02
Who experiences it
- Championship football clubs
- Smaller Premier League clubs
- Clubs with analysts, paid data and extensive video but no dedicated ML engineering team
Loading evidence file
Loading evidence file
Give analysts one place to search every match, inspect the supporting footage and create a coach-ready opposition report.
The operational problem
Clubs own more event data and video than an analyst can reasonably inspect before the next fixture. Finding one precise tactical pattern means moving between disconnected tools, manually cutting clips and rebuilding the same report structure every week.
Who experiences it
Current workflow
Translate a coaching question into multiple event-data filters
Search matches one at a time and locate the matching video manually
Review clips, reject false positives and annotate useful moments
Copy evidence into a presentation and rebuild the narrative
Rarely revisit the forecast after the match to improve the method
Interactive product demonstration
This is a front-end product thesis built with synthetic local data. Interact with the controls to inspect the proposed workflow; it does not represent a deployed customer system.
7 of 9 qualifying sequences across 6 matches. Most often through a third-player release or the No. 6 receiving behind the first line.
Sample: n=9 · All formations · analyst review requiredWhat the proposed system does
Interprets a natural-language tactical question as visible, editable search logic
Links every result to the exact supporting video moment
Reports sample size and confidence instead of hiding uncertainty
Keeps analyst approval in the loop before evidence reaches a coach
Turns approved tendencies and clips into an editable opposition report
Compares pre-match expectations with what occurred after the fixture
Required customer data
Data availability, rights and quality are assessed before a live pilot is scoped.
Expected measurable outcomes
Target outcome: reduce time spent locating, clipping and assembling evidence.
Pilot hypothesis: maintain a measurable acceptance threshold for surfaced moments.
Measure how many report claims remain linked to inspectable match evidence.
These are target outcomes and pilot hypotheses. Baselines, thresholds and measurement methods are agreed with the design partner before work begins.
08 / 30 / 60 / 90
The scope increases only when the previous stage establishes that the data, workflow and economics support it.
01 / evidence gate
Ingest 10 historical matches, configure 10 tactical concepts and recreate one historical opposition report.
02 / evidence gate
Ingest 20–30 matches, synchronize video and event data, add tactical search and automatic clips, then run four weeks in shadow mode.
03 / evidence gate
Deploy into the live analysis workflow with permissions, feedback, reporting and data integrations.
Typical price range
CAD $40,000–$65,000
Target pilot range. Final scope depends on data licences, video synchronization and integration requirements.
Prices shown in Canadian dollars, excluding applicable taxes and third-party data or licensing costs.Risks and limitations
11 / Design-partner intake
We are looking for operators who can test the product thesis against real constraints and help define what an operational version must do.
Discuss this project