This works really well until you're dealing with potato quality video, weird fonts, and shirts plastered with ads, or even pretty good quality video and white shirts with white lettering (https://www.arsenal.com/news/how-volunteer-part-no-more-red).
> Does your team and department taking of analytics primarily on the footballing side? Like player performance?
Yes, we work primarily on the footballing side and across the spectrum in that space: Player/team performance for the men's first team, women's first team, and boys academy age groups U16 and up, player recruitment / squad planning, etc.
> How does your teams work typically get incorporated
We produce a mix of interactive tools, regular static reports (e.g. opposition analysis, post-match analysis, etc.), and live dashboards that come from specific stakeholder requests such as coaching staff or execs, or that we build proactively to address a specific football-related question.
> what does the day to day look like?
It really varies from day to day and role to role within the team. A data engineer might be adding another data provider to an entity resolution ETL pipeline, a research scientist might be incorporating feedback from first team coaching staff into a work-in-progress model, a data analyst might be putting together an in-depth opposition analysis report for an upcoming match, and an operations analyst might be helping train operators on a new data labeling task.
> Do you manage your own tech stack as well?
We do manage most of our tech stack, although we get a lot of support on front-end from a great sister team in the IT dept.
Hmm, well the super secret stuff we’re working on comes directly to mind, but if I set that aside, boring entity resolution is actually a big pain point.
Regardless of their sophistication, 3rd party data products in football tend to rely on manually collected and maintained player metadata. It can be unreliable. If I could reliably have a durable unique ID for every player, manager, and team in world football along with reliable timestamps for every moment each said player entered and left play, that would be pretty great. When joining together disparate data sources, discrepancies in things this simple cause all sorts of pain downstream.
There definitely is hope for AI aiding with the less subjective elements of refereeing, such as offside, ball out of bounds, keeper moving off of line before a penalty is taken, etc.
It has already had a big impact for ball crossing the goal line judgments if you put computer vision in the broader AI category (e.g. Hawkeye in the PL)
Aside from 3rd party data, we contract with a nonprofit in Laos where operators collect event data for us (~2000 data points per match) across roughly a dozen leagues worldwide.