This Data Sources Statement explains the public and third-party data sources used by EventFlow Data, how data is processed, and important limitations.
EventFlow Data is a read-only analytics and research infrastructure service. It does not provide trading, execution, order routing, account operations, custody, wallet, or investment-advice functionality.
1. Source Categories
1.1 Prediction-Market Data
Sources include public Polymarket market-data, Gamma, Data API, CLOB public read endpoints, and WebSocket feeds where available.
Typical fields include events, markets, market questions, resolution text or references, tags, categories, condition IDs, token IDs, prices, spreads, order-book levels, trades, trade timestamps, trade size, side, transaction hashes, public wallet activity, public wallet positions, market holder distributions, and leaderboard or aggregate public-wallet data where available.
We use public read endpoints and market-data feeds. EventFlow Data does not use customer private keys or customer venue credentials and does not submit orders.
1.2 Public Blockchain Data
Sources include public Polygon blockchain records and related RPC or block-explorer data.
Typical fields include block numbers, block timestamps, transaction hashes, wallet addresses, token transfers, ERC-1155/CTF activity, split, merge, redemption, settlement, approval, and contract-event logs.
Blockchain records are public, replicated, and generally immutable. Our indexes, joins, and derived views may lag, contain parsing errors, or be corrected over time.
1.3 Crypto Reference Data
Sources include Deribit public REST endpoints and other public crypto-reference data sources if disclosed in product documentation.
Typical fields include instruments, index prices, book summaries, mark prices, implied-volatility fields, expiry, strike, option type, and other public reference data for crypto-linked markets.
1.4 Public News and RSS/Atom Feeds
Sources include configured public RSS and Atom feeds. The current RSS connector uses conditional GET headers where possible, including ETag and Last-Modified, to reduce redundant fetching.
Typical fields include title, body or summary, URL, source reference, author or publisher metadata where present, published time, fetched time, and feed cursor metadata.
1.5 Telegram Public Channel Posts
Sources may include Telegram public channel posts available to an authorized Telegram Bot API integration. The connector is credential-gated and unavailable without a bot token. The bot must have access to the relevant channel updates.
Typical fields include channel reference, message ID, post text, channel title, post timestamp, URL, update cursor, and fetch status.
We do not ingest private Telegram chats through the public Service.
1.6 X Public Posts
Sources may include public X posts retrieved through the X API v2 recent-search endpoint. The connector is credential-gated and unavailable without a bearer token.
Typical fields include post ID, text, author ID, creation timestamp, URL, source query, and fetch status.
1.7 Customer-Configured Views
Authenticated users may configure watchlists, saved views, selected markets, or integration preferences. These are customer settings, not external market sources.
2. Derived Data and Analytics
We normalize, deduplicate, join, sequence, snapshot, and replay source data to create derived analytics, including:
- L2 order-book snapshots and replay history.
- Price history and candles.
- Trade-time BBO and inferred aggressor-side context.
- Public-wallet activity ledgers and cash-flow PnL reconstruction.
- Wallet toxicity scores based on counterparty markout and market-wide drift.
- Wallet skill scores based on statistical testing against luck baselines.
- Behavior flags such as fresh-whale, bot, farmer, concentration, and activity-pattern indicators.
- Markout, realized spread, microprice, imbalance, cumulative volume delta, large-trade, absorption, and fill-intensity metrics.
- News classifications, settlement-clause summaries, relevance, materiality, confidence, fair-probability research outputs, and latency-aware signal evaluation.
- Data freshness, latency, gap, reconciliation, confidence, and quality flags.
Derived outputs are research analytics. They are not trading instructions, investment advice, proof of intent, proof of identity, or guarantees of future results.
3. Update Behavior
Update behavior varies by source:
- WebSocket market-data streams may update near real time when source connections are available.
- Public REST endpoints may be polled on configured schedules and are subject to rate limits, pagination, and provider availability.
- RSS/Atom feeds may be polled with conditional GET and may return "not modified" responses.
- Telegram and X connectors run only when valid credentials and source configuration are present.
- Blockchain data may be delayed by block production, confirmations, RPC lag, provider indexing, and reorg handling.
- Deribit and other reference sources may be updated on scheduled polling intervals.
No update interval is guaranteed unless expressly stated in a signed agreement.
4. Data Quality Controls
We use controls designed to make source status visible, including:
- Source timestamps, ingest timestamps, and latency metrics.
- Deduplication by source identifiers, transaction hashes, or event keys where available.
- Sequence numbers and replayable capture where implemented.
- Raw plus structured retention where implemented.
- Gap logs, freshness indicators, and stale-data flags.
- Reconciliation gates for wallet cash-flow PnL and public-wallet activity.
- Confidence, coverage, and quality fields on derived metrics where applicable.
- Error classification and fail-soft behavior so one failing source does not necessarily stop the entire pipeline.
Quality controls reduce risk but do not eliminate errors.
5. Known Limitations
Public and third-party data may be incomplete, delayed, unavailable, rate limited, corrected, deleted, edited, inconsistent, or subject to schema changes.
Specific limitations include:
- Public wallet addresses are pseudonymous. We do not claim to know the real-world identity behind a wallet unless that identity is lawfully public and expressly shown from a source.
- Wallet PnL, skill, toxicity, and behavior flags depend on captured coverage, public source availability, cost-basis assumptions, and reconciliation checks.
- Social and news content may be deleted, edited, rate limited, unavailable, inaccurate, or taken out of context.
- LLM-generated outputs may be wrong, incomplete, or sensitive to prompt and source coverage.
- Absorption or iceberg indicators are candidates based on public evidence, not proof of hidden orders.
- Market prices, depth, and public positions can change quickly and may be stale by the time they are viewed.
- Blockchain records are public and immutable, but our indexing, labeling, and derived interpretations may change.
- Third-party source terms may restrict downstream use. Users are responsible for their own compliance.
6. No Trading or Advice
Data Sources and derived analytics are provided for research and informational use only. They are not recommendations to buy, sell, hold, hedge, copy, follow, or trade. EventFlow Data does not execute, route, place, cancel, custody, settle, or automate trades for users.
7. Responsible Use
Users must not use source data or derived analytics to:
- Harass, threaten, dox, identify, or target individuals.
- Make legally significant decisions about people, including employment, credit, housing, insurance, education, healthcare, law enforcement, or eligibility decisions.
- Violate privacy, market-abuse, sanctions, financial, gambling, platform, or other applicable laws.
- Misrepresent source data, omit quality flags, or present probabilistic research outputs as facts or advice.
- Violate third-party source terms or API restrictions.
8. Source Attribution and Changes
We may add, remove, replace, rename, or modify sources, fields, metrics, connectors, polling behavior, retention, and quality rules as sources and the Service evolve. Product documentation or data dictionaries may provide more specific field-level details.
9. Contact
Questions about data sources, field definitions, or quality flags: contact@yield3.app
Privacy questions involving public identifiers: contact@yield3.app