Fortune
This site documents the fortune dbt project — the analytics layer of the neo-fortune platform that turns raw Bet365 odds and event feeds into match-level betting strategy outputs on ClickHouse.
Layers
Section titled “Layers”The pipeline is organised in five layers, each with its own section in the sidebar:
| Layer | What it covers |
|---|---|
| Raw | Source tables loaded by the Scrapy crawler — the untyped JSON landing zone. |
| Sources | First-cast, typed projections of the raw JSON into well-defined columns. |
| Calculations | Per-event derivations: pre-match and in-play odds aggregates, scores, team strength, expected goals, probabilities, and betting strategy metrics. |
| Mart | Wide, query-ready tables that join calculations into a single row per match (or per match-minute). |
| Elementary | Test and run-result views built on top of the Elementary dbt package, exposed by event_date for trend analysis. |
Per-model pages
Section titled “Per-model pages”Every model in the dbt project has its own page containing:
- A Mermaid lineage diagram showing the immediate parents (upstream refs/sources) and children (downstream refs).
- A columns table listing every column with its description and the formula or expression that produces it.
Use the sidebar to navigate; pages are grouped by the layer the model lives in.