Data sources & methodology
Schmatz combines public filings and economic series with historical market data, then computes charts, classifications, backtests, and research narratives. Source availability, licensing terms, coverage, and freshness differ by dataset; a source being publicly accessible does not automatically make every record public domain.
Principal source categories
- SEC EDGAR — company identifiers, filings such as 8-K, 10-K, 10-Q, Form 4, and Schedule 13D, and XBRL company facts used to compute selected fundamentals. Filing content is linked or attributed to the SEC where practical.
- Federal Reserve Economic Data (FRED) — selected interest-rate, volatility, currency, and other economic or market series used as research context and model regime inputs. Individual FRED series may carry source-specific notes or restrictions.
- Historical market-data services — daily price, volume, corporate-action, and related records accessed through public endpoints or software libraries. Coverage and adjustment quality vary, and the upstream provider’s terms still apply.
- Public news and filing metadata — headlines, links, timestamps, and government filing metadata used to connect events with price moves. Schmatz does not republish full third-party articles.
The operator may compare customer-facing results with separately licensed research datasets for quality assurance. Records from a restricted dataset are not intentionally exposed through customer APIs.
How data reaches the product
- Scheduled ingestion jobs request new or corrected records and write them to local tables. The end-of-day price refresh uses a rolling lookback so recent corrections can replace cached rows.
- Validation checks normalize symbols and dates, reject malformed rows, compare priority coverage, and track freshness. Missing or stale data can still occur and is surfaced through freshness or coverage states in the app.
- Research code computes returns, normalized performance, event classifications, similarity results, portfolio simulations, and summary statistics from the stored corpus.
- The browser receives the subset required for a requested view. A stock chart may receive dated close or OHLC points; a backtest receives its equity curve, summary statistics, assumptions, and selected trade details. Schmatz does not expose unrestricted table access.
What the app displays
- End-of-day price and normalized-performance charts
- Computed returns, drawdowns, risk statistics, and comparison series
- Backtest equity curves, assumptions, validation state, and simulated trades
- Classified events, similarity matches, and model outputs
- Selected fundamentals computed from SEC XBRL facts
- Headlines, links, filing references, and AI-assisted editorial synthesis
Source labels describe provenance at the level the interface can support. A citation or source badge does not guarantee that the upstream record is error-free, complete, or fit for trading.
What the app does not provide
- A bulk downloadable archive of historical OHLCV records
- A general-purpose SQL or raw-database endpoint
- Tick-by-tick or guaranteed real-time market data
- A warranty that every listed security has complete five-year coverage
- A substitute for a licensed market-data feed or official filing record
Freshness, corrections, and coverage
Most customer-facing price views are end-of-day, not live quotes. Market holidays, upstream outages, symbol changes, delistings, rate limits, and failed scheduled jobs can delay or interrupt a refresh. The app should be read together with its displayed “as of,” freshness, and coverage notices. Cached values may be corrected after an upstream source revises them.
Backtests and AI-assisted text
Backtests are simulations over the available historical corpus and inherit its missing data, survivorship, timestamp, adjustment, and execution-model limitations. AI systems may compose prose from a selected package of facts and tool results, but model-generated text can still misstate or omit context. Computed numbers remain the responsibility of deterministic code; users should verify important facts against the cited filing or an independent data source.
Questions or corrections
If a value appears stale, incorrectly adjusted, misattributed, or inconsistent with its cited source, send the ticker, URL, displayed date, and expected value through the contact form. Licensing or redistribution concerns can use the Legal category.