News API
originArticle revisions from a third-party market-data provider. Original source URLs and availability timestamps are retained.
Licensed market-data feedsynced 4m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Keep source revisions and market timestamps
SnowflakeNews and daily bars land separately in Snowflake; preserve when each fact became available.
Distinguish missing data from a quiet market
HumanTrack API cursors and trading sessions so a failed fetch never becomes a zero-volume observation.
1 to 2 of 2
Daily bars API
originDaily price and volume observations, keyed by ticker and trading session.
Licensed market-data feedsynced 4m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Keep source revisions and market timestamps
SnowflakeNews and daily bars land separately in Snowflake; preserve when each fact became available.
Distinguish missing data from a quiet market
HumanTrack API cursors and trading sessions so a failed fetch never becomes a zero-volume observation.
1 to 2 of 2
news
declaredAppend-only Snowflake landing table. Each provider revision remains available for audit.
RAW.MARKETloaded 4m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Keep source revisions and market timestamps
SnowflakeNews and daily bars land separately in Snowflake; preserve when each fact became available.
Distinguish missing data from a quiet market
HumanTrack API cursors and trading sessions so a failed fetch never becomes a zero-volume observation.
1 to 2 of 2
daily_bars
declaredSnowflake landing table for split-adjusted OHLCV data and source timestamps.
RAW.MARKETloaded 4m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Keep source revisions and market timestamps
SnowflakeNews and daily bars land separately in Snowflake; preserve when each fact became available.
Use the approved research universe
Google DocsThe Google Docs research brief defines ticker coverage, corporate actions and reporting policy.
Distinguish missing data from a quiet market
HumanTrack API cursors and trading sessions so a failed fetch never becomes a zero-volume observation.
1 to 3 of 3
stg_news
parseddbt staging model that normalizes news revisions, resolves security IDs and preserves publication time.
ANALYTICS.STAGINGbuilt 3m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
dbt joins must respect the research cutoff
HumanClean article revisions, validate daily bars and aggregate news without future information.
1 to 1 of 1
stg_daily_bars
parsedValidated daily bars with stable security IDs. Adjustment conventions follow the research playbook.
ANALYTICS.STAGINGbuilt 3m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Use the approved research universe
Google DocsThe Google Docs research brief defines ticker coverage, corporate actions and reporting policy.
dbt joins must respect the research cutoff
HumanClean article revisions, validate daily bars and aggregate news without future information.
1 to 2 of 2
int_news_by_ticker
parsedEligible news grouped by security and as-of cutoff before semantic extraction. Future revisions are excluded.
ANALYTICS.INTERMEDIATEbuilt 2m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
dbt joins must respect the research cutoff
HumanClean article revisions, validate daily bars and aggregate news without future information.
1 to 1 of 1
int_price_features
parsedLagged price returns and volume changes, computed only from sessions available at the forecast cutoff.
ANALYTICS.INTERMEDIATEbuilt 2m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
dbt joins must respect the research cutoff
HumanClean article revisions, validate daily bars and aggregate news without future information.
Evaluate trend forecasts out of time
AgentNumerical price and volume features feed a separate ML model with walk-forward validation.
1 to 2 of 2
news_signals
declaredLLM-extracted company sentiment and topics, with model confidence and a review state. Not a stock-price forecast.
Python · enrich_news.pyran 2m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Separate news sentiment from confidence
AgentPython extracts company signals with an LLM; confidence is not a probability of stock movement.
1 to 1 of 1
trend_forecasts
declaredFive-session return estimates from lagged price and volume features. Model evaluation uses walk-forward splits.
Python · predict_trends.pyran 2m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Evaluate trend forecasts out of time
AgentNumerical price and volume features feed a separate ML model with walk-forward validation.
1 to 1 of 1
fct_stock_research
parsedResearch mart joining semantic signals and numerical forecasts at security and as-of grain. Provenance remains separate for each model.
ANALYTICS.MARTSbuilt 1m ago1 to 4 of 4
Impact
Choose a change to see its impact.
Knowledge items
Use the approved research universe
Google DocsThe Google Docs research brief defines ticker coverage, corporate actions and reporting policy.
dbt joins must respect the research cutoff
HumanClean article revisions, validate daily bars and aggregate news without future information.
Separate news sentiment from confidence
AgentPython extracts company signals with an LLM; confidence is not a probability of stock movement.
Evaluate trend forecasts out of time
AgentNumerical price and volume features feed a separate ML model with walk-forward validation.
Publish two traceable research views
Google DocsBigQuery serves semantic signals and trend forecasts to BI, with separate review states.
1 to 5 of 5
news_signals
declaredBigQuery semantic-signal serving table for analyst BI dashboards. Review status and source provenance accompany the labels.
BigQuery · researchpublished 1m ago1 to 3 of 3
Impact
Choose a change to see its impact.
Knowledge items
Separate news sentiment from confidence
AgentPython extracts company signals with an LLM; confidence is not a probability of stock movement.
Publish two traceable research views
Google DocsBigQuery serves semantic signals and trend forecasts to BI, with separate review states.
1 to 2 of 2
trend_forecasts
declaredBigQuery numerical-forecast serving table for BI. Forecast horizon, model version and training cutoff remain visible to analysts.
BigQuery · researchpublished 1m ago1 to 3 of 3
Impact
Choose a change to see its impact.
Knowledge items
Use the approved research universe
Google DocsThe Google Docs research brief defines ticker coverage, corporate actions and reporting policy.
Evaluate trend forecasts out of time
AgentNumerical price and volume features feed a separate ML model with walk-forward validation.
Publish two traceable research views
Google DocsBigQuery serves semantic signals and trend forecasts to BI, with separate review states.
1 to 3 of 3