Query the warehouse¶

Analyst

Use this page when you want to query OddsFox Pipeline data, not operate it. OddsFox Pipeline ships code and local warehouse tooling, not a hosted dataset. Analysts query the DuckDB file produced by a local or self-managed run. For a linear first session, see First query. For the full analyst map, start with Analysts. Term shortcuts live in Analyst shortcuts.

Reference ladder

Chooser → dictionary → public contracts → warehouse reference; do not treat staging/raw as APIs. This page is the chooser and trust guide.

Shortest Path¶

I already have a warehouseI need a warehouse first

Open oddsfox.duckdb (or the file in DUCKDB_PATH). See Configuration for path precedence.

Follow Quickstart, or ask an operator to run a scope. See Analysts for the full analyst map.

Query Rules¶

Query *_marts first. These are the supported public query surfaces.
Use *_observability when checking freshness, coverage, run health, or data quality findings.
Treat *_raw, *_ops, *_staging, and *_intermediate schemas as internal or debugging surfaces.
Prefer fully qualified table names, such as polymarket_wc2026_marts.polymarket_wc2026_market_hourly_odds.
Filter by event_slug, event_id, or market status fields when narrowing WC2026 hourly analysis. Closed and resolved markets remain in the mart.

Historical closed and resolved rows are intentionally retained. Do not assume a row is live because it appears in a mart.

Open With Python¶

import duckdb

con = duckdb.connect("oddsfox.duckdb", read_only=True)
rows = con.sql("""
    select
        event_slug,
        question,
        odds_hour_utc,
        close_odds,
        event_volume_usd_lifetime_reported
    from polymarket_wc2026_marts.polymarket_wc2026_market_hourly_odds
    where is_active
      and not is_closed
    order by event_slug, question, odds_hour_epoch desc
""").df()

Use read_only=True for notebooks and analysis so you do not compete with a running Dagster/dbt writer.

Which Table Should I Use?¶

Goal	Start Here	Notes
WC2026 Polymarket hourly odds	`polymarket_wc2026_marts.polymarket_wc2026_market_hourly_odds`	One row per `market_id`, `odds_hour_epoch`; primary-outcome CLOB prices (`primary_outcome_label`) with market and event metadata.
WC2026 in-game match minutes	`polymarket_wc2026_marts.polymarket_wc2026_match_minute_odds`	Dense minute series for all 104 matches; requires the match-minute path, not ordinary hourly ingest alone.
Spain–Argentina final historical L2 depth	`polymarket_wc2026_marts.polymarket_wc2026_match_order_book`	Long-form independent bid/ask levels for both PMXT outcome-token snapshot streams; requires the unscheduled PMXT backfill.
WC2026 fixtures and results	`oddsfox_reference.international_results_wc2026_matches`	One row per `match_id`, loaded from the active Scraper reference bundle.
WC2026 team status	`oddsfox_reference.international_results_wc2026_team_status`	Join on `canonical_team_name` or `team_name`; same reference-bundle prerequisite as fixtures/results.
Current Kalshi stage prices	`kalshi_wc2026_marts.kalshi_wc2026_stage_markets`	Filter to `is_actionable_live_market`.
Kalshi stage hourly series	`kalshi_wc2026_marts.kalshi_wc2026_stage_market_hourly_odds`	Use `progression_*_price` for stage progression semantics.
Current Kalshi group-winner prices	`kalshi_wc2026_marts.kalshi_wc2026_group_winner_markets`	Use `group_winner_price`.
Kalshi group-winner hourly series	`kalshi_wc2026_marts.kalshi_wc2026_group_winner_market_hourly_odds`	One row per `market_ticker`, `odds_hour_epoch`.
WC2026 finalized Polygon settlement minutes (advanced)	`polymarket_wc2026_marts.polymarket_wc2026_polygon_settlement_minute_odds`	Fixed 150/210-minute scheduled windows; empty sides remain null; fill counts are normalized economic legs.

Trust Before Analysis¶

For hourly odds:

Confirm the market's enclosing event is volume-eligible (event_volume_usd_lifetime_reported >= 100000).
Inspect market status fields (is_active, is_closed, is_resolved) when you need current vs historical rows.
If coverage looks sparse, inspect polymarket_wc2026_observability.polymarket_wc2026_ingestion_run_observability.
Keep historical rows when you explicitly want closed or resolved markets.

Useful observability tables:

Source	Table	Use
Polymarket WC2026	`polymarket_wc2026_observability.polymarket_wc2026_ingestion_run_observability`	Ingestion run telemetry and request counts.
Polygon settlement WC2026	`polymarket_wc2026_observability.polymarket_wc2026_polygon_settlement_data_quality`	Published scan/seed match, finalized chunk coverage, exact dense inventory, and hard publication state.
Polygon settlement WC2026	`polymarket_wc2026_observability.polymarket_wc2026_polygon_settlement_quality_issues`	Sparse/no-fill, derived-leg, pair-deviation, secondary-RPC, and structural findings.
Kalshi WC2026	`kalshi_wc2026_observability.kalshi_wc2026_data_quality`	Stage/group-winner stale or missing live odds and coverage findings.
Kalshi WC2026	`kalshi_wc2026_observability.kalshi_wc2026_ingestion_run_observability`	Kalshi ingestion telemetry.

Next: use Query recipes for examples, then the Data dictionary for table-by-table semantics.