Adjusted Claims Percent haircuts a convertible claim by the chance it stops being a claim at all and turns into equity instead. To do that it needs a weight, and a weight needs a schedule that says which claims get haircut and by how much.
One such schedule is published and standing: the four-band step schedule on the Claims Grade page, live since 2026-06-14. This page publishes a second one, with five bands and different weights, as a candidate. It is written down and dated so it can be scored against evidence rather than argued from preference.
Nothing here is launched. No figure on this site is computed from this schedule, and none will be until a public calibration says it earns that. The document exists to be tested, not to be adopted.
Terms expanded on first use below: Adjusted Claims Percent (ACP), in the money (ITM), out of the money (OTM), Calibration 2 (CAL-2).
This is a candidate prior schedule, submitted to the Calibration 2 candidate family for scoring beside the published four-band step function. It is a discussion document with a date on it. Its standing is exhausted by that sentence.
What it is not. It is not ratified canon. It does not replace, amend, or supersede the four-band step schedule published 2026-06-14 on the Claims Grade page, which remains the standing stated prior named as such in Specification v1 section 3.1. It does not alter the section 8 calibration gate. It does not alter anything pre-registered for Calibration 2, whose rulebook was locked before any candidate was scored and stays locked.
The reason for the care is the same reason the calibration pages exist. A prior that can be revised after the evidence arrives is not a prior, it is a description of the evidence. So a new schedule enters as a candidate with a publication date, takes its scoring like every other candidate, and becomes canon only by clearing the stated bar. Writing one down is cheap. Promoting one is not.
CANDIDATE. Published 2026-08-15 for scoring. Not canon, not launched, not carried by any figure on this site.
Promotion path: the Calibration 2 disposition, on its published terms. There is no other route to standing, and this document does not create one.
A weight is the stated prior probability that an instrument resolves as a claim against the treasury rather than converting to equity. It applies to face in the ACP numerator only. The denominator is untouched, and so is every measurement surface.
Moneyness is the stock price divided by the conversion price, struck at the surface's as-of.
Direction, pinned. Every weight on this page is a standing weight, the fraction of the claim that remains senior. It is not a conversion weight. The invariant of Specification v1 section 3.1 holds here as everywhere: standing weight plus conversion weight equals 1. The published four-band schedule is expressed in the same direction, which is what makes the two comparable at all.
Band boundaries. Left closed, right open, matching the published schedule's explicit "at or above 130 percent". This convention is carried deliberately from the Calibration 2 rulings so that both schedules are scoreable under one reading rather than two.
| Band | Moneyness | Weight | Basis |
|---|---|---|---|
| Deep ITM | At or above 130% | 0.10 | Stated |
| ITM | 100 to 130% | 0.35 | Stated, CAL-2 target cell |
| Straddle | 85 to 100% | 0.65 | Stated, CAL-2 target cell |
| OTM | 70 to 85% | 0.85 | Stated |
| Deep OTM | Below 70% | 1.00 | Stated, behaves as debt |
Weights are standing weights on face. Bands are left closed, right open. Two cells are marked as Calibration 2 targets because those are the cells the study is positioned to speak to; they are stated here and stay stated until a study replaces them.
The bottom band, below 70 percent, behaves as debt and matches Claims Grade rubric score 3. The band thresholds at 130 percent and 70 percent deliberately mirror the rubric's Moneyness dimension, so the two surfaces speak one vocabulary rather than two dialects of the same idea.
Weight 1.00 is not a forecast. It is measurement unadjusted, a claim persisting in its default state with no event required. Every weight below 1.00 is a forecast of an event that has not happened, and in a glass box a forecast never reaches certainty. That asymmetry is the whole reason the top of the scale is a hard 1.00 and the bottom of it is a floor rather than a zero.
Three further grounds bound the floor above zero.
Error costs are asymmetric. Overweighting a claim understates common equity, which is conservative from the common shareholder's seat. Underweighting overstates it, which is the exact failure this framework exists to expose. A deep in the money instrument whose stock breaks below strike snaps back to full face instantly.
Weight zero would render an impossible state. The claim erased with no shares delivered anywhere, the note neither repaid nor converted, violates the either-state principle. Full conversion lives in the two-state bracket, where both sides of the world move together.
Moneyness does not settle resolution where settlement is the issuer's election. An instrument the issuer cash-settles resolves as a claim at any moneyness.
The candidate differs from the standing prior in every convertible cell, and it splits one band into two. Publishing the difference is the point. A candidate that is not stated against the thing it would replace cannot be scored against it either.
| Moneyness | Published, standing | Candidate |
|---|---|---|
| At or above 130% | 0.25 | 0.10 |
| 100 to 130% | 0.50 | 0.35 |
| 85 to 100% | 0.80 | 0.65 |
| 70 to 85% | 0.80 | 0.85 |
| Below 70% | 1.00 | 1.00 |
| Non-convertible | 1.00 | 1.00 |
The published schedule carries one band from 70 to 100 percent at 0.80, shown here across the two rows the candidate splits it into. Both columns are standing weights on face.
The published schedule stands as published, and the reasoning for that is on the Claims Grade page in its own words:
The weights are stated priors, published so they can be calibrated against observed conversions as the record grows, not asserted as truth.
That sentence is why this document can exist without disturbing anything. A stated prior invites a competing stated prior. What it does not invite is a quiet substitution, which is why the candidate publishes beside the standing schedule with the deltas shown rather than in place of it with the deltas absorbed.
Specification v1 section 8 holds the conversion weight behind a calibration gate. No computed conversion-weighted figure appears on any tracker surface until a calibration demonstrates, in public, that traded convertible instruments price within stated tolerance of the model's implied values, with the tolerance stated before results.
That gate holds today. Calibration 1 ran under it and no candidate cleared, the no-pass clause fired, and the conversion weight stayed unlaunched. Calibration 2 is pre-registered, locked, and carries no result. This document changes none of that.
Two consequences follow, and they are stated rather than implied.
No computed figure appears on this page or because of it. The schedule is published as stated weights, which is what the standing prior on the Claims Grade page already is. Publishing a weight is not the same act as publishing a figure computed from it, and only the second is gated.
A document cannot satisfy its own gate. The section 8 condition is an empirical result clearing a tolerance fixed in advance. No ratification, no publication and no dating of a schedule meets that condition, including this one. The gate's stated condition is the only thing that opens the gate.
CEBE framework, cebetracker.io. Candidate schedule submitted to the Calibration 2 family. The standing stated prior is the four-band schedule on Claims Grade. The canonical method definition is Adjusted Claims % Specification v1, and the gate this document leaves untouched is section 8.