Protocol releasev0.1.0No comparative results

Solana wallet intelligence,
tested before marketed.

A preregistered protocol for comparing position accounting, funding provenance, coordination, leader/follower inference, freshness, null honesty and evidence traceability. The rules are public now. Provider scorecards are not.

Version0.1.0
Release statusProtocol only
Comparative resultsNot published
Registered providers0
Frozen public cases0
Protocol test definitions30
Comparative release gate25–50 independently adjudicated cases
What this release proves

The benchmark cannot move its goalposts after seeing the answers.

Version 0.1 fixes the questions, admission rules, abstention semantics and reporting requirements before a provider is tested. It does not prove a performance advantage for Conyr or anyone else.

01 · Evidence classes

A capability claim is not a measurement.

The public report will keep three kinds of evidence separate. Only an executed run against the frozen manifest can produce a score.

Measured result

Not released

Requires an executed, timestamped run against the frozen manifest. Version 0.1 contains no measured provider results and names no winner.

Documentation observation

Not scored

A provider may document a capability, but documentation is used only to design a fair request. It never becomes accuracy, coverage or freshness points.

Conyr internal finding

Excluded

Internal research can nominate a future case. It cannot serve as independent benchmark truth until the evidence is frozen and separately adjudicated.

02 · Seven evaluation axes

Accuracy, depth, timeliness—and the courage to say unknown.

Each axis defines its unit of analysis, independent evidence requirement, reported measures and abstention behavior. A broad feature checklist cannot substitute for these tests.

01

Position accounting

5 tests · definition only

Does the provider reconstruct completed positions instead of treating profitable exits as independent wins?

Unit

One wallet and token mint over a frozen slot range

Independent ground truth

An independently reconstructed inventory ledger containing buys, sells, transfers, fees, lot allocation, remaining inventory and source transaction signatures.

Measures
  • Completed-position outcome accuracy
  • Realized PnL absolute and percentage error
  • Open inventory quantity error
  • Partial-history and unsupported-inventory abstention accuracy
Abstention rule

A documented unknown is correct when the frozen ledger cannot establish cost basis. Coercing unknown inventory to zero or profit is an error.

Preregistered tests
PA-01Closed round trip
PA-02Partial exits
PA-03Open inventory
PA-04Transfer-acquired inventory
PA-05Fee-aware lot allocation
02

Funding provenance depth

5 tests · definition only

How far can a provider follow capital without inventing an origin?

Unit

One wallet and its pre-cutoff native-SOL funding path

Independent ground truth

A transaction-signature-backed path adjudicated hop by hop, with service and infrastructure roots verified separately from the provider under test.

Measures
  • Direct-funder accuracy
  • Valid path depth before termination
  • Recognized-root accuracy
  • False attribution rate
Abstention rule

Unknown is preferable to an unsupported entity attribution. Unresolved paths are not scored as self-funded wallets.

Preregistered tests
PV-01Direct funder
PV-02Multi-hop path
PV-03Recognized infrastructure root
PV-04Unresolved root
PV-05Ambiguous funding event
03

Coordination discovery

4 tests · definition only

Can the provider find repeated on-chain relationships without equating co-occurrence with common ownership?

Unit

One token episode and an adjudicated wallet cohort

Independent ground truth

Positive and negative wallet relationships reviewed from funding, execution, timing, size, recurrence and infrastructure evidence; intent is never used as ground truth.

Measures
  • Relationship precision and recall
  • False grouping of shared infrastructure users
  • Evidence-domain coverage
  • Calibrated confidence or explicit abstention
Abstention rule

The benchmark grades detected on-chain relationships, not claims about legal ownership, identity or malicious intent.

Preregistered tests
CD-01Repeated timing relationship
CD-02Shared private funder
CD-03Shared infrastructure negative
CD-04Hot-token co-occurrence negative
04

Leader/follower inference

4 tests · definition only

Can the provider recover a repeat, directed trading relationship and support its direction?

Unit

One wallet pair across multiple shared token episodes

Independent ground truth

A blinded pair adjudication using repeated event order, lag distribution, shared-mint support, size resemblance and counterexamples where the apparent direction reverses.

Measures
  • Directed-edge precision and recall
  • Direction accuracy
  • Lag and support-count error
  • Single-event false-positive rate
Abstention rule

One coincident trade is insufficient. A provider can abstain when repeat support does not clear its published threshold.

Preregistered tests
LF-01Repeated directed edge
LF-02Direction reversal negative
LF-03Single-event negative
LF-04Lag and support disclosure
05

Freshness

4 tests · definition only

How long after an eligible on-chain event does the tested result become available?

Unit

One live, pre-registered event observed from a common Solana slot reference

Independent ground truth

The chain slot and event signature, an external observation timestamp and the first successful provider response containing the eligible update.

Measures
  • Availability latency p50 and p95
  • Timeout and stale-response rate
  • Event-to-derived-intelligence latency
  • Observation-window sample count
Abstention rule

Documentation language such as real time is not a measurement. Freshness is reported only from timestamped live trials.

Preregistered tests
FR-01Raw-event availability
FR-02Position-state update
FR-03Provenance update
FR-04Stale-response detection
06

Null honesty

4 tests · definition only

Does the provider preserve the difference between zero, empty, unavailable, unsupported and unknown?

Unit

One deliberately selected boundary or incomplete-evidence case

Independent ground truth

An adjudicated answer state paired with the evidence that makes a numeric zero, empty set, unknown value or unsupported request appropriate.

Measures
  • Answer-state classification accuracy
  • Unsupported-as-zero rate
  • Unknown-as-fact rate
  • Machine-readable partial or error semantics
Abstention rule

A clear unknown or unsupported answer can receive full credit. A plausible-looking value without sufficient evidence cannot.

Preregistered tests
NH-01Measured zero
NH-02Valid empty set
NH-03Unknown value
NH-04Unsupported or partial response
07

Evidence traceability

4 tests · definition only

Can a user inspect the chain evidence behind the answer?

Unit

One normalized provider answer from any other axis

Independent ground truth

The frozen case record and its public transaction, slot, address, mint and time references.

Measures
  • Transaction-signature coverage
  • Slot and timestamp coverage
  • Method or data-version disclosure
  • Stable evidence-reference coverage
Abstention rule

A result is not made correct merely by attaching a link. References must resolve to evidence relevant to the normalized answer.

Preregistered tests
ET-01Transaction references
ET-02Slot and event time
ET-03Method and data version
ET-04Stable evidence resolution
03 · Case admission

Zero frozen cases is the honest v0.1 result.

The repository contains internal wallet studies and curated labels, but those are not independent truth for a comparison maintained by Conyr. No internal case has been promoted into the benchmark fixture.

Public identifiers

Only public Solana wallet addresses, token mints, slots and transaction signatures may identify a case.

Frozen boundary

Every case fixes the last eligible slot or time before any provider is queried.

Independent truth

Ground truth is reconstructed from chain evidence without reading the answers being scored.

Balanced set

The cohort must include positives, negatives, thin-history cases, ambiguous paths and legitimate abstentions.

No allegation

Coordination cases describe observed relationships, not legal ownership, identity, fraud or intent.

No cherry-picking

A case cannot enter because Conyr already returns a favorable or distinctive answer for it.

v0.2 gate25–50 cases

The next release must publish independently adjudicated positive and negative wallet or token cases, their evidence references, frozen cutoff and manifest hash before any tested provider is queried. Until then, the empty `cases`, `providers`, `runs` and `results` arrays in the manifest are intentional.

04 · Run protocol

Freeze, capture, normalize, adjudicate, report.

The sequence prevents case selection and scoring rules from drifting after a provider response is visible.

01

Freeze before querying

Publish the case manifest and content hash first. Record every case identifier, cutoff, adjudicated answer state and evidence reference before collecting provider output.

02

Register the surface

Record provider, product, API version, paid tier, test date and configuration. Use an interface available to ordinary customers on that disclosed tier.

03

Capture without repair

Record timestamps, status, retries and a raw-response hash. Normalization may map fields, but it cannot fill missing values or silently turn unknown into zero.

04

Adjudicate blind

Where formats permit, reviewers see normalized answers without provider identity. Disagreements are resolved against public chain evidence and the rationale is retained.

05

Report denominators

Every rate ships with sample size, eligibility, missing count and uncertainty. Per-axis measurements appear before any optional weighted summary.

05 · Reporting contract

No denominator, no score.

Every published rate must expose what was eligible, missing and uncertain. Version 0.1 defines no overall winner because different users value the seven axes differently.

Always publishSample size, eligible count, missing count, metric definition, point estimate and uncertainty interval.
Run identityProvider, product, API version, disclosed tier, run timestamp, retry policy and response hashes.
EvidenceTransaction, slot, time and case references sufficient to audit each normalized answer where data rights permit.
Unavailable outputRecord it as unavailable or untestable. Do not invent a zero and do not silently remove the case.
CorrectionsAppend a versioned correction with rationale. Never replace the historical result without a trace.
Composite rankingNone in v0.1. Per-axis measurements come first; any future weighting must be separately disclosed.
06 · Limitations

A fair protocol still has boundaries.

These constraints will remain visible beside future results rather than being buried in a methodology appendix.

No result yet

This page is a preregistered protocol, not evidence that Conyr or another provider wins any axis.

Historical coverage differs

Provider retention and launch dates can make a case testable for one surface and unavailable for another. Availability is reported, not guessed.

Price models differ

PnL comparisons must hold price source, fees, lot method, transfers and cutoff constant or publish the remaining semantic difference.

Identity is not observable

On-chain patterns can support relationships and provenance; they do not independently prove a person or beneficial owner.

Latency is time-dependent

Freshness requires repeated live trials and can change by load, plan, region and incident state. Documentation is not a latency sample.

Publisher conflict

Conyr maintains this protocol and is also a potential participant. Frozen cases, blind review and public evidence reduce—but do not eliminate—that conflict.

07 · FAQ

What can be cited today.

The short answer is deliberately conservative: the protocol exists; the comparison does not yet.

Does v0.1 say Conyr is the best Solana wallet-intelligence provider?

No. Version 0.1 publishes the rules before the cases and measurements. It contains no comparative scores and declares no winner.

Why publish an empty case fixture?

Because Conyr's internal observations and curated labels are not independent cross-provider ground truth. Publishing them as truth would make the benchmark easier for its maintainer. The empty fixture makes that limitation machine-readable.

When can comparative results be published?

After 25 to 50 public wallet or token cases are frozen, independently adjudicated, evidence-linked and hashed before any tested provider is queried.

Can a provider receive credit for returning unknown?

Yes. When available evidence cannot support a value, a precise unknown or unsupported response is more correct than a plausible fabricated number.

Machine-readable release

Inspect the same protocol without the prose.

The JSON manifest exposes the release state, evidence classes, cohort contract, seven axes, run rules, scoring policy and empty result fixture. It contains no credentials, private data, query shapes or database information.