MSARSE

MSARSE stands for Micro-Sector Adaptive Racing Strategy Engine. It cuts a racing lap into micro-sectors of about 30 meters and compares two laps sector by sector, to show where a driver loses time. It was a team hackathon project, coded over two days in November 2025 on the official Toyota GR Cup data for Circuit of the Americas.

The race data has no GPS. To place each telemetry sample on the track, a conversion script generates coordinates, so every position in the app is synthetic. That limits everything downstream: a map of where time is lost is only as real as the positions under it.

A Python package with a data loader, a Haversine distance calculator, the micro-sector engine, a lap comparator and plotting. A command-line tool and converters for the race, lap and weather files. A Streamlit app, and a FastAPI backend with a small web front end. Seven test files, some of them property-based. Despite the name, there is no machine-learning or strategy model in it.

The pipeline runs end to end, but its results are not meaningful yet. The project's own run summary reports lap times of 17.7 and 999.1 seconds, which a sampling step has plainly broken. It was team work, and this page does not single out who did what. No result from the event is on record.

Fix the lap timing first. Then find real positions, or drop the map and compare laps by distance alone. And capture a comparison of two different laps: the only screenshot on file shows a lap compared with itself.