Before a claim can be graded, it has to be gradeable.
Below are six claims of the kind a firm might actually publish. Each one is run
through the same assessment RexUnda uses on real claims. Four of them do not
pass — and what the machine says about why is the part worth reading.
Synthetic exampleThese claims were written for this page. No firm submitted them, nothing here is stored, and nothing here is graded.
“NVDA rises over the next trading day.”
The simplest shape the policy can grade: one instrument, one direction, one window it has measured.
resolvable
This claim is resolvable: it commits to enough, and RexUnda has measured this kind of claim before. It is not yet graded — resolvable is not correct.
The most common shape of all, and the one that cannot be graded: with no window, there is no moment at which the claim ever comes due.
underspecified
This claim is not yet resolvable because it does not commit to 2 required elements (window, window.start_policy). Supply them and resubmit — nothing about the claim itself is being rejected.
machine reasons
MISSING_HORIZONMISSING_HORIZON_ANCHOR
state duration_value (> 0) and duration_unit (one of: hours, trading_days, calendar_days)
state when the clock starts — "into year-end" has no anchor, and an anchor chosen later is chosen with hindsight
An instrument the book has never priced. The claim is well-formed; the machine simply has no basis to grade it and says so.
unsupported
This claim is complete, but RexUnda cannot grade it today: subject.instrument_type. This is a limit of what we have measured, not a judgement of the claim.
A claim passing this assessment means only that it commits to enough for an
outcome to be checked later. It is not a view on whether the claim will hold,
and nothing on this page grades anything.
The refusals are not failures either. A claim the machine declines to grade is
a claim it has no measured basis for — saying so is the honest answer, and it
is why the boundary is drawn where it is rather than wherever would be
convenient.