Data Quality
Coverage, refresh timing, and reliability notes for TickerDB
TickerDB is most useful when the state labels are dependable. This page explains what is covered, when data updates, and what integrity safeguards sit behind the API.
Coverage
TickerDB covers US stocks, ETFs, and supported crypto pairs. Public responses expose asset_class so clients can distinguish stocks from ETFs and crypto, and stock-only fundamental sections are omitted for non-equities.
Freshness
TickerDB is end-of-day only. Responses expose data_status for the freshness mode and as_of_date for the specific session date represented by the snapshot.
TickerDB is end-of-day only. Stocks and ETFs refresh by ~00:30 America/New_York (Eastern time, EST/EDT). Crypto refreshes by ~00:10 UTC, including weekends.
Corporate Actions
Historical classifications are computed on adjusted market data so stock splits and similar corporate actions do not create false technical regime shifts. If a response is available for a given historical date, it is meant to reflect the post-adjustment view of that session.
Universe Quality
Only active tradable assets are published into the current API surface. Unsupported or inactive symbols are excluded from watchlist additions and summary lookups.
Historical weekly stock snapshots can retain inactive, acquired, delisted, or bankrupt symbols so a point-in-time model universe is not reconstructed from today's survivors. Those historical records do not activate a symbol in current daily pipelines or make it eligible for watchlist additions.
Historical Market-Cap Evidence
Market-cap state is explicit. large, mega, and ultra_mega contain an exact point-in-time value and satisfy the Value Divergence Model's large-cap gate. large_or_larger_proven also satisfies the gate when threshold evidence is conclusive but an exact cap is unavailable. nano, micro, small, mid, and below_large_proven are excluded as below the gate.
unknown means the server cannot establish which side of the historical threshold the stock belonged on. It is not treated as small. The row is individually excluded with value_divergence_eligible=false, null score and band, and value_divergence_exclusion_reason=market_cap_unknown.
When the tier is large_or_larger_proven, below_large_proven, or unknown, the API returns market_cap=null. It never fills that field from a stale fallback value; use market_cap_tier as the authoritative threshold evidence.
Weekly Publication Integrity
Explicit weekly-date search responses include weekly_integrity. Its row counts distinguish unknown, proven-large, and proven-below-large observations. publication_complete is fail-closed: it is false when the publication marker is missing or stale, or when row-level eligibility contracts do not match their market-cap evidence.
An unknown stock is omitted individually and does not invalidate an otherwise verified week. The whole week becomes risk-off only when universe_incomplete=true. In that case every stock is ineligible and no stock has a score or band.
Classification Integrity
TickerDB stages publish candidates before promoting them into the serving tables. Snapshot rows, support and resistance rows, sector aggregates, and band transitions are validated during that publish flow so incomplete runs do not silently replace the active dataset.
How To Use This Safely
Use as_of_date and the band stability metadata together. The date tells you which session is being described, and the stability metadata helps you judge whether a label is brand new or well established.