Spatiotemporal Tracking Pipeline
Tracking study · Python
I built this to test how noise changes football tracking data.
It starts with SkillCorner-style player data, adds Gaussian
noise and 8 m position spikes, and compares four filters. I then
compare the recovered speeds and the changes in Spearman
pitch-control maps. The README says the moving average worked
best in the noise setting I tested. The repo has single-match
and cross-match scripts, tables, figures, and JSON results.
Python · NumPy · tracking data · Spearman pitch control
Synapse OS
Agent monitoring · JavaScript
I built Synapse OS to see what a group of AI agents is doing. It
can import agent setups, draw their connections, show costs,
token use, and timing, and flag possible hallucination risks. It
can pause an action when a person needs to approve it. The core
uses Node.js, Express, and Socket.IO. The frontend uses React,
Vite, Tailwind, Recharts, and React Flow. The core and frontend
are separate public repos, and there are no published releases.
React · Vite · Tailwind · Node.js · Express · Socket.IO · React
Flow
MPLSoccer Tools
Football analytics · Python
I use this repo to practice reading football event data. The
scripts use public StatsBomb matches to make pass networks, shot
maps, xG charts, heatmaps, freeze frames, pitch-control
examples, and Voronoi diagrams. It is a collection of study
projects rather than one finished application.
Python 3.7+ · mplsoccer · Matplotlib · NumPy · Pandas · SciPy
xAssassin
Football analytics concept
I'm building xAssassin to study the off-ball movements that
event data misses. It looks at space creation, decoy runs, and
pressing-trigger failures from coordinated tracking data. It is
still in development.
Python · tactical football analysis