This module is being developed from the local tracking-data demo project. It moves beyond static box-score analysis by connecting event data, tracking frames, route prototypes, decision regret, state windows, role fit, error mining, and player-level decision profiles.
/model_folder/futrixmetrics-tracking-data-demo The prototype already contains FastAPI delivery, report pages, tracking/event databases, M1-M7 model outputs, and LTCE warehouse tables.
The project is not just a running-distance page. Its current direction is to turn local movement, ball state, pressure, passing lanes, and action outcomes into decision intelligence that coaches and analysts can review frame by frame.
The demo warehouse links action records with tracking clips, local teammates, opponents, ball control state, start/end locations, pressure indicators, route labels, and player summaries. That gives each decision a local spatial context instead of treating it as an isolated event.
The local project treats M1-M7 as one connected training and reporting pipeline: route extraction feeds decision quality, decision quality feeds state windows and counterfactual review, and the later modules summarize role fit, error type, and final player profile.
Clusters pass, carry, and shot paths into route labels, then builds player-level top routes and style fingerprints.
Scores real actions against learned alternatives and separates reasonable, suboptimal, and poor decisions.
Segments performance into five-minute windows to identify hot, stable, stressed, and fatigue-drop phases.
Generates candidate alternatives and compares expected gain against the action that actually happened.
Learns behavior embeddings and compares players with role centroids to support position-fit discussions.
Classifies failures into pressure, vision, decision, and technical error families for training priority.
Combines M1-M6 into player-level decision profiles, peer similarity, and recruitment-ready summaries.
The current demo includes a working player-level decision profile page. It is still a prototype, but it already shows how route behavior, decision quality, LTCE signals, and correction recommendations can become a product surface.
The LTCE design upgrades M2 and M4 from event-template alternatives toward tracking-native local counterfactuals: which lanes were actually reachable, how nearby players could respond, and why one option ranks higher than another.
First-version LTCE recalls carry lanes, pass lanes, pressure escape lanes, and contest-risk variants from local tracking context.
The design focuses on ball, carrier, nearby teammates, and nearby opponents over a short 1.5-2.5 second window.
Candidate ranking is designed around retention, progress, pressure relief, creation, risk, and shape cost signals.
The current data supports two-dimensional local tracking, speed, pressure, teammates, opponents, route labels, and event outcomes. It does not yet provide body orientation, ball height, or detailed body-pose labels, so the page describes LTCE as a practical local-window model rather than a full 22-player simulator.
This project is still moving from research prototype toward product module. For deeper project details, demo access, data collaboration, or commercial partnership discussions, contact us and we will share the appropriate demo route.