WA Waltzing Analytics Football Data Solutions
Services / Data Infrastructure

Data infrastructure your analysts can build on

Pipelines, databases and tools that turn raw feeds from Opta, StatsBomb, IMPECT, Wyscout and SkillCorner into one clean, connected source for every model, report and dashboard at the club.

Discuss a project All services
What you get

Deliverables

Data architecture design

A blueprint for your club's data: sources, storage, a shared data model for players, matches, events and tracking, and who uses what.

Provider integration

Automated ingestion from provider APIs and files, including IMPECT, StatsBomb, SkillCorner, Wyscout and Opta, with validation on every load.

Cross-provider ID matching

One player, one ID: players, teams and matches matched across providers so event, tracking and market data join cleanly.

Automated pipelines

Scheduled, competition-agnostic pipelines that rebuild metrics and model outputs after every matchday without manual steps.

Analysis database

A fast, queryable database, in DuckDB or your existing warehouse, that analysts can use directly from SQL or Python.

Dashboards and internal apps

Web dashboards and apps for staff: player search, team benchmarks and report viewers, in your house style.

Automated reporting

Match, player and season reports generated straight from the pipeline as PDFs or web pages.

Natural-language data queries

A local language model that answers questions about your data in plain language by writing SQL, running entirely on club hardware.

Documentation and handover

Code, documentation and training so your own team owns, runs and extends the system.

How it works

One pipeline from raw feed to decision

  1. 01

    Sources

    Opta, StatsBomb, IMPECT, Wyscout, SkillCorner, Transfermarkt

  2. 02

    Ingest and validate

    APIs, schema checks and cross-provider ID matching

  3. 03

    Club data model

    Players, matches, events and tracking in one database

  4. 04

    Models and metrics

    xG, delivery value, goalkeeper model, recruitment ratings

  5. 05

    Delivery

    Reports, dashboards, apps and exports

Every service on this site runs on the same architecture. Raw feeds are parsed and validated once, stored in one data model and reused by every model and report, so numbers agree across departments.

Systems are built to be handed over: plain Python and SQL, open file formats such as Parquet, version control and documentation.

Events and passes loaded per Eredivisie season
Four seasons, 1,239 matches and 1.88 million events, re-parsed every time the models are retrained.
Five-step model pipeline diagram
A model pipeline from raw event state to action value, one of the pipelines this architecture runs.
Technology

Built with

  • Python
  • SQL
  • DuckDB
  • Parquet
  • REST APIs
  • Git
  • Streamlit
  • Local LLMs

Related work

Tired of data living in ten different places?

Tell me which data your club licenses and who needs it. I will propose an architecture that fits your staff and budget.

Discuss a project Email directly