# SkillCorner vs event data: what additional evidence are you buying?


**Tracking-derived data should be evaluated against a specific movement question, not purchased simply because it contains more observations.** SkillCorner describes physical metrics, XY tracking and Game Intelligence combining tracking and event data. Opta Vision similarly describes synchronised event and tracking information. ([skillcorner.com](https://skillcorner.com/us/sports/soccer))

### How do events, event snapshots and continuous tracking differ?

| Evidence structure | What the documented approach contains | Suggested recruitment use | Boundary to test |
|---|---|---|---|
| Event records | Recorded actions and locations; Wyscout markets an event-data pack | Examine passing, carrying, shooting and other defined actions | Which off-ball questions remain unanswered? |
| Event-linked spatial snapshots | Statsbomb 360 captures player positions at on-ball events | Inspect surrounding structure at relevant actions | Do not treat event snapshots as a continuous movement record |
| Continuous tracking-derived records | SkillCorner describes player and ball tracking, including off-camera extrapolation | Investigate movement sequences and physical context | Separate observed locations from estimated locations |
| Merged event and tracking products | Opta Vision describes synchronised datasets and derived metrics | Connect movement, passing choices, pressure and outcomes | Validate synchronisation and the provenance of each output |

These distinctions follow the suppliers’ descriptions, rather than an independently established quality hierarchy. ([hudl.com](https://www.hudl.com/products/wyscout/data-api))

### What changed in Statsbomb 360 in October 2026?

Hudl’s **29 October 2026** announcement describes AI-powered player identification within 360 snapshots. It says the enhancement is delivered by API, included in existing Statsbomb subscriptions without an additional charge, and available for qualifying matches from **1 August 2026** across more than 80 competitions that passed its quality threshold. This is a scoped release statement, not proof that every historical match or target competition has identified frames. ([hudl.com](https://www.hudl.com/blog/statsbomb-360-upgrade-adds-ai-powered-player-id))

**Buyer implication:** before adding a tracking product for a positioning question, test whether newly available incumbent functionality answers enough of it. Conversely, do not assume identified snapshots satisfy a requirement for continuous runs, speed or movement between events.

### How should the tracking suppliers be compared?

| Documented approach | Public description | Questions to put into the trial |
|---|---|---|
| SkillCorner | Single-camera tracking with off-camera extrapolation; physical and Game Intelligence products | Which records are observed or estimated? Which outputs are included in this package? |
| Opta Vision | Event/tracking synchronisation; a diffusion model generates out-of-shot player trajectories | How are generated trajectories flagged and validated for the proposed use? |
| Hudl Tracking Data | Single-camera player-location, physical and off-ball data connected to events and video | Does the proposed entitlement cover the club’s competitions, seasons and API requirements? |

The offering descriptions are supplier statements; the trial questions are our recommendations. ([skillcorner.com](https://skillcorner.com/us/sports/soccer))

Do not turn certification or a headline accuracy figure into a blanket endorsement. A study published in *Sports Engineering* on **11 May 2026** examined automatic event detection using broadcast-derived tracking from one World Cup match. It found promising performance for some events, but sensitivity to camera feed, visibility and underlying tracking quality. That is evidence for task-specific validation—not a ranking of the suppliers above. ([link.springer.com](https://link.springer.com/article/10.1007/s12283-026-00549-4))

**Suggested acceptance tasks:** inspect the same player’s recovery runs, movements to receive and actions around possession losses; check identity continuity after substitutions; compare outputs around camera cuts; and require explanations for missing or estimated observations.