The scenarios model converts source-backed operating drivers into downside, bear, base, and bull cases. It is designed to make sensitivities auditable: every case has explicit drivers, deterministic calculations, named deltas, and refusal codes when the case is just a story. This is backed by a deterministic API runner, not a forecast approval engine.
ALCHEMIST takes revenue growth, EBITDA margin, leverage, multiple, working capital, and capex assumptions, then emits a locked scenario table. It preserves the difference between a sensitivity, a management case, and an unsupported narrative.
COPYABLE SCENARIO PACKET Company: SampleCo Software Fiscal periods: CY2025 actuals baseline; CY2026E operating cases Purpose: Deterministic scenario table for a banking discussion Source references: - FY2025 Form 10-K filed 2026-02-28, revenue and adjusted EBITDA reconciliation: https://investors.samplecosoftware.example/sec/2025-10-k - Q1 2026 board plan extract dated 2026-04-21, pages 8-13 - Draft lender model v4, "Scenario_Output" tab, received 2026-05-06 - Customer-retention cohort backup for bull case is requested but not supplied Source facts and assumptions: - Base case revenue growth: 7.0% - Base case EBITDA margin: 27.0% - Base case CY2026E EBITDA: $4.10bn - Base case implied equity value: $27.4bn - Downside: 2.0% revenue growth, 23.0% EBITDA margin, $3.12bn CY2026E EBITDA, $18.7bn equity value - Bear: 4.0% revenue growth, 25.0% EBITDA margin, $3.55bn CY2026E EBITDA, $22.0bn equity value - Bull: 10.0% revenue growth, 29.0% EBITDA margin, $4.72bn CY2026E EBITDA, $34.1bn equity value - Bull customer-retention evidence is missing - Downside covenant cushion falls below 12.0% - No board minutes, lender-approved case, pipeline conversion backup, pricing study, or churn cohort file is supplied Screen rules: - Reproduce the four cases and identify the deltas from base. - Flag covenant cushion below 12.0% as WATCH. - Accept the bull case for arithmetic only and flag the missing retention support. - Preserve the difference between source plan, sensitivity, lender case, and approved forecast. - REFUSE to call any case management-approved, board-approved, lender-approved, financeable, or investable without evidence.
The base case ties to the source plan. Bull retention assumptions are accepted for math but flagged as not yet source-proven.
EXPECTED ALCHEMIST SCENARIO OUTPUT Verdict: COMPUTE - 1 UPSIDE PROOF GAP Source receipts: FY2025 10-K; Q1 2026 board plan extract pages 8-13; draft lender model v4 Fiscal basis: CY2025 actuals baseline; CY2026E scenario cases Assumption mode: deterministic API runner using supplied packet facts only; no forecast approval, lender approval, or board approval claim Case table: - Downside: 2.0% growth, 23.0% margin, $3.12bn EBITDA, $18.7bn equity value, WATCH covenant cushion under 12.0%. - Bear: 4.0% growth, 25.0% margin, $3.55bn EBITDA, $22.0bn equity value, WATCH pricing pressure persists. - Base: 7.0% growth, 27.0% margin, $4.10bn EBITDA, $27.4bn equity value, PASS ties to source plan. - Bull: 10.0% growth, 29.0% margin, $4.72bn EBITDA, $34.1bn equity value, WATCH retention proof missing. Deterministic flags: - PASS: cases use supplied drivers and values. - WATCH: downside covenant cushion needs credit review. - WATCH: bull retention support is missing, so upside is arithmetic-only. - MISSING EVIDENCE: board minutes, lender-approved case, pipeline conversion backup, pricing study, churn cohort file, and bull retention proof are not supplied. - ABSTAIN: no management approval, board approval, lender approval, financeability view, or investment recommendation supplied. Safe conclusion: The scenario table is useful for sensitivity discussion, but it does not approve a forecast, transaction, or financing decision.
| Case | Revenue growth | EBITDA margin | CY2026E EBITDA | Implied equity value | Flag |
|---|---|---|---|---|---|
| Downside | 2.0% | 23.0% | $3.12bn | $18.7bn | Covenant cushion under 12%; lender case not supplied |
| Bear | 4.0% | 25.0% | $3.55bn | $22.0bn | Pricing pressure persists per Q1 churn memo |
| Base | 7.0% | 27.0% | $4.10bn | $27.4bn | Ties to source plan; not board-approved evidence |
| Bull | 10.0% | 29.0% | $4.72bn | $34.1bn | Arithmetic-only upside; retention proof missing |
REFUSE Live deterministic API runner; this page computes scenario tables from supplied packets while preserving forecast-approval and recommendation boundaries.
REFUSE Refuses named scenarios when drivers are not provided or cannot be tied to a source case.
REFUSE Refuses to label a case as management, sponsor, lender, or board-approved without evidence.
REFUSE Flags circular logic, mixed time periods, and unsupported margin expansion.
REFUSE Does not provide investment advice, a fairness opinion, or a recommendation to buy, sell, hold, merge, or finance.
Paste this into an LLM and compare whether it separates computable cases from unsupported claims:
Create a four-case scenario table for SampleCo using only these source refs: FY2025 10-K, Q1 2026 board plan extract pages 8-13, and draft lender model v4. Fiscal basis is CY2025 actuals baseline and CY2026E scenario cases. Source plan: base revenue growth 7.0%, EBITDA margin 27.0%, CY2026E EBITDA $4.10bn, implied equity value $27.4bn. Downside is 2.0% growth and 23.0% margin with covenant cushion under 12.0%. Bear is 4.0% and 25.0%. Bull is 10.0% and 29.0%, but customer retention support is missing. Show deltas, flags, missing evidence, and do not call any case board-approved, lender-approved, financeable, or investable.
Scenario labels ALCHEMIST Keeps downside, bear, base, and bull tied to supplied facts. Ordinary LLM May rename cases or infer management intent.
Proof gaps ALCHEMIST Accepts bull arithmetic but flags missing retention support. Ordinary LLM Often treats upside assumptions as credible narrative.
Credit flags ALCHEMIST Names the downside covenant cushion watch item. Ordinary LLM May overlook covenant pressure.
Boundary ALCHEMIST Refuses approval, recommendation, or board posture. Ordinary LLM May turn scenario math into advice.
Edit the scenario packet. The deterministic runner expects period labels, driver rows, valuation outputs, source refs, approval evidence, and missing-proof hooks; it recomputes case rows and refuses to promote sensitivities into approved forecasts, financeability claims, or recommendations.
JOURDANLABSFetching API manifest metadataComputing verdict, blockers, refusals, metadata, and audit trail.
Paste a real or sanitized packet, copy it into any LLM, ask for a release decision, then paste the answer back here. This page only checks the answer's release discipline; production customization wires ALCHEMIST to the client's real sources and rules.
Separate computable sensitivity math from approved forecasts; preserve receipts and period labels; flag unsupported upside and refuse board, lender, management, financeability, or investment approval language without evidence.
Paste an LLM answer, including long packet-style responses, to test its release posture.