Methodology

Data sources

RobinIndex reads three public sources: the Robinhood Chain Blockscout explorer (tokens, contracts, verification status, holder and transfer counters, network statistics), the official Robinhood Chain documentation (canonical contract addresses, maintained as a curated config reviewed by hand), and the GitHub search API (public repositories that mention the chain).

RobinIndex never reads private data, never requires wallets, and never executes transactions.

Refresh cadence

Heavy indexing runs every hour: token and contract discovery, verified-contract sync, official docs, GitHub discovery, deduplication, recategorization, Index Score refresh, historical snapshots, and Pulse change detection.

Live metrics refresh up to every 60 seconds: network counters and per-token holder/transfer counters for pages currently being viewed. These are cached server-side so many visitors share one upstream request.

Freshness labels (fresh / stale / unavailable) always show which layer you are looking at.

Canonical source rules

A contract is labeled Canonical Source only when its address appears in official Robinhood Chain documentation. The lowercase contract address is the primary identity everywhere.

Official metadata (name, category, canonical status) always wins over explorer data; the explorer fills in activity numbers (holders, transfers, verification, first/last seen).

Community tokens are never matched to canonical assets by ticker — a community token named AAPL does not inherit anything from the canonical AAPL stock token.

Deduplication

Every hourly run merges duplicate project pages that point at the same contract address. Claimed pages are never deleted automatically — conflicts involving a claimed page go to manual review.

Index Score

Index Score measures how complete the information on this page is. It is NOT a safety score, an endorsement, or investment advice.

The score (0–100) adds points for available information: name, description, website, logo, linked contracts, verified source code, canonical provenance, extra links, a linked repository, and a completed ownership claim. It never measures safety, legitimacy, or quality.

Missing Metadata

A token is labeled Missing Metadata when its name or symbol is empty, whitespace-only, or contains no letters or digits (for example an emoji-only symbol). Such tokens stay in the Raw Index and the Missing Metadata view but are excluded from featured sections and activity rankings.

Live metrics

Live metrics come from the Blockscout API with a 60-second server-side cache, request coalescing, and a cooldown after provider failures. When the provider is slow or down, RobinIndex shows the last known value labeled Stale, or Unavailable — never a fake zero.

Pulse calculation

Pulse compares each entity's current state with its last known baseline once per hourly run. Deltas are only computed when both the old and new values exist — a first observation never becomes a fake change event. Sudden implausible drops are flagged as possible source resets and excluded from rankings. Summaries for 1h/24h/7d are precomputed after each run.

Logo source priority

1) Curated issuer logos for canonical contracts (Wikimedia Commons, verified and attributed, keyed by official contract address). 2) A logo provided by a claimed project. 3) Blockscout token metadata for that exact address. 4) A deterministic monogram fallback. Logos are never AI-generated, never recolored, and never assigned by ticker.

Claim & review status

Anyone can claim a project page without a wallet. Ownership can be proven with a domain challenge (file, DNS TXT, or meta tag) or a GitHub repository challenge; challenges are random, single-use, and expire after 48 hours. A moderator approves every claim — verification speeds review up but never bypasses it, and canonical pages always require manual approval.

Reviewed means a moderator looked over the page. Neither status is an endorsement.

Limitations

Holder counts may not represent unique users. Transfer counts are not financial volume. Explorer data can lag or reset. Verification of source code says nothing about intent. RobinIndex indexes public data as-is and may be incomplete or outdated at any moment.