Features

Basketball-first, sport-agnostic by design

OCF v1 covers basketball across the FIBA, NBA, NCAA, and NFHS rulesets, but the underlying architecture — entities, actions, frames, and named positions resolved per ruleset — is not basketball-specific. Future versions can extend the same model to other sports without breaking existing documents.

Real-world coordinates

Every position in an OCF file is expressed in real-world units — meters for FIBA, feet for NBA, NCAA, and NFHS — with the origin at midcourt center. This keeps diagrams grounded in actual court geometry instead of an arbitrary pixel grid, so the same drill renders correctly at any scale or aspect ratio.

Named court positions

Rather than forcing every position to be a raw coordinate pair, OCF defines a registry of named positions — top_of_the_key,left_elbow, inbound.baseline_left, and more — grounded in official ruleset geometry. Coaches can author drills by naming positions instead of guessing coordinates.

Semantic actions

Frames describe what players do — move,cut, screen, defend,dribble, pass, shoot,rebound, pickup — rather than raw geometric lines. This makes drills legible to both humans and machines, and is a prerequisite for the LLM-driven generation OCF is designed toward.

Automatic ball possession

A top-level balls[] array (carried_by /at / dead) supports multiple balls per drill. Possession transfers are inferred automatically from passand pickup actions, so authors don't have to manually track and restate who's holding the ball in every frame.

Hybrid frames

Frames combine explicit start_state /end_state anchors with action sequences, and support outcome branches (make / miss / …) to model continuum drills where the next step depends on what happened in the last one.

Progressive detail

Only player and type are required per action — variants, tags, and timing are all optional. Coaches can sketch a drill quickly with minimal annotation, or go back and add full detail later, without the schema forcing an all-or-nothing choice.

LLM-friendly

The semantic action model is structured enough that a language model can reliably generate valid OCF documents from a natural-language drill description — a first step toward OCF's end goal of text-to-diagram generation.

Accessibility

OCF ships with WCAG-guided color scheme recommendations for rendered diagrams, so tools built on the format start from an accessible default instead of retrofitting contrast and color-blindness support later.

FIBA import path

OCF defines a coordinate mapping from the FIBA Europe Coaching internal format, giving existing FIBA-authored content a path into the open standard instead of starting from scratch.