Every claim the Investment Research demo makes is either mechanically gated, derived from declared inputs, or listed below as an open gap. This page is the benchmark of record for the suite — including the part most product pages omit: the register of what has not been built or proven.
No published standard enumerates investment research capability. That is worth saying plainly, because the other estates in this suite can point at one and this estate cannot.
| Standard | What it actually governs |
|---|---|
FINRA 2241 | Analyst conflicts of interest, disclosure and firewalls. Conduct, not capability — it says nothing about what analysis should be performed. |
Reg AC | Analyst certification that views are their own honest opinion. A certification requirement, not a scope of work. |
MiFID II | Research unbundling and inducements. Commercial arrangement, not analytical coverage. |
CFA ROS | Research objectivity standards — independence and integrity of process. Again conduct. |
Every one of these is a conduct regime. None enumerates the analytical work a research function should be able to do, and buy-side and sell-side would draw that list differently from each other in any case. Inventing a citation here would be worse than admitting there is none.
So the denominator is self-declared, with a published criterion. An item is in the 21 if it is analytical work a research desk is routinely asked for and whose absence a portfolio manager or IC would notice — earnings response, initiation, rating change, credit recovery, merger arbitrage, macro sweep, activist situation, and so on. Each is enumerated on this page with the lane that exercises it.
Excluded by design. Trade execution and order routing (an OMS/EMS function, not research). Portfolio construction and optimisation (a PM function). Client suitability and advice (a wealth function under a different regime). Proprietary alpha models (institution-specific by definition). Each is excluded because it belongs to a different system, not because it is hard.
What 69.0% therefore means. Thirteen areas in full and three in part of twenty-one that we enumerated and defended above. It is checkable against the register printed on this page and against nothing else — which is a weaker claim than KYC's FFIEC citation, and is stated as such rather than dressed up.
No file ships unless the full battery prints ALL 6 IR GATES PASS — packaging is literally grep-gated on that string. Every founder-caught field defect became a permanent assertion.
| GATE | WHAT IT PROVES |
|---|---|
| GATE 1 · CLAIMS | Bans stale counts, era text, absolutist compliance language, weak-hash relics — including en-dash variants that once evaded ASCII bans. |
| GATE 2 · STRUCTURE | Tag balance and full script parse on every page. |
| GATE 3 · VISUAL CANON | One nav, one byte-identical footer, banner canon, version stamps, feature presence — 60+ assertions. |
| GATE 4 · HEADLESS RUNTIME | Pages must RUN clean, not merely parse — null-for-absent-ids shim born from a field crash. |
| GATE 5 · REAL DOM | jsdom executes every page: zero errors, computed-visible nav, BOTH auth paths of the demo must yield all scenario pills; pure-SHA parity vs node crypto. |
| GATE 6 · VALUATION ORACLE | Independent reimplementation of every engine formula, asserted against the page's own declared inputs — ten derivations, six composites, drift-bans. |
Displayed figures are computed at runtime by IRVAL v1.0 from declared inputs — and Gate 6 reimplements every formula independently and asserts the outputs against the page's own data. Conviction tiers are composites of five declared dimensions (threshold: ≥75 HIGH · ≥50 MEDIUM). You can reproduce all of it yourself: export the ⬖ Research Binder and drop it on ⛊ Verify — your browser recomputes every SHA-256 and every derivation with no server involved.
| SCENARIO | TICKER | BASIS | DECLARED INPUTS | FORMULA | DERIVED |
|---|---|---|---|---|---|
| SC01 | VYRA | price | PT $215 · price $183 | (PT − price) / price | +17.5% |
| SC02 | SFG | price | PT $240 · price $213.70 | (PT − price) / price | +12.3% |
| SC03 | PHC | recovery | entry 100 · recovery 55 | (recovery − entry) / entry | −45.0% |
| SC04 | VLTN | price | PT $300 · price $395.30 | (PT − price) / price | −24.1% |
| SC05 | CLSP | price | PT $465 · price $410 | (PT − price) / price | +13.4% |
| SC06 | Nimbus | arb | target $36.25 · offer $38.50 | (offer − target) / target | +6.2% arb |
| SC08 | — | price | PT $540 · price $478.70 | (PT − price) / price | +12.8% |
| SC09 | — | price | PT $420 · price $480 | (PT − price) / price | −12.5% |
| SC10 | — | price | PT $38 · price $72 | (PT − price) / price | −47.2% |
| SC12 | HLX | price | PT $96 (candidate) · price $82.50 | (PT − price) / price | +16.4% |
Routing lanes come from the fingerprinted policy pack (IR-2026.07) — the same pack whose hash stamps every ledger row. Change one rule in the ⚖ Sandbox and watch the fingerprint move; the restricted-list hard block is displayed locked: statutory-class controls are not sandboxable.
| SC | SCENARIO | LANE | HUMAN AUTHORITY | ENGINE-DERIVED |
|---|---|---|---|---|
| 01 | Earnings Flash | PRINCIPAL_FAST | Auto-clear · rated note certified + principal-released | YES |
| 02 | Coverage Initiation | PRINCIPAL | Analyst sign-off | YES |
| 03 | Credit · IC Committee | COMMITTEE | IC signatures | YES (recovery basis) |
| 04 | PM Ad Hoc Downgrade | PRINCIPAL | Analyst sign-off | YES |
| 05 | Perpetual Monitor | PRINCIPAL | Analyst sign-off | YES |
| 06 | M&A Event-Driven | COMMITTEE | IC signatures | YES (arb basis) |
| 07 | FOMC Macro Sweep | PRINCIPAL | Analyst sign-off | N/A — portfolio-level reweight, no single PT by design |
| 08 | Activist / 13D | PRINCIPAL | Analyst sign-off | YES |
| 09 | Earnings Miss Downgrade | PRINCIPAL | Analyst sign-off | YES |
| 10 | Short Thesis Initiation | COMMITTEE | IC signatures | YES |
| 11 | Restricted / MNPI | BLOCK | Interdicted — no research produced | N/A — blocked upstream by design |
| 12 | Analyst Override | PRINCIPAL | Candidate sealed · override + rationale ledgered | YES (candidate derived) |
| 13 | Evidence Conflict | COMMITTEE | Abstained · routed with conflict quantified | N/A — abstained by design (dual basis preserved) |
A benchmark that only lists what works is marketing. These are the open gaps, stated plainly; each one is pilot or production scope, and several are hard gates.
