Research foundation · Phase 0–1

Public‑data research layer

First confirm where the data come from, what timestamp they represent, and whether they have been revised before discussing formulas and results. This page is a source and data‑contract catalogue, not a real‑time quotes dashboard.

Catalog status Loading… Awaiting version info

Data contract

What does this layer protect against?

Source traceable

Each result must be traceable to the official URL, snapshot hash, parser version and licensing notes.

No time ambiguity

as-of、observation、filed、release、retrieved-at 與 vintage 分開保存。

Do not fill failures with placeholder values

When a source is overdue, rejected, or its format changed, display stale／error; do not pretend new data by using zeros or the last result.

Source registry

Registered data sources

Phase 0: first log capabilities and boundaries; not enabled does not mean data are available.

Reading source registry…

Dataset catalog

Check datasets and quality status

Currently the catalog only describes datasets; catalog metadata is not treated as observation values.

Reading dataset catalog…

Research contract

Minimum data provenance for research queries

source_idSource registration ID

snapshot_idOriginal response SHA‑256

as_ofTimestamp of data validity

retrieved_atTime to system acquisition

frequencyFrequencies such as daily/weekly/monthly

adjustmentadjusted／as-reported 等口徑

quality_statusfresh／partial／stale／invalid／unavailable

license_refSource terms‑of‑use index

To obtain data values the next‑stage server‑side adapter must pass schema, time, unit, completeness and authorization checks; this static page will not place FRED API keys, SEC credentials or other secrets in the browser.

Continue learning

From data contracts back to quantitative research

Quant research workflow5MA offset, volatility Z‑Score, out‑of‑sample backtest and research checklist. Z-Score Taiwan index futures / 富台期 pairingContract unit, Basis, term structure, rollover and shutdown conditions. Volatility Z‑Score tool Z-ScoreUse verified series for local formula work; when sources are unavailable do not impute fake values.

Phase 2 · Features & backtest

Features and backtest research workspace

Perform feature engineering and held‑out benchmark backtests with dated, traceable series; does not provide trade signals.

Open quant features and backtesting workbenchView volatility characteristics, missing‑count, next‑bar execution definition, trading costs and walk‑forward windows.Read the research methodology first.Understand data scope, out‑of‑sample splits and how to avoid future information leakage.