About Sport Tech...

AI is making its way into every aspect of our lives, and sport is no exception. Athletes, teams, and sport organizations are all asking the same question: when will AI become a reliable partner in sport analytics?

Broadly speaking, sport analytics can be grouped in two main families:

1.    Technical analysis – focused on improving the execution of a specific movement, such as a tennis serve, a golf swing, a gymnastics flip, or a runner’s gait. The goal of such analytics is to help athletes gain efficiency and precision.

2.    Tactical analysis – focused on how teams or groups of athletes work together to make a difference in the game. This typically involves tagging events and generating stats on one’s own team or the opponent, ranging from objective data (passes, goals) to more subjective insights (quality of a transition from attack to defense).

Not surprisingly, team sports organizations are particularly eager to use AI to automate tactical analysis, by turning raw game footage into structured events and statistics.

From what we observe combining the current state of the AI sport market and the needs of our customers, AI in sport can be thought of in a 3-tier approach when it comes to tactical analysis:

Tier 1: Basic events

AI identifies simple game phases and actions – the kind of things any casual fan could recognize.

• Football: passes, corners, penalties, goals

• Tennis: rally length, forehand/backhand, serve

Tier 2: Advanced insights

AI starts providing richer data, including player positioning, individual statistics as well as more precise event qualifications – insights that require a trained eye.

• Football: player positions, lineups, player-by-player metrics

• Tennis: shot classifications such as passing shots, lobs, smashes

Tier 3: Comprehensive, sport-specific analytics

AI delivers tactical data which would typically require the eye of a sport expert.

• Football: pass completion rates, defensive line height, pressing efficiency

• Tennis: spin variations (slice, topspin), precise in/out decisions

Where do we stand in 2025?

• Tier 1: Several sports tech providers already offer AI-generated data at this level, with growing reliability. These models often require just a single video feed.

• Tier 2: This level of data remains a challenge in sport. Tracking all players on a field requires multiple camera angles whilst accurate event qualification are calling for extensive annotations to train the models. Reliability is improving but not yet consistent.

• Tier 3: Obviously the Graal when it comes to AI produced data. Currently production of data reaching  this level of detail and expertise is extremely limited.

AI in sport is advancing quickly — yet, in 2025, human tagging remains the backbone of reliable data generation. We’re still at the early stages of this journey, and the coming years will show how fast AI can truly scale from basic (Tier 1) to advanced, expert-level analytics (Tier 3).

What’s certain, in any case, is that this progress won’t come cheap. It will demand significant processing power and massive amounts of annotated data. The “free” generative AI tools many of us use today have created a false sense of value — remember that if it is free you are the product!

September 2025