for advisers

Before it goes out,
know which claims can be checked later.

Most market claims cannot be graded — not because they are wrong, but because they were never stated in a form that comes due. RexUnda tells you which is which, in advance, and keeps the record of what was said and what stood behind it. This page is the whole boundary: what it grades, what it refuses, what a firm would receive, and what is not built yet.

Check a claim — free, nothing stored How we keep our own
what it grades today

Four shapes, and it says so.

A claim is gradeable when it names an instrument, a direction, a clock, and something to be measured against. These are the forms the book has actually measured — not the forms it could imagine measuring.

absolute
goes up
move_up
absolute
goes down
move_down
relative
beats the market
outperform
relative
trails the market
underperform

Instruments: equity, etf. Windows: 4h, 24h, 72h. Compared against: SPY. Anything outside that is refused below rather than approximated.

what it refuses

A refusal is an answer, and it names the reason.

These are well-formed claims the machine declines to grade. A vendor who hides the boundary is selling you a discovery you will make later anyway. Every refusal carries machine-readable reasons — 22 of them exist — so the gap is specific rather than a shrug.

refused
goes up by at least X%
asks for a size the book has never graded against. Our own floor is calibrated from measured volatility, not chosen by the claimant, and scoring against a threshold you pick would move every rate we publish.
refused
goes down by at least X%
asks for a size the book has never graded against. Our own floor is calibrated from measured volatility, not chosen by the claimant, and scoring against a threshold you pick would move every rate we publish.
refused
beats the market by at least X%
asks by HOW MUCH something beat the benchmark. We grade excess against SPY as a direction only — whether it beat the market, never by how far — so the size of an outperformance is a quantity this book has never measured.
refused
trails the market by at least X%
asks by HOW MUCH something trailed the benchmark. We grade excess against SPY as a direction only — whether it trailed the market, never by how far — so the size of an underperformance is a quantity this book has never measured.
refused
reaches X% at any point
asks whether a price was ever reached at any point in the window. The book records where a window opened and where it closed, not the route between — so a touch that happened and reversed is invisible to it, and always will be under this record.
refused
stays above X% for the whole window
asks whether a level held for every moment of the window. That is a statement about the whole path, and the book keeps two prints per window — a claim that failed briefly and recovered would read as a pass, which is worse than refusing it.
refused
stays below X% for the whole window
asks whether a ceiling held for every moment of the window. The book keeps an opening and a closing print, so a breach that reversed before the close would be scored as if it never happened.
refused
stays between two levels for the whole window
asks that a price hold between two levels for the whole window — a path question twice over. ⚑ And it cannot be STATED here at all: the contract carries one threshold, so there is no field for the second side of a band. We will not accept a single number and treat it as the claim, because that would grade half of what you said.
refused
closes above a set price
names an absolute price rather than a move. The closing print exists, so this is computable — but a level is the one threshold a split or a dividend silently invalidates, and the house has never declared how a price level is adjusted for corporate actions. Grading it would mean choosing that rule after seeing the claim.
refused
closes below a set price
names an absolute price rather than a move. The closing print exists, so this is computable — but a level is the one threshold a split or a dividend silently invalidates, and the house has never declared how a price level is adjusted for corporate actions. Grading it would mean choosing that rule after seeing the claim.
refused
beats another named company
compares one instrument to another company rather than to the market. Every excess number this book holds is computed against SPY specifically — not against a configurable benchmark that happens to be set to SPY — so a peer comparison is a measurement that has never been made, not a parameter that has never been changed.

Also refused: instruments the book has never priced, claims with no window, and horizons it has no measured reference for. Six worked examples show four of these being turned down, with the exact reason each one could not be graded.

what you actually get

A judgment on the claim, and the reasons behind it.

Today, live and free: a structured claim goes in, and the machine returns one of three verdicts — resolvable, underspecified, or unsupported — with machine-readable reasons attached. If a claim cannot be graded, the output names exactly which criterion is missing or which part falls outside the envelope above. That refusal is the finished edge of this product, and it is the artifact worth having before something goes out.

⚑ Resolvable means a grade is possible. It never means a claim was right. Nothing on this page grades your calls.

Live today
The verdict and its reason codes, at /claim-check. No account, no stored row, no charge.
Built, uncalled
Sealing, the frozen text hash, the immutability rule and the pinned grading standard exist in the database and are wired to nothing. Zero tenants, zero claims, and no claim of yours has been graded — because none exists.

A record being complete is not the same as a claim being right. RexUnda keeps the first and never asserts the second — and it does not carry your obligations for you.

what is not being sold

Four things a firm would reasonably assume, and shouldn't.

not built
Portfolio-level grading
One claim at a time. There is no batch intake, no book-wide rollup, and no aggregate score for a firm's published output.
not priced
Options and derivatives
The book has never priced them, so a claim on one is refused rather than approximated.
no reference
Long-horizon thesis packets
A quarter or a year is a different question, not a longer version of the same one, and there is no measured reference to judge it against.
never
A stamp you can point at
This is the machine-auditable backup file behind your own work. It is not a third-party sign-off and does not discharge anything you are responsible for.
how to start

You cannot buy this yet.

The founding cohort is five slots at a $199/mo founding rate. Not open yet: it onboards after securities-counsel review, September target. Nothing is charged, no account is created, and there is no checkout on this site.

What you can do today costs nothing and never stores your claim: structure a claim and see whether it could be graded at all. Most cannot, and the reason is specific. If the boundary above fits the work you actually publish, that is the conversation worth having when the door opens.

Check a claim Join the waitlist