Original CryptoSignalsReview dataset research

How the 3,200-Record Transparency Report Was Built

The method is designed to make absence, source concentration, and classification limits visible. It deliberately avoids market-size claims, provider rankings, and invented scores.

Analysis by CryptoSignalsReview Evidence Desk of 3,200 candidate records frozen 2026-07-05. Published 2026-07-10. Source commit 5dcd30b2b1a0. Dataset release. No provider paid for inclusion or status.

Read the reproducible method

No signup, payment, wallet connection, or credentials are required.

Unit of analysisUnique CSR slug

3,200 unique slugs in the snapshot.

Snapshot source5dcd30b2b1a0

Committed manifest, isolated from concurrent worktree changes.

Snapshot date2026-07-05

All records carry the same inventory update date.

Direct answer

The report deterministically aggregates the committed Provider Atlas manifest at source commit 5dcd30b2b1a0da9bacbaeec08244190dae19a49f. The denominator is 3,200 unique provider slugs. Missing fields remain missing, multi-platform term counts may overlap, quarantine is a relevance classification rather than a misconduct finding, and no metric is converted into a rating or recommendation.

Step 1: freeze the source

The generator can read a named Git reference through git show. This run used commit 5dcd30b2b1a0da9bacbaeec08244190dae19a49f, so the report is insulated from concurrent uncommitted Provider Atlas regeneration. The manifest’s own generated date is 2026-07-05. Re-running against a later commit should create a new measurement, not silently rewrite the old one.

Step 2: define the unit

One record equals one unique CSR slug. The snapshot contains 3,200 unique slugs and 3,133 unique display names. Display names can repeat, so name uniqueness is not used as the denominator. A record can describe a channel, group, provider candidate, project community, marketplace listing, or quarantined non-provider candidate.

The inventory is not a random sample. It combines public category directories, public group ratings, a supported-groups marketplace, manual provider seeds, and small Discord and WhatsApp source families. Results describe this inventory only.

Step 3: preserve missingness

A blank route, subscriber count, alias list, search-intent field, or editorial dossier remains blank. The calculation does not treat a missing website as proof that no website exists, or a missing subscriber count as zero subscribers. Coverage tables answer how much data is present, not what is true outside the dataset.

Step 4: calculate counts and shares

Measure typeFormulaOverlap ruleInterpretation
Status sharerecords in exact status / 3,200Mutually exclusive in current snapshotInventory workflow state.
Platform mentionplatform label contains term / 3,200Terms can overlapLabel concentration, not market share.
Field coveragenon-empty field / 3,200Fields can coexistCompleteness of stored metadata.
Source familyrecords with exact sourceKind / 3,200Mutually exclusive sourceKindDiscovery-source composition.
Quarantine typequarantined records in profileType / 3,200Exact current profile typeWhy a record is excluded from recommendation framing.
Collision groupnormalized value appears on at least two provider slugsOne record can occur in one group per fieldEntity-resolution queue, not duplicate proof.
Exact source concentrationrecords on exact trimmed sourceUrl / 3,200Every row contributes to one stored URLCollection-path concentration, not market share or independent corroboration.
Source-registry reconciliationprovider sourceUrl exactly matches a declared sources entryEvery row is matched or unmatchedInternal source-metadata agreement, not provider or source validation.
Field coverage by sourcepopulated field / records with exact sourceKindEach field uses its source-family denominatorImporter and source availability, not provider quality.
Country-language source cross-tabexact stored country and language labels grouped by sourceKindEvery row enters one exact pair and one source familyCollection taxonomy, not operator location, jurisdiction, audience geography, or current language.
Subscriber observationfinite nonnegative integer value / 3,200Missing remains missing; no imputationMetadata coverage, not active audience or customer count.
Subscriber quantilenearest-rank value among observed rowsObserved rows onlyDistribution of stored values, not a market estimate.
Metadata bundlefour-bit route/alias/subscriber/key presence maskEvery record enters exactly one of 16 possible patternsField co-occurrence and importer signature, not evidence score.
Platform-route cross-tabplatform label term crossed with matching route fieldLabels can overlap; each route field is tested separatelyStored classification agreement, not activity or ownership.
Source-family quarantine sharequarantined rows / rows with exact sourceKindOne mutually exclusive sourceKind per recordCSR classification within a collection path, not source or fraud rating.
Cornix reviewer denominatorstored reviewer count on 85 exact-sourceKind rowsMissing star values stay missing; reviewer counts are not deduplicatedMarketplace sample size, not verified-customer or unique-person count.
Cornix activity dependenceweekly value compared with monthly value * 3 / 13One residual test per Cornix rowField lineage, not independent activity or outcome proof.

Step 5: normalize conservatively for comparison

Name and alias comparison version nfkd-mark-lower-separator-v2 uses Unicode NFKD decomposition, removes combining marks, lowercases without a national locale, and folds non-letter and non-number runs into separators while retaining letters and numbers from every script. It is not full Unicode case-folding, homoglyph or confusable detection, transliteration, phonetic matching, or semantic similarity. Route version platform-route-v2 applies only declared Telegram and Discord endpoint rules; generic URL query strings, fragments, paths, and www hosts remain intact. Explicit dedupe keys use opaque trim-and-lower comparison version trim-lower-v1, not name normalization.

Original values remain the evidence; comparison values only create review queues. A collision group contains at least two provider slugs with the same non-empty key. 2 names produce an empty key and remain ungrouped. Counts labeled “excess” retain one record per group only as a mathematical baseline; they are not deletion or merge recommendations.

Step 6: separate classification from verification

Review status, result-sheet status, verification level, and quality quarantine are read from explicit manifest fields. The report does not infer a positive status from subscriber count, source family, country label, language label, platform, aliases, Cornix presence, or page position. It also does not infer fraud from quarantine, a repeated name, route reuse, a repeated dedupe key, or missing proof.

Data dictionary

FieldMeaning in this reportNot equivalent to
sourceKindDiscovery-source family assigned by CSROfficial ownership or independent corroboration
sourceUrlExact stored route through which the candidate entered the inventoryAn official provider route, independent source, or citation-quality score
platformNormalized platform labelCurrent activity or market share
reviewStatusCSR workflow conclusion displayed on the recordRegulatory status or user rating
verificationLevelCurrent evidence gateProvider quality score
resultSheetStatusWhether CSR completed a result-sheet reviewA provider’s own screenshot or testimonial
qualityStatusQuarantine or correction workflow markerFraud allegation
subscribersStored public catalog count where availableActive users, customers, or profitable followers
aliasesAlternative names used for resolutionEndorsements or affiliated entities
dedupeKeySource-derived comparison key where populatedVerified entity identity or automatic merge instruction
normalized nameVersioned NFKD, mark, lower, and separator comparison keyCanonical provider name, full case-folding, or confusable detection
canonical routePlatform-specific endpoint comparison keyCurrent ownership, authorization, or service continuity
source URL concentrationRecord shares across exact trimmed sourceUrl valuesProvider-market share, web traffic, or source independence
source registry matchExact provider sourceUrl also appears in the manifest sources registryProvider validation, source authority, or independent corroboration
country labelExact stored geographic taxonomy on one provider rowOperator location, legal jurisdiction, audience geography, or market availability
language labelExact stored language taxonomy on one provider rowCurrent message language, support language, contract language, or translation quality
subscriber quantileNearest-rank position among finite nonnegative integer observationsTypical market audience, active users, or customers
metadata bundle maskPresence order: direct route, alias, subscriber observation, explicit keyEvidence quality, identity resolution, or verification status
label-without-routePlatform term appears in the stored label while its matching direct URL field is blankProof that the provider lacks the platform outside this snapshot
Cornix reviewer countFrozen marketplace denominator attached to a record’s star metadataUnique people, verified customers, independent reviews, or CSR rating
Cornix weekly activityStored monthly count transformed by 3/13 in this snapshotIndependent observation, successful signals, trades, or follower outcomes

Reproduction commands

node tools/generate-crypto-signal-provider-transparency-evidence-library.mjs --git-ref 5dcd30b2b1a0da9bacbaeec08244190dae19a49f --distribution-ref 60368e8bdc778678738af1b1534253bdcbd3bd27 --distribution-json-url https://gist.githubusercontent.com/TheCryptoSimon/943c15bdd232eca0047603e33200dba0/raw/ffb007a02c83a78e8f0d505bfd7e712fca57cc50/summary.json --distribution-csv-url https://gist.githubusercontent.com/TheCryptoSimon/943c15bdd232eca0047603e33200dba0/raw/c3a134b5e031a67537d21ae898de785f1ff50df8/summary.csv --distribution-version 91c6a4d3f781fce63acd8b9689a8511b60269370 --distribution-page-url https://gist.github.com/TheCryptoSimon/943c15bdd232eca0047603e33200dba0 --published-at 2026-07-10 --modified-at 2026-07-10
node tools/check-crypto-signal-provider-transparency-evidence-library.mjs --verify-public-distributions
node tools/audit-evidence-library-information-gain.mjs --library crypto-signal-provider-transparency-evidence-library --fail-on-risk

The generated manifest stores the source commit, aggregate metrics, page metadata, word counts, repository commit, public release identity, byte sizes, and checksums. The canonical landing page links to immutable aggregate JSON and CSV files; the live data-table article exposes the same measurements to readers.

Cite and download this release

Use the canonical dataset landing page, snapshot date 2026-07-05, source commit 5dcd30b2b1a0da9bacbaeec08244190dae19a49f, and the metric’s stated denominator. The aggregate files below are frozen in public release 91c6a4d3f781fce63acd8b9689a8511b60269370; repository commit 60368e8bdc778678738af1b1534253bdcbd3bd27 preserves the matching internal build.

CryptoSignalsReview Evidence Desk. CryptoSignalsReview Provider Transparency Snapshot 2026-07-05. Published 2026-07-10. https://cryptosignalsreview.com/crypto-signal-provider-transparency-evidence-library/

Aggregate reuse terms: CSR permits quotation and reuse of the aggregate JSON and CSV with attribution to CryptoSignalsReview, the canonical landing page, snapshot date, and frozen source commit. This permission covers CSR’s aggregate analysis only; it does not grant rights in third-party names, marks, routes, or source-catalog content. Reuse must preserve the non-representative, non-verification, and non-performance boundaries.

Dataset boundary

This is a census of records in the CryptoSignalsReview candidate inventory at snapshot 2026-07-05, not a representative survey of every crypto signal provider or every messaging channel. Directory inclusion is discovery evidence only. Missing fields stay missing, platform labels can overlap, and no count proves provider quality, legality, profitability, or safety.

Frozen source: 5dcd30b2b1a0da9bacbaeec08244190dae19a49f. Unit: one unique CSR provider slug. Position: coverage is not endorsement.

Official context sources

These sources explain why identity, disclosure, complete performance evidence, and resistance to urgency matter. They do not validate any record in the CSR dataset.

Continue the transparency research