Where to start
Choose the entry point that matches what you're trying to learn.
17 league pipelines, one rating scale
Every league below is rated on the same absolute Current Ability scale. An SHL first-liner at CA 110 sits exactly where you'd expect relative to an AHL top-six forward or NHL bottom-six player.
What's shipped
The pipeline ships alongside each Premier Pivot Rosters / TBL Database update. Each release builds on the last.
Honest about how this is built
This project uses substantial AI assistance — originally ChatGPT, currently Claude (Anthropic). The AI does the implementation work for the modeling, scripting, and documentation. The human (xECK29x) provides hockey expertise, oversight, validation, and direction throughout.
That setup matters for two reasons. First, the AI is a powerful tool but it requires deep domain knowledge to guide — without hockey instincts informing every decision, you get reasonable-looking nonsense. Second, transparency is part of the contract with the community: the methodology is auditable precisely because the documentation exists.
This repo is the public face of that transparency. The code lives in a private repo (selectively shared with research collaborators) because it includes API integrations and scraping tools that aren't appropriate to publish openly. But every methodology decision, every formula, every weight is documented here — and that's where you can challenge it, propose changes, or just understand what's happening under the hood.
Got feedback?
The whole reason this repo exists is auditability. If you think a rating is wrong, a formula is off, or a methodology choice deserves debate — file a GitHub issue, post on the EHM forums, drop a thread on r/EHM. That's all in scope.