How a club can own its hockey data
A reference architecture where the club owns the data and models — and any AI is just a swappable consumer.
The data and models live in infrastructure the club controls — it never leaves your hands.
Because the gold layer is the contract, you can swap AI providers anytime — the data layer never changes.
The 30-second version
- Sources feed ingestion; ingestion feeds a data lake and models the club owns.
- Dashboards, your own apps and any AI read from the same governed gold layer.
- Because you own the data layer, you can swap AI providers without re-plumbing anything.
Why ownership is the whole point
Plenty of vendors will happily hold your club's data for you — and quietly make you dependent on their tool. The alternative is simple and durable: the club owns the **data lake and the models**, and everything else (dashboards, internal apps, AI assistants) is just a consumer that reads from it. Own the layer that's expensive to rebuild, and rent the layers that are easy to swap.
Sources → ingestion
League APIs, public scrapers and your own tracking or scouting data flow in through small, scheduled ingestion jobs. The rule at this stage is to keep an **immutable raw copy** in infrastructure you control. If a model is wrong next month, you reprocess from raw instead of re-scraping a season.
The layer the club owns
Inside the ownership boundary sits the club's private **data lake** — raw (bronze), cleaned (silver) and analysis-ready (gold) — plus the **models** that turn events into [expected goals](/glossary/expected-goals), [possession](/glossary/possession) and scouting profiles. This is the part that compounds in value, so it lives where the club controls it and never leaves.
Serve it anywhere — and swap AI freely
Dashboards, your own internal apps, and any AI provider (ChatGPT, Claude, Copilot) all read the **same governed gold tables** with read-only access. Because the gold layer is the contract, switching AI providers — or adding a new dashboard — changes nothing about the data underneath. You are never locked in.
What this buys you
- **Continuity:** the data and models survive any vendor change. - **Flexibility:** adopt the best AI of the moment without a migration. - **Trust:** sensitive data stays in infrastructure the club controls.
Build it yourself
- 1Decide the boundary first: which layers must the club own? (Data lake + models.)
- 2Put ingestion on a schedule and keep an immutable raw copy.
- 3Model raw events into governed gold tables — your contract for everything downstream.
- 4Give dashboards and AI read-only access to gold; never let a consumer own the data.
Key terms
Newsletter
Weekly hockey analytics
Data-driven takes on performance, scouting, and team strategy. No fluff — just the numbers that matter.
No spam. Unsubscribe anytime.
Related explainers
A reference architecture for hockey analytics
A full, opinionated blueprint tying data, models, AI and delivery into one system.
How a hockey analytics data stack works
The end-to-end path from raw game data to an answer you can act on — drawn as one diagram.
Using Claude to analyze your hockey data
How an LLM turns your data stack into a question-and-answer machine — safely.