Stable TapeSignals
How Stable Tape assesses changes
Stable Tape records and preserves the evidence before assessing it. A Signal would be a finding about an unusual change, supported by comparable records and rules set in advance. Most daily comparisons would not qualify.
A Signal would use the Daily Tape record and its verified daily comparison, called The Move. It could not repair, replace or reinterpret that evidence.
Signal status
- Public Signal methodology
- Rules for identifying unusual changes from sufficient evidence are under review.
- In development
- Public Signals issued
- None. No public Signal evaluator or ledger is active.
- None
- Current role
- Research and methodology development only. Stable Tape publishes its records and verified daily comparisons independently of Signal.
- Research
Methodology lineage: the earlier Daily Assessment used selection rules v5 and fixed presentation templates. Each canonical record distinguishes an assessment, an assessment withheld for lack of governed constituent evidence, or no evaluation. The last record carrying a Daily Assessment is 12 August 2026. That process is separate from the public Signal methodology in development. Read the published methodology.
See the latest verified daily comparisonSee how a Signal qualifies
Why a supply figure can change
00 · What the filtering is for
Supply figures move every day, and a changed figure alone does not establish what happened. The proposed Signal rules would identify unusual changes only where stronger evidence supports them.
Some of the movement is real but ordinary. Supply expands and contracts as holders enter and leave; a large mint on a quiet day and a large mint on a busy one are not the same event, and neither is remarkable on its own.
Some of it is not movement at all. A source revises a figure, a chain is reindexed, or a constituent appears or disappears from a provider's coverage. The reported figure changes without a corresponding change in actual supply.
And some of it cannot be compared. Where the prior reading is unavailable or was captured on different terms, the 2 figures are not measuring the same thing, so a difference between them is not a change.
A changed figure alone does not establish a change in supply.
Stable Tape preserves the record before assessing it. Each day's reading is frozen, dated and attributed to its source before interpretation. A comparison under the published rules requires preserved records for both observations. Where the required prior observation is missing or terms are not comparable, the governed comparison cannot turn that absence into ordinary 0 movement.
A future Signal would carry its evidence with it: which record, which comparison and which rule. Most daily comparisons would not qualify.
The Methodology records the Daily Assessment rules and the developing Signal work. This page explains the intended distinction without treating Signal as active.
4 possible assessment outcomes
01 · The shape of the answer
The methodology in development considers 4 types of subject and 4 possible determinations. Only 1 would be a Signal. The example below illustrates that proposed evaluation; no public Signal evaluator runs today.
-
Market
No new Signal
The aggregate could be evaluated. It moved, and the movement was ordinary for this record set.
-
Stablecoin Larkspur Dollar
New Signal
A qualifying condition that was not present at the previous comparison, with every required piece of evidence attached.
-
Chain Calder
Continuing state
A previously identified condition continues, without new information that warrants another report.
-
Issuer Thale Reserve
Unavailable
The current level is known. The earlier reading does not meet the comparison rules, so no change can be stated.
Insufficient evidence leaves the outcome unknown.
What would support a Signal
02 · How a change qualifies
A future published Signal would be 1 short sentence. Underneath it would sit a larger record: what was observed, which earlier records it could be compared with, which rule applied, how it relates to anything already reported, and what may be said in public. Open each stage.
Illustrative example. Not an issued Signal.
Larkspur Dollar supply rose by $96M against the previous record, its largest single-record expansion in this archive.
-
01What was observedMeasurement
The dated record the claim is bound to, and the exact scope it is allowed to describe. A Signal may only describe the part of the record that was admissible and comparable. Where only part of an object can be measured, the claim is limited to that part or it is not made at all.
- Record
- Illustrative
- Measurement basis
- Illustrative
- Permitted scope
- Only assets that meet the inclusion and comparison rules
-
02What can be comparedEligibility
A change is a statement about 2 observations, so the prior one has to exist and the 2 have to be comparable. In the governed Signal evaluation described here, a missing prior reading is not treated as 0, a gap is not filled in, and a renamed or reclassified entity cannot silently inherit the history of the thing it replaced. If the comparison cannot be made, the proposed outcome is unavailable rather than a number.
Comparability is judged for the specific claim being made. A missing input for 1 type of subject does not block an evaluation that never needed it.
- Comparison record
- Illustrative
- Identity continuity
- Required, and proved rather than assumed
- Missing prior reading
- Comparison withheld. Missing data is never treated as 0
-
03Which rule appliesEligibility
Each type of subject has a small set of named rule groups, and each group states in advance what evidence it needs and what it takes to pass. Each way of qualifying has its own requirements, so a single striking number cannot qualify on its own.
Rules are fixed before the date being assessed.
- Family
- Illustrative
- Methodology version
- In development. Not effective
- Thresholds
- Not yet calibrated or approved
-
04How overlapping candidates are handledEvent resolution
1 underlying occurrence can pass several rules at once. A large move in a single stablecoin can also register as a move in its issuer, its chain and the aggregate. Publishing all 4 would be 4 reports about 1 thing.
Candidates connected by a stated mechanical relation are grouped, and 1 is selected to represent the group in public. Happening on the same day is not a relation. Grouping never changes whether a candidate qualified: the others remain in the record as supporting members and stay inspectable.
- Grouping basis
- A stated evidence relation. Never coincidence
- Effect on qualification
- None. Grouping cannot change whether a candidate qualifies
- Supporting members
- Retained and auditable
-
05What may be publishedDistribution
Wording is assembled from facts the evidence already carries. Copy may not add a cause, a motive or a consequence that the record does not support, so a Signal describes what changed rather than why someone did it. Distribution decides how a finished Signal is carried. It cannot create one, suppress one, or make one sound larger.
Under the methodology in development, no Signal is published. Nothing it produces may be published.
- Copy authority
- Recorded facts only
- Effect on eligibility
- None
- Current publication state
- Not enabled
The order is fixedMeasurement, then eligibility, then event resolution, then distribution. No later stage may change an earlier result.
The 4 determinations
04 · What an evaluation can conclude
These 4 are the whole public vocabulary for the outcome of an evaluation. They describe 1 thing only: what the evidence supported for that object on that record.
New Signal
NEW_SIGNAL
A new qualifying condition, or a stated escalation of one already reported, under the rules in force for that record.
Continuing state
CONTINUING_STATE
A condition already identified is still running, without the new information that would warrant reporting it again. The condition remains visible on the record it belongs to.
No new Signal
NO_NEW_SIGNAL
The evaluation was possible and it was carried out. Nothing met the rule. This is the ordinary result on an ordinary day.
Unavailable
UNAVAILABLE
The evidence or the comparison was not sufficient to reach a determination at all. The reason is recorded.
3 separate questions, often confused
Assessment outcomes, methodology status and delivery are recorded separately.
What the evidence supported
New Signal, continuing state, no new Signal, unavailable. The result of 1 assessment.
What the methodology is
Draft, shadow, live, retired. This describes the rules, independently of any particular day. Internal trial evaluation publishes nothing.
What happened afterwards
Reviewed, published, delivered, corrected, withdrawn. This describes what happened to the output after the assessment was final.
A Signal can be determined and never published. A methodology can be live while 1 of its types of finding is not. These statuses are recorded separately.
Assessing changes across different sizes of stablecoin
05 · Entity fairness
2 ways of ranking a move give opposite answers. A methodology that picks either one on its own will be wrong in a predictable direction, and the same names will win every time.
A large stablecoin
- Level
- $148,000M
- Change
- +$1,180M
- Relative
- +0.80%
Large in dollars. Unremarkable for something this size.
A small stablecoin
- Level
- $610M
- Change
- +$96M
- Relative
- +15.74%
Small in dollars. Extraordinary against its own history.
By absolute size the first move is more than 12 times the second. By percentage the second is close to 20 times the first. Neither comparison settles whether either is worth reporting.
Neither test is used alone. A change must meet a minimum size and be unusual compared with the entity's own history. A large entity does not qualify for routine behaviour because the dollar figure is big, and a tiny entity does not qualify because a small base produced a large percentage.
Nothing about an entity's size, popularity or commercial relationship with Stable Tape enters that test.
What is settled here, and what is not
Settled in principle. Size alone does not determine eligibility. Percentage movement alone is never sufficient. Every route by which a smaller entity can qualify must also carry a minimum economic size. Sponsorship, popularity and expected engagement are excluded from eligibility entirely.
Not settled. Every number that would make those principles operable. The floors, the percentile cutoffs, the variability measures and the minimum history each family needs are open calibration questions. None has been approved, and none is stated on this page.
Not claimed. That entities of different sizes will produce Signals at similar rates, or that this design has been shown to be unbiased in practice. Fairness has to be measured across size bands once there is a record to measure, and a bias found that way is fixed by changing the rules going forward, not by editing past outcomes.
Evidence required for a Signal
06 · What exists today
A subject type can have a valid current reading while lacking a valid comparison, an approved rule or permission to publish. These requirements are checked separately.
| Subject type | Observations | Comparison | Enough evidence to apply a rule | Methodology | Publication |
|---|---|---|---|---|---|
| Market | Available | Available | Not yet established | In development | Not enabled |
| Stablecoins | Available | Not yet established | Not yet established | In development | Not enabled |
| Chains | Partial | Not yet established | Not yet established | In development | Not enabled |
| Issuers | Partial | Not yet established | Not yet established | In development | Not enabled |
How to read that table
Market. There is a usable current level and a usable prior-record comparison. That is the input a Signal would need. It is not evidence that a rule exists to judge it, that an evaluator runs, or that anything is authorised for publication.
Stablecoins, chains and issuers. Prospective directional history for individual constituents had not yet begun accumulating at the date checked. A first level is not a movement, and history reconstructed after the fact cannot stand in for it. Comparison also depends on stable identity and a consistent measurement basis, which are tracked separately from the numbers themselves.
Partial. Chain and issuer figures exist, but accounting them by source key is not the same as a settled canonical universe for those objects. The missing evidence is stated explicitly.
Not yet established means the input has not been proved sufficient. It is not 0.
What has to happen first
- 01A first admissible observation for the object.
- 02A first pair of observations that meets the comparison rules.
- 03Enough history for the particular rule being applied.
- 04Independent review, then calibration of every threshold.
- 05An internal trial period that evaluates observations without publishing results.
- 06An explicit decision to authorise publication.
Each condition requires evidence or approval. Elapsed time alone cannot satisfy it, and no timetable is set.
The proposed principles for every Signal
07 · The Signal Constitution
A methodology states thresholds. Above it sits a shorter document that states what any Stable Tape Signal methodology is allowed to do. A methodology may be stricter than these principles. It may not be looser.
Constitution
The principles
What any Signal methodology is allowed to do. Changed rarely, and never to reach a particular outcome.
governs
Methodology version
The rules in force
The families, the evidence each one needs and the thresholds. Versioned, and fixed before the records it judges.
evaluates
Determination
The outcome for 1 subject
1 of the 4 states, bound to the exact record and the exact rule version that produced it.
The 3 layers are versioned separately. A change to one is not a change to another.
DraftThe Signal Constitution is drafted and under review. It is not in force, and it carries no effective date.
Evidence before interpretation
A Signal may only arise from observations the record actually supports. No carried-forward value, no gap filled with a 0, no reconstruction presented as something measured at the time, and no economic story attached to a number that does not prove it.
Rules fixed before the outcome
Rules are published and versioned before the records they judge. A change applies to later records only. An effective version is never edited in place to alter a determination already made under it.
Independent eligibility
Popularity, follower counts, expected engagement, commercial relationships and editorial preference are excluded inputs. 2 entities with identical evidence get the identical outcome.
Consistent event resolution
Related candidates are grouped by stated rules, not by a daily quota or an editor's sense of how much is too much. Only findings supported by the evidence qualify; there is no output quota.
Explanation attached
Every Signal must be reproducible from its own evidence. A reader should be able to see which record, which comparison and which rule produced it, and a reader who disagrees should be able to say precisely where.
Accountability when things change
Corrections and withdrawals are recorded as additions to the record. Nothing is deleted, and criticism of the method gets a reasoned answer or is logged as an open question.
What versioning actually commits Stable Tape to
Rules can change, and they will. Each change is dated and applies going forward, and the rules that produced a past determination stay attached to it.
Each layer is versioned on its own: the principles, the methodology, the way things are measured, the register of entity identities, and the templates that turn a result into a sentence. Every determination names the exact versions it used. A change to one is not a change to the rest, and a wording change cannot carry a rule change.
2 version lines are separate and should not be run together. The principles document is at version 1. The methodology being drafted against it is the 7th in its own series, most of which were never public and several of which never left draft.
A hash proves that a rule set has not changed since it was published. It proves nothing about whether the rule was a good one or whether the source data was right. Both remain open to challenge.
How the rules change
08 · Version history
Rules will change. Every change is dated, applies prospectively, and leaves earlier readings untouched.
| Layer and version | Status | Effective | Rule hash | What it did |
|---|---|---|---|---|
| Constitution Signal Constitution v1 | Draft | Not assigned | Not assigned | Defines the principles every Signal methodology must follow. Still under review. |
| Selection rules Daily Assessment v4 | Superseded | Not published | Not published | Introduced 1 shared selection process and 1 public reading per record. Superseded by v5. |
| Selection rules Daily Assessment v5 | In force | Not published | Not published | Added repetition, reversal, dominance and cooldown rules, and separated selection rules from wording. This is the version used today. |
| Wording Presentation template v5-t6 | In force | Not published | Not published | Controls how the selected fact is written. It cannot change which fact was selected. |
| Candidate Signal v6 | Research, never public | None | None | Re-read frozen records to test whether a day contained new information. It never affected anything published. The published note on it. |
| Next Signal methodology | In development | Not assigned | Not assigned | The proposed rules identify unusual changes supported by sufficient evidence in market, stablecoin, blockchain and issuer observations. They are still in development. The published note on it. |
Why there are gaps
The 2 absences differ. "Not published" means the version operated, but Stable Tape never issued an effective instant or a rule hash for it. "Not assigned" means the version has not reached a state where either exists.
Neither cell is filled with an estimate.
The layers are versioned separately. Moving from template v5-t4 to v5-t6 changed the wording and not the selection, so it is not a selection-rule version.
Signal history: findings about unusual changes begins with the first issued Signal.
09 · The archive
Every issued Signal will have a permanent page tied to the record and methodology version that produced it. None has been issued yet.
No Signals yet.
The archive begins with the first issued Signal. Examples and test candidates never appear here.
Only issued Signals appear in the archive.
What each entry will carry
None of this exists yet.
- Identifier
- A stable Signal ID and 1 permanent address that keeps working
- Issued
- The date and time it was issued
- Subject
- The type of subject, the entity, and the family under which it qualified
- Determination
- The outcome, and the methodology version that reached it
- Evidence
- The record it was bound to, and the comparison it was measured against
- Reasoning
- Why it qualified
- Current validity
- Whether it still stands, and what changed if it does not
Kept together, these let a reader return years later and recover the rules that applied, the record behind them, why the Signal qualified, what was published, and what changed afterwards.
How later evidence can change a Signal’s status
10 · Corrections
Data can change after publication. When it does, we preserve the original statement, record the new evidence, and show what changed.
We keep 3 things separate: what was published, what changed later, and whether the original Signal still stands. If a Signal is withdrawn, the original remains visible with the reason and date of the change.
Original statement
Published from the evidence available at the time. The original remains permanently accessible.
New evidence
A provider restates a figure, or Stable Tape discovers an error in identity, measurement or wording.
Review
The Signal is re-tested under the methodology version that originally produced it.
Retained or updated
The Signal either keeps its status or receives a new one. The reason is added to its revision history.
A provider restatement does not automatically make the original fixing wrong. If Stable Tape captured the source faithfully, the original remains the record of what the source reported at the time. The revised evidence is recorded and evaluated separately.
-
The Signal
Larkspur Dollar supply rose by $96M against the previous record, its largest single-record expansion in this archive.
-
Original determination
New Signal under the methodology in force when it was issued.
-
Current status
Withdrawn
-
Revision history
1 entry: what changed, why it changed, and when the status changed.
-
The revision event
Records the underlying change and links every affected record or Signal.
The original statement and its methodology are preserved. Later status changes are recorded separately.
Not defined hereStatus labels are provisional. The revision system will define the final vocabulary for Signals and other revised records.
A withdrawn Signal should still show what it said, when it was issued, which methodology produced it, the record and comparison behind it, why its status changed, and when.
In developmentThis policy is still in development. Automated source-revision monitoring is not live yet. When the revision register launches, Signal status changes will link to the underlying revision event.
Where a Signal would lead
11 · Distribution
A Signal should lead back to the record it came from, and to the reasoning that produced it.
Daily Tape
Published
The dated records themselves. This is where any Signal would point. Read the latest Daily Tape.
Stable Briefing
Published
The daily edition, published today under the current methodology. Read the Stable Briefing.
Methodology
Published
The rules that select and phrase what is published today, including the thresholds behind them. Read the Methodology.
A page for each Signal
Not enabled
The intended destination: the claim, its evidence, its comparison, its limitations and its version bindings on 1 page. No such page exists yet because no Signal has been issued.
Signal alerts and feeds
Not enabled
No Signal notification, feed or interface is offered. There is nothing to subscribe to.
Commercial support does not determine Signal eligibility.
Daily Assessment lineageThe candidate research that preceded thisThe Daily Tape