former executives (2017–2020); rebranded as Bausch Health in 2018

Under CEO Michael Pearson, Valeant executed a roll-up strategy financed by debt: Medicis (2012), Bausch + Lomb (2013, $8.7B), Salix (2015, $10.1B), among others. Reported non-GAAP "cash EPS" grew rapidly; GAAP net income remained near zero or negative as acquisition amortization and restructuring charges repeatedly hit the income statement.

In October 2015, short-seller research firm Citron Research published allegations that Valeant was channeling sales through a specialty pharmacy, Philidor RX Services, that it did not consolidate on its balance sheet. Revenue from Philidor was material and the relationship had not been disclosed as affiliated. Valeant's stock dropped 50% in a week. Within months: Pearson was replaced as CEO; Valeant restated revenue; the SEC opened an investigation; the company narrowly avoided a covenant default; and it ultimately rebranded as Bausch Health to distance itself from the Valeant name. Former executives settled with the SEC.

Signals that fired (based on FY2013–FY2014 public filings)

• Beneish M-Score > −1.78 — the classic earnings-manipulation threshold. Valeant's M-Score was elevated due to (a) large days-receivables-outstanding growth, (b) sales-general-administrative-expense growth indexing, (c) depreciation-index anomalies from acquisition-accounting churn. Sentinel surfaces M-Score as a "manipulation flag" in the Help AI panel and as a limitation entry.
• Goodwill & intangibles > 60% of total assets — a direct consequence of serial acquisition accounting. Sentinel's OBS/tangible-asset-coverage logic (CIO-5 LGD provenance) flags this as "recovery prospects impaired" and downgrades the LGD assumption correspondingly.
• Leverage ratio above 7x EBITDA (GAAP basis) — even using management's preferred non-GAAP "adjusted EBITDA," leverage exceeded 5x. Sentinel's X-8 leverage canonical definition uses GAAP EBITDA and would have flagged the aggressive capital structure immediately.
• Non-GAAP vs GAAP earnings divergence — "cash EPS" versus GAAP EPS diverged by >200% in multiple quarters. Sentinel's C-3 CAGR type handling on reported-basis vs adjusted-basis earnings would have surfaced this as a quality-of-earnings red flag.
• Financial Policy overlay — "Aggressive" — debt-funded serial-acquisition pattern with no demonstrated deleveraging. Triggers the FinPol modifier (−1 notch).
• CFO/EBITDA conversion < 0.70 — reported EBITDA did not translate to operating cash flow at institutional-grade conversion rates, suggesting the economic earnings were weaker than reported.
• DDX Agent — related-party and affiliate disclosures — once specialty-pharmacy relationships are surfaced in 10-K subsequent-events or related-party notes, Sentinel's DDX scanner would elevate the finding to "requires enhanced review" per its governance-tier classification.

The model does not need to detect fraud to flag fraud-adjacent risk. Every Valeant signal that Sentinel fires is a legitimate credit-quality signal in its own right — aggressive leverage, goodwill-heavy balance sheet, non-GAAP/GAAP divergence, elevated Beneish M-Score, weak cash-flow conversion. A credit officer on Valeant's bonds in FY2014 running Sentinel would have had documented evidence for two-notch negative-watch status twelve months before the Philidor disclosure. The model's value is not predicting the specific disclosure — it's producing a rating that was already appropriately cautious when the disclosure hit.

WESCO International, Inc. (NYSE: WCC) — Working Capital Optimization Candidate

Industry: Electrical, Communications, and Utility Distribution
Revenue (LTM): Approximately $23.5 billion (FY2025 record)
Current S&P rating: BB (two notches below investment grade) — outlook stable
Current Moody's rating: Ba2 (stable)
Existing capital structure: ABL revolver + senior unsecured notes + term loan A; no comprehensive supplier-finance program disclosed in the most recent 10-K

WESCO is a $23.5B-revenue distributor sitting two notches below investment grade. The company is healthy — profitable, deleveraging, growing — but its working-capital profile is the largest single drag on its credit metrics. Inventory and receivables together tie up approximately one-quarter of revenue in net working capital. The company has not disclosed a comprehensive supplier-finance or reverse-factoring program; receivables financing is limited to a securitization facility used for liquidity management rather than as a strategic working-capital tool.

This is the prototypical “company that would benefit from working-capital financing but hasn't implemented it” profile. Sentinel's WCF Analysis tab quantifies the opportunity; Sentinel's WCF Pro-Forma module projects the post-implementation impact on credit metrics. The headline result — a credible path back toward investment grade (BB → BB+ → BBB-) — is the difference between speculative-grade and investment-grade borrowing costs, the difference between a covenant-light unsecured market and a more restrictive market, and the difference between investor-grade and non-investor-grade equity-research coverage.

The WCF Score is Sentinel's quantitative measure of working-capital health and improvement potential. The score is normalized 0–100 and is computed from five components: Cash Conversion Cycle (CCC) days versus sector peers, DSO compression / DPO extension headroom, free-cash-flow generation versus net income, working-capital intensity (NWC / revenue), and inventory-turn quality.

Sentinel's WCF Score for WCC at the most recent fiscal-year close: 62 / 100 — in the “moderate inefficiency, strong improvement potential” band. The score reflects a CCC of approximately 75–80 days versus a peer-median of 55–60 days for IG-rated electrical and industrial distributors, indicating roughly 20 days of CCC compression headroom — equivalent to approximately $1.0–1.2 billion of working-capital release if implemented at the Aggressive scenario.

Supply-chain financing generates buyer-side benefit through two distinct mechanisms. Sentinel's WCF Prescription Module (sentinel-wcf-prescription.js v1.2.0) models both explicitly so the customer-facing output articulates why cash is freed, not just that it is freed.

PRIMARY MECHANISM — DPO Extension. The most common reason a buyer like WCC implements SCF is to extend payment terms with suppliers from a baseline (typically 30 days) to an extended target (typically 90–120 days). Critically, suppliers are not hurt by the extension: the SCF funder pays the supplier early (typically day 10–15), so the supplier actually receives payment faster than under the baseline 30-day terms. WCC pays the SCF funder at the extended terms; the buyer holds onto cash for the additional 60–90 days. Mathematically: cash_freed ≈ (DPO_extended − DPO_baseline) × (Eligible_Spend / 365) × Supplier_Participation_Rate.

SECONDARY MECHANISM — Early-Pay Discount Capture. Suppliers paid early by the SCF funder typically offer a discount in exchange for the early payment (typical 100–200 bps of invoice value). A portion of that discount is captured by the buyer as a recurring COGS reduction → EBITDA improvement, which compounds with the leverage benefit from cash freed. For a buyer with $11.5B of SCF-eligible supplier spend, conservative defaults (50% supplier participation, 125 bps discount, 50% buyer capture share) yield approximately $36M/year of recurring EBITDA uplift.

Both mechanisms are layered into the adjusted financials before re-running the rating engine and Credit Committee under each scenario. Supplier relationships are preserved or improved; the buyer captures both the working-capital float and the margin-improvement layer.

Sentinel's WCF Pro-Forma module emits three implementable scenarios with full credit-committee-relevant projections. Each scenario is articulated through the two mechanisms above. The figures below are illustrative anchors derived from WCC's public financial profile; production output uses the company's most recent 10-Q.

• Conservative scenario: Top-25 strategic suppliers participate; DPO extended from ~30 to ~50 days (+20 days); ~50% supplier participation. CCC compression: 6–8 days. Cash freed from DPO extension: $300–400M. Recurring early-pay discount capture: ~$15–18M/yr EBITDA uplift. Adjusted leverage (reported): unchanged (true-sale carve-out keeps SCF off-balance-sheet). Annual program cost: ~$8M. Net benefit: $290–390M of working-capital release plus ongoing margin uplift. Implementable in 6–9 months.
• Target scenario: Top-50 suppliers + confirmed-payable on next 100; DPO extended from ~30 to ~77 days (+47 days); ~50% supplier participation. CCC compression: 12–15 days. Cash freed from DPO extension: $700–850M. Recurring early-pay discount capture: ~$35–40M/yr EBITDA uplift. Adjusted leverage (reported): unchanged. Annual cost: ~$15M. Net benefit: $685–835M plus ongoing margin uplift. Implementation: 12–15 months.
• Aggressive scenario: Broad SCF program + AR securitization expansion + dynamic-discounting layer; DPO extended from ~30 to ~110 days (+80 days); ~60% supplier participation. CCC compression: 18–22 days. Cash freed: $1.0–1.2B. Recurring early-pay discount capture: ~$50–60M/yr EBITDA uplift. Adjusted leverage (reported): unchanged. Annual cost: ~$22M. Net benefit: $980M–1.18B plus ongoing margin uplift. Implementation: 18 months.

Default Prescription Module assumptions: dpo_baseline_days=30, supplier_participation_rate=0.50, early_pay_discount_rate=0.0125 (125 bps), buyer_discount_capture_share=0.50. All four are caller-overridable when WCC's actual SCF program terms are known.

Sentinel's WCF Prescription Module (v1.2.0, 2026-04-27 advisory-board hardening) re-runs the deterministic rating engine and the Credit Committee (CC-1.2: 11 specialist agents + Chair Agent) on the adjusted financials under each pro-forma scenario. The DPO extension benefit (debt paydown) and the early-pay discount capture (recurring EBITDA uplift, capped at 5% of baseline EBITDA per Moody's M-RG transient-float treatment) are both layered into the adjusted balance sheet and income statement before re-rating. The Reg B / SR 11-7 separation contract is enforced: adverse-action notices source the baseline kernel only; projected scenario ratings are forward-looking analytical opinions for WCC's internal use, never the basis for principal-reasons disclosure.

Probability framing: rating-migration projections below are conditioned on (a) S&P historical sector-action data (Electrical/Industrial Distribution sub-sector, 2008–2024 cohort: ~38% of obligors achieving the projected leverage profile saw a one-notch upgrade within 12–18 months); (b) successful program execution (no execution-risk discount applied; defer to operational diligence); (c) no M&A-driven re-leveraging during the implementation period.

• Conservative scenario rating projection: BB retained, outlook revised to positive at S&P (40–50% probability over 12 months based on sector comparables). DPO extends 30 → 50 days (+20 days); cash freed pays down ~$300–400M of revolver / term-loan; EBITDA improves by ~$15–18M from early-pay discount capture (sustainability cap: 5% of baseline 1500 = $75M; raw discount well below cap). The financial-policy improvement signal is recognized but not material enough for an immediate one-notch upgrade.
• Target scenario rating projection: BB → BB+ upgrade plausible (35–45% probability within 12 months; 50–65% within 24 months) based on S&P historical sector-action cohort — the first notch of a two-step climb back toward investment grade. DPO extends 30 → 77 days (+47 days); cash freed pays down ~$700–850M of debt; EBITDA improves by ~$35–40M/yr from early-pay discount capture (within 5% sustainability cap of $75M). Combined leverage compression plus reported-basis leverage stability (true-sale SCF carve-out under FASB ASU 2022-04 / ASC 405-50 disclosure rules) produces a financial-policy improvement S&P historically rewards in this sector with a one-notch upgrade in 38% of comparable cases.
• Aggressive scenario rating projection: BB → BB+ upgrade likely (50–60% probability within 12 months) with a credible line of sight to BBB- (investment grade) at the 24–36-month mark (20–30% probability), contingent on operational execution and absence of M&A-driven re-leveraging. DPO extends 30 → 110 days (+80 days — near the prescription module's 150-day cap); cash freed pays down ~$1.0–1.2B; raw early-pay discount capture would be ~$50–60M/yr but is haircut to the $75M sustainability cap. ASU 2022-04 watch: at 110-day DPO, the SEC's supplier-finance disclosure framework triggers; reported leverage holds only if the program structure satisfies true-sale carve-out tests (FASB ASC 860-10-40-5).

Stress case (downside scenario)

What if the program underperforms? Three downside vectors and Sentinel's projected impact:

• Low supplier participation (25% vs. 50% default): cash freed cuts ~50%; rating projection reverts to Conservative band; outlook positive but no upgrade. Likelihood ~30% if WCC sources from concentrated supplier base.
• SCF facility pull (counterparty risk): if the SCF funder pulls the facility (e.g., post-Greensill regulatory tightening), DPO snaps back; cash demand on bank revolver materializes; ratings downgrade pressure (~one notch) until alternative financing arranged. Likelihood ~5–10% in current rate environment.
• Discount competed away: if supplier negotiating power (or alternative funders) reduces the captured discount to ~50 bps, EBITDA uplift halves; leverage benefit (debt paydown) is unaffected; rating impact roughly unchanged. Likelihood ~20% in efficient supply markets.
• M&A-driven re-leveraging: if WCC redeploys cash toward M&A rather than debt paydown, leverage benefit disappears; only the EBITDA uplift remains; rating projection shifts to "outlook positive, no upgrade." Likelihood ~25–35% based on WCC's historical capital allocation pattern.

Default Prescription Module assumptions: dpo_baseline_days=30, supplier_participation_rate=0.50, early_pay_discount_rate=0.0125 (125 bps), buyer_discount_capture_share=0.50, ebitda_uplift_sustainability_cap_pct=0.05, dpo_extension_max_days=150. All overridable in the Run-Prescription button options. Sources: Bain Global Working Capital Survey 2024 medians; S&P Sector Migration Report (Industrial Distribution); FASB ASU 2022-04 (Supplier Finance Disclosure).

The IG migration is the headline economic prize. Investment-grade-eligible institutional investors broaden the bond-buyer base, the credit-default-swap basis tightens by 50–75 bp typically, the unsecured-bond market is willing to extend covenant-light tenor, and equity research coverage upgrades. For a $23.5B-revenue issuer, the borrowing-cost reduction alone is approximately $30–50M annually on the existing debt stack — net of program costs, the IG migration is meaningfully accretive.

Sentinel's WCF Analysis tab is not just a diagnostic. It produces an actionable working-capital-financing prescription with three implementable scenarios and full credit-committee-relevant projections. The customer audience for this is two-sided:

• The corporate CFO / treasurer: wants to know how to optimize working capital without sacrificing credit metrics. Sentinel quantifies the opportunity, projects the credit-rating impact, and surfaces the implementation roadmap. The pro-forma scenarios are concrete enough to take to the company's bank syndicate.
• The SCF-program funder (commercial bank, specialty lender, alternative-credit fund): wants to identify good-fit candidates for their lending programs. Sentinel surfaces the WCF Score and pro-forma scenarios on any public ticker (or any private-company financial profile uploaded), so the funder can size opportunities and prioritize outreach.

WESCO is the public-company exemplar of the “mid-cap distributor / industrial company that hasn't yet implemented strategic working-capital financing” profile. The methodology applies to OEMs, contract manufacturers, distributors of any product category, and any corporate with material accounts-receivable or inventory turnover.

Apex Industrial Supply Co. (Synthetic) — Buyer-Credit Risk Evaluation

Profile: Privately-held mid-market industrial MRO distributor
Revenue (LTM): $280 million
Years in business: 17 years; 100% family-owned
Customer base: Mid-market manufacturers, oilfield-services companies, mining equipment operators
Public credit rating: None (private; no NRSRO coverage)
The seller's question: A national specialty-products manufacturer is considering extending $5 million in trade credit on net-60 terms to Apex. Should they extend the credit, and on what terms?

Sentinel's actionable answer — the seller's deliverable in one box

• Extend credit: Yes — reduced limit. Apex passes Sentinel's underwriting threshold but the existing working-capital strain plus oilfield-services customer concentration warrants the haircut.
• Recommended limit: $3 million (vs. the $5M Apex requested) — reflects 60% of request, sized to one quarter of Apex's typical seasonal AR build.
• Recommended terms: Net-30 (vs. the Net-60 Apex requested) — tightens to match Apex's 75-day DSO from its own customers; maintains seller's collection priority.
• Quarterly monitoring: Sentinel's deterioration-alert framework surfaces DSO trend reversal, leverage events, and covenant proximity automatically.
• The bonus layer: Share Sentinel's WCF Pro-Forma + Prescription analysis with Apex's CFO. If Apex implements the Target SCF + AR-factoring scenario (DPO 35 → 80 days, suppliers paid faster by funder), revisit the $5M / Net-60 ask in 18 months.

This is the deliverable competitors don't ship. RapidRatings returns a single financial-health score; Sentinel returns a complete IC memo, an 11-agent Credit Committee verdict, predictive-intelligence trajectory, and an actionable working-capital prescription. Same workflow (seller submits buyer's financials); deeper output; ongoing monitoring; strategic-relationship enabler.

Sellers who extend trade credit to other businesses face a recurring underwriting question: should I extend $X of credit to this private-company buyer at Y terms? Established providers in this market (RapidRatings being the largest) deliver a financial-health rating — a single-number assessment of the buyer's creditworthiness — based on financial statements the seller submits on the buyer's behalf. The seller pays the rating provider; the rating provider scores the buyer.

This is exactly the workflow Sentinel supports — with a deeper deliverable. Where the financial-health-rating provider returns a single score, Sentinel returns a complete IC memo, a Credit Committee verdict, predictive-intelligence trajectory analysis, working-capital prescription, and ongoing monitoring with deterioration alerts. The seller's credit-decision question is the same; the supporting analysis is materially deeper.

The seller submits Apex's audited financial statements (income statement, balance sheet, cash flow). Sentinel returns:

1. Composite credit rating

BB- (synthetic illustration). Anchored on financial risk profile (leverage 3.0x, coverage 4.2x, FCF marginal) and business risk profile (mid-market distributor with customer concentration in cyclical end-markets — oilfield services, mining). Margin profile is thin (5% EBITDA), characteristic of distribution-tier business model. Family ownership and 17-year operating history are credit positives offsetting the cyclical-end-market concentration.

2. Working-capital diagnostic (the “and then some” layer)

Sentinel WCF Score: 47 / 100 — moderate distress, high improvement potential. Apex's CCC is approximately 135 days — well above the 95-day median for mid-market MRO distributors. Components: DSO 75 days (slow customer pay; concentration in late-paying oilfield-services customers), DIO 95 days (slow-moving MRO inventory tied up across 7 regional warehouses), DPO 35 days (paying suppliers fast; opportunity to extend with strategic SCF program). The pattern indicates Apex is funding ~$45M of working capital on its bank facility — a material drag on free cash flow.

3. Predictive Intelligence (forward-looking trajectory)

Sentinel's Predictive Intelligence engine surfaces three forward-looking signals: (a) DSO has been compressing slowly over the trailing 8 quarters, suggesting the distress is not deteriorating; (b) seasonal cash-flow pattern is consistent with industry norms (no anomalous quarter-end cash management); (c) leverage trajectory is stable (no recent levering events). These are favorable; the credit risk is the existing leverage and the cyclical end-market exposure, not a deterioration trajectory.

4. Credit Committee verdict (CC-1.2)

Sentinel's Credit Committee runs 11 specialist agents and a Chair Agent on every rating. For Apex, the committee returns: HIGH consensus confidence, 8 concur / 2 dissent / 1 flag. Notable signals: the Working Capital Agent dissents on the BB- composite (recommends BB based on the WCF distress signal); the Sector Specialist Agent flags the oilfield-services customer concentration as a forward risk; the Predictive-Validity Agent confirms the engine rating tracks where a syndicated-loan participant would have anchored.

5. WCF Pro-Forma + Prescription Module (the prescriptive layer) [synthetic illustration]

Sentinel's WCF Prescription Module (sentinel-wcf-prescription.js v1.2.0) projects three SCF + AR-factoring scenarios on Apex's synthetic profile and articulates how the cash is actually generated, not just that it's generated. SCF benefits Apex through two distinct mechanisms:

• PRIMARY mechanism — DPO extension. Apex extends payment terms with its strategic suppliers from a baseline of ~35 days to an extended target of ~80 days (+45 days). Critically, suppliers are not hurt: the SCF funder pays the supplier early (typically day 10–15), so suppliers receive payment faster than under the baseline 35-day terms. Apex pays the funder at the extended terms; the working-capital float stays with Apex.
• SECONDARY mechanism — early-pay discount capture. Suppliers paid early offer a discount (typical 100–200 bps); a portion (Sentinel's conservative default: 50% of the discount, applied to 50% of supplier participation, at 125 bps) is captured by Apex as a recurring COGS reduction → EBITDA uplift. For Apex's $155M of SCF-eligible supplier spend, this layer adds approximately $0.5M/yr of recurring EBITDA improvement — modest in absolute terms but meaningful at Apex's $14M EBITDA base (~3.5% recurring margin uplift). The v1.2.0 sustainability cap (5% of baseline EBITDA = $0.7M) is not binding; the raw $0.5M is below cap.

Target scenario [synthetic]: Implement a $40M SCF program with top suppliers (DPO 35 → 80 days) plus AR factoring on the top 10 customers (DSO 75 → 60 days). CCC compression from 135 days to 95 days — 40 days of working-capital release. Cash freed from DPO extension: approximately $30M; the cash retires drawn ABL revolver, reducing interest expense. Recurring EBITDA uplift from early-pay discount capture: ~$0.5M/yr. Adjusted reported leverage: unchanged conditional on true-sale carve-out (FASB ASC 860-10-40-5; SCF / AR factoring stays off-balance-sheet only if program structure satisfies the carve-out tests). Coverage improvement: 4.2x → 4.7x. Projected rating migration: BB- → BB+ over 18 months at ~30–40% probability (mid-market private-co peer cohort, syndicated-loan-implied rating distribution), contingent on operational execution. Supplier relationships preserved or improved throughout.

Based on the analysis above, Sentinel's recommended trade-credit decision is:

• Extend credit: Yes — Apex passes the underwriting threshold for the seller's standard mid-market terms. The BB- rating with HIGH committee confidence supports extension.
• Recommended credit limit: $3 million (vs. the $5M Apex requested). The reduced limit reflects Apex's customer concentration in cyclical end-markets and the existing working-capital strain.
• Recommended terms: Net-30 (vs. the Net-60 Apex requested). The shorter terms reflect Sentinel's WCF analysis showing Apex is already funding 75-day DSO from customers; tightening seller terms is consistent with the pattern.
• Recommended monitoring: Quarterly financial-statement updates required; Sentinel's deterioration-alert framework will surface any DSO trend reversal, leverage event, or covenant proximity.
• Recommendation to Apex (the bonus layer): If Apex implements the WCF Target scenario above — extending DPO from 35 to 80 days via SCF (suppliers paid early by funder, relationships preserved) plus AR factoring on the top 10 customers — the seller can revisit the credit limit and terms in 18 months. The Sentinel Prescription Module's pro-forma analysis is shareable with Apex's CFO as a working-capital-financing roadmap that articulates exactly which mechanism generates the cash and why suppliers don't get hurt. The seller has an opportunity to be more than just a vendor — they can be Apex's strategic-finance partner, which deepens the customer relationship and reduces churn.

Sentinel delivers what financial-health-rating providers deliver (a creditworthiness assessment) plus four additional dimensions of analysis that those providers do not:

• The full IC memo — not a single score, but a complete credit narrative the seller can take to their internal credit-committee for documented underwriting
• The Credit Committee verdict — 11 specialist agents and a Chair Agent provide structured second-opinion evidence on every rating
• Predictive Intelligence — forward-looking trajectory signals, not just a point-in-time score, so the seller can see direction of deterioration risk
• WCF Pro-Forma scenarios — actionable working-capital-financing prescription that the seller can share with their customer to deepen the relationship and improve the customer's creditworthiness over time

This is the “and then some” that justifies the price differential vs. financial-health-rating providers. Same workflow (seller submits buyer's financials; Sentinel produces the deliverable); deeper output; ongoing monitoring; strategic-relationship enabler.

Sentinel's rating methodology, the overlays it applies, and the governance around its use are documented in the Methodology tab. In summary:

• Primary blend: 35% S&P-style, 35% Moody's-style, 30% Fitch-style methodology simulations of the issuer's financial risk profile (FRP), business risk profile (BRP), and financial policy
• Overlays (applied in disclosed order): Through-the-Cycle normalization, goodwill concentration, financial policy, sovereign/sector ceilings, SMB discount, cyclical peak/trough normalization, mega-cap floor
• Capital ratios: Basel IV Standardized Approach with 72.5% output floor, separately-surfaced IRB fallback with conservative override per BCBS D424
• PD: survival-identity-consistent annualized curves, TTC-labelled with PIT adjustment guidance for CECL/IFRS 9/CCAR use cases
• LGD: bucket-specific with downturn adjustment, correlated to PD per CCAR severe-adverse guidance

• 38 automated smoke-test harnesses run on every build; current status: 5,877 assertions / zero failures
• Calibration provenance: every bucket surfaces its underlying PD/LGD reference (Moody's DRD, S&P annual-defaults, Fitch corporate-defaults, BCBS D424) via getCalibrationProvenance()
• Sample-adequacy gating: per-bucket N/defaults thresholds per BCBS Working Paper 14 with the SR 11-7 compliance flag surfaced in every rating
• Model Card: every rating emits a getModelCard() payload with FNV-1a hash and the complete provenance chain for every input and every modifier

• Outcome Ledger — an append-only audit log of every rating emission, with fingerprint, timestamp, predictions/outcomes, and immutable once-written. See the Outcome Ledger tab.
• Drift Monitor — five drift signals (overlay cap consumption, multi-notch outlier events, basis inconsistency, migration pattern drift, sector-overlay drift) running continuously over the Outcome Ledger. See the Drift Monitor tab.
• Challenger Model Benchmark — Altman Z-Score, Ohlson O-Score, and Zmijewski model comparisons on every rating; SR 11-7 divergence flag fires at >2-notch disagreement with two or more challengers.

• Rating-accuracy pilot: 31-ticker S&P 500 vintage-matched sample (seed=42) measuring agency-notch agreement. MAE 2.61 notches, within-3 notches 71%, Spearman ρ 0.518. Acceptance thresholds documented (exact≥25%, within-1≥65%, within-2≥85%, |μ±|≤0.5). This measures level agreement against agency notches and is distinct from the default-discrimination backtest below.
• Default-discrimination backtest (executed): point-in-time historical backtest on 188 realized defaults (2001–2025) vs. a survivor control. S&P 500 cohort AUC 0.984 / Gini 0.967; full Russell 3000 small-cap breadth AUC ~0.95 with recall holding at 91.7%; same-rating-tier control 0.92–0.96. Backtest, not live performance; subject to survivorship limits in deep history. Full method in the SR 11-7 validation pack.
• Known bias: on the 31-ticker rating-accuracy pilot, Sentinel produces an average +1.23 notch IG bias on franchise/scale names. This is disclosed in the Known Limitations page and in every rating's limitations block.

• Every engine release carries a semantic version with a signed commit and corresponding Model Card hash
• Sector overlays (ceiling + uplift) are frozen in source as named constants with associated provenance comments referencing the empirical justification (e.g., rating-agency published sector caps)
• Calibration coefficients are frozen post-initial-fit and never re-fit on the same sample. Any recalibration triggers a major-version bump and a new Model Card hash
• Overlay additions/removals require a documented change memo and a smoke-test harness addition

• Per-rating Model Card — 20+ provenance accessors rolled up via getModelCard()
• Overlay attribution — every notch adjustment cites the rule that fired, the coefficient, and the methodology source
• Limitations block — every rating emits a dynamic list of applicable limitations keyed to the specific data and overlays used
• Adverse-action-ready rationale — IC Memo §2 is structured for Reg B adverse-action compatibility in lender use cases

• Not an NRSRO. Sentinel Credit, Inc. is not a Nationally Recognized Statistical Rating Organization and the outputs of this tool are analytical model outputs, not official credit ratings. They must not be represented as official ratings.
• No fair-lending or ECOA attestation as of this release. If the rating is used as an input to a consumer-credit decision subject to Regulation B, the institution deploying it is responsible for maintaining its own adverse-action reasoning consistency under its MRM framework.
• SR 11-7 alignment is self-attested. Independent third-party model-validation review is in progress. The SR 11-7 Evidence Pack documents the current evidence base.
• SOC 2 certification is not yet complete. Type I attestation targeted within 6 months of launch; Type II will follow the standard observation window. Until then, no attestation letter can be issued.

• Model outputs depend on the accuracy and completeness of publicly filed financial statements. Issuers with material off-balance-sheet exposures not disclosed in SEC filings will be under-reported in Sentinel's leverage and coverage metrics.
• Coverage of private companies requires manual PDF upload or direct field input; extraction accuracy depends on PDF quality and layout.
• LTM (last-twelve-month) computations require at least four sequential quarters of reported data; companies with interrupted filings default to the most recent full fiscal year with a seasonality-footnote flag.

• Launch baseline (2026-04-25): 50-ticker S&P 500 dual-fidelity backtest against a refreshed agency benchmark. Raw scoring engine: within-1 notch 52%, within-3 notches 76%, MAE 2.36 notches, Spearman ρ 0.51. The "raw" tier is the unadjusted three-agency submodel blend (35% S&P / 35% Moody's / 30% Fitch) before any post-composite overlays. This is the headline accuracy claim Sentinel makes to customers.
• Adjusted (post-overlay) tier: within-1 notch 36%, within-3 notches 70%, MAE 2.90. The post-composite overlay stack (TTC trough, peak ceiling, GW adj, FP lev, PI calibration, franchise, MVA, Cluster A, distress check, X-12b cap) is currently net-negative against agency match: it helped 10 names, hurt 22, was neutral on 18. The adjusted tier remains in the payload for transparency and is shown alongside raw in every IC Memo, but it is NOT the customer-facing accuracy claim. Per-overlay re-audit (PI calibration, franchise registry, peak ceiling sub-$50B cyclicals) is on the v1.1 roadmap.
• Acceptance thresholds (production target): within-1 ≥ 65%, within-2 ≥ 85%, |μ±| ≤ 0.5. Current within-1 of 52% is below this target; the 80% within-1 figure used in early customer conversations is an aspirational target, not a current measurement. The path to within-1 ≥ 65% requires structural changes to the qualitative-side scoring of legacy/cyclical mega-caps (industry_risk grid expansion, ticker fallback tier-5 adoption), not additional overlays.
• Franchise/scale IG bias: on the legacy 31-ticker pilot, Sentinel produced an average +1.23-notch bias on large investment-grade names. The 2026-04-25 50-ticker run shows the bias has shifted: legacy/cyclical mega-caps (DAL, UAL, T, F, DVN) are now substantially over-rated (raw notch Δ of −6 to −10) because the engine's qualitative scoring of competitive_advantage and scale_scope_diversity gives credit for size that agencies impose offsetting industry-decline penalties for. See structural-decline disclosure below.
• Confidence flags on the benchmark: 10 of 50 tickers in the 2026-04-25 benchmark are flagged medium- or low-confidence (CZR, GT, NWL, PARA, LYV, SPB, HPQ, EMR, HII, LVS) where the most recent S&P credit research note could not be located within the validation window. These are excluded from any subset accuracy claims when more conservative reporting is required.
• Expansion to 125-ticker stratified S&P 500 (Q3 2026) and Russell 3000 (Q4 2026) is planned.

• Aerospace & cyclicals: peak/trough EBITDA normalization is applied per SentinelCyclicalNormalizer logic. Companies at cyclical peaks may be flagged with the CYCLICAL_PEAK_SMOOTHED disclosure.
• Structural-decline / legacy-mega-cap divergence (STRUCTURAL_DECLINE_INDUSTRY_STRESS): for a documented set of industry clusters — legacy wireline and wireless telecommunications at scale, legacy major-network passenger airlines through and after the COVID-era recovery, sub-investment-grade auto OEMs, sub-mega-cap upstream Energy & Production, and live-entertainment platforms — Sentinel's three-agency submodel blend produces a materially higher rating than NRSRO consensus on the affected clusters (typically 5 to 10 notches less conservative than agency). The cause is well-understood: rating agencies impose industry-level structural-decline caps and BCA/SACP modifiers on these clusters that override the strong-quantitative-profile-from-scale signal that Sentinel's qualitative grid (industry_risk × competitive_advantage × scale_scope_diversity) produces. Sentinel does not currently replicate the agencies' qualitative cap on these clusters. This is a documented modeling choice, not a model error: Sentinel's view is that the long-horizon cash-flow and balance-sheet profile of issuers in these clusters does not fully justify the agencies' qualitative downward override, and we surface our independent quantitative read alongside the agency benchmark rather than back-fit to it. The runtime CCO-Review divergence flag (severity critical when |Sentinel − agency| ≥ 5 notches AND industry is in the documented cluster set) surfaces affected ratings prominently. Per-issuer cluster identification is available to enterprise customers under the MSA-gated methodology pack. Future work (v1.1+) is targeted at narrowing — but not eliminating — the divergence via expanded tier-5 industry_risk classification for structural-decline clusters.
• Banks & financial institutions: rated on the Financial Institutions tab with different frameworks (Basel capital adequacy, asset-quality metrics). The standard Rate a Company flow is not appropriate for banks.
• Insurance: not yet covered. Commercial insurers and reinsurers require specialized methodology not yet implemented.
• Municipal, sovereign, and structured credit: not covered.
• REITs: partially covered via standard corporate flow; specialized REIT covenant treatments and FFO-based metrics are on roadmap, not shipped.

• Beneish M-Score, Ohlson O-Score, Zmijewski challenger models are applied as benchmark comparisons per SR 11-7 §V.3. Divergence flags fire at >2-notch disagreement with two or more challengers.
• Generative layer: certain narrative explanations, when enabled, are produced via large language models. All LLM calls are recorded in the Outcome Ledger with a GENERATIVE_LAYER_DISCLOSURE flag. Numerical ratings are never produced by the LLM layer.
• Real-time news overlay: when enabled, the NEWS_EVENT_MODIFIER applies downward-only adjustment within the X-12 overlay envelope. Upward adjustments from positive news are not permitted.

• The 5-cell CCAR/DFAST-style ladder (base, revenue-down-10, margin-down-20, rates-up-200bp, combined) is illustrative, not a comprehensive stress framework. Institutions requiring full DFAST or ICAAP scenarios should supplement with internal scenario design.
• Stress-test scenarios do not capture jump-to-default tail risk.
• Scenario outputs are deterministic point estimates. No Monte Carlo distribution is produced for headline results; full-distribution covenant breach probability is available via the Predictive Intelligence tab.

• Rate-limited free tier: 10 ratings per day per account, 3 per minute burst. Paid tiers have expanded quotas; Enterprise has unlimited.
• Outcome Ledger stored in browser IndexedDB; users should periodically export via CSV/JSON for long-term preservation. Institutional deployments with centralized ledger storage are on roadmap.
• Third-party data sources (SEC EDGAR, FMP, Intrinio) have independent uptime and accuracy risks. Sentinel displays an error-sentinel banner when upstream fetches fail.