How to hire a data scientist in sport
Hiring a sports data scientist is harder than most equivalent searches outside sport. The candidate pool is small, the bidding wars are real, and many of the people worth hiring are heads-down on something they care about and not actively looking.
This guide is for the hiring exec, director of football, head of operations or CTO running this search - typically for the first or second time. It covers the things we wish every brief told us up front.
Get the role spec right before you go to market
The single biggest cause of stalled searches is a job spec that combines incompatible requirements. 'Senior ML engineer who can deploy production pipelines AND can also produce coach-facing match reports' is two jobs.
Decide whether you're hiring (a) a builder who'll ship systems, (b) a translator who'll work with coaches day-to-day, or (c) a leader who'll set the strategy and grow a team. The expectations, comp band and assessment shape are different for each.
Where the candidates actually are
Outside the top three or four leagues, very few sports data scientists are actively looking. Most of the hires we make come from one of three pools: (1) people already in the sport you're hiring for who've been quietly courted, (2) adjacent sports (NFL → football, basketball → football, etc.), and (3) technical industries - quant finance, gaming, computer vision research - where the analytical depth is strong and we backfill the sport context.
The first pool is small and contested. The second is where most of the leverage sits, if you can resist the urge to require sport-specific experience as table stakes. The third works for the most senior IC and leadership roles where pure technical depth is the binding constraint.
How to run an assessment that actually predicts the work
Drop the take-home exercise. The strongest candidates won't do them, and the take-homes you do get back tell you very little about how someone reasons under pressure.
Instead: run a 90-minute working session using one of your own real, messy datasets. Watch how they explore it. Listen to the questions they ask. Watch how they handle being told their first hypothesis was wrong.
Pair this with a structured discussion of a problem they shipped before - not the rehearsed STAR-format answer, but a back-and-forth where you can probe why they made the decisions they did.
Counter-offer and onboarding - what we see go wrong
Counter-offers in this market are common. Be ready: have your final number authorised before you go to offer, and move within 48 hours of a verbal acceptance.
Onboarding fails most often when the data scientist arrives and finds the underlying data infrastructure isn't where they were told it was. Have an honest conversation about your data state during the search - better to lose a candidate at the offer stage than two months in.
Want help running the search?
Animo Group runs sports data science searches across football, NFL, NBA, NHL and SportsTech vendors. The person you brief is the person doing the work - no account managers, no junior researchers running first calls. Reach out via the contact form.
