Problem
Raw battle logs show what happened, but they do not explain performance trends, matchup context, or whether a model result is reliable enough to use.
StarrLabs
A battle research companion that makes player history useful.
StarrLabs is a public Brawl Stars analytics companion for finding player profiles, reviewing battle history, understanding performance patterns, and turning collected evidence into clearer post-battle insights.
From a technical product-management view, the priority is trustworthy analysis: preserve point-in-time data, explain where each result comes from, protect player identity, and only promote prediction or classification work when honest evidence gates pass.
Product
StarrLabs combines player discovery, verified profiles, Battle Archive, Victory Score analysis, opponent context, and explainable post-battle research.
System
The platform runs on an automated Pi collection and operations layer with PostgreSQL history, immutable evidence datasets, bounded API work, backups, and model-readiness telemetry.
Outcome
Players get a readable research record for their battles while new analytics earn trust through provenance, calibration, and explicit promotion gates.
Technical product direction
Deepen the analysis without outrunning the evidence.
The next product direction is richer Victory Score contributors, matchup and synergy context, stronger account controls, and carefully gated model families that remain separate when their battle formats or data boundaries differ.