A third look across the ice.
Most hockey stats measure what already happened. The interesting work is figuring out what the box score missed — what coaches, scouts, and players are actually watching for. That is what this site is for.
The premise
The name comes from the ice. A defenseman's first read is the puck. The second is the man. The third — the one that separates good from elite — is the play developing two passes ahead. Public hockey analytics is mostly stuck on the first two. We are trying to publish the third.
That means stats no one else uses, deliberately. If a number is already on every other site we reach for it only to give context. The headline metrics here are ones we built because we couldn't find them elsewhere: shift-level success against an opportunity-adjusted league baseline, an implied-purpose classifier that infers whether a shift was meant to attack or defend, a Net Rating that respects position and pace, and an expected-goals model fit only on the chip-tracking era.
What we will not do
- Publish a stat we cannot defend. Every headline metric has a writeup with the formula, the calibration, and the failure modes.
- Treat the box score as ground truth. Goals are noisy; shots are noisy; hits and giveaways are scorer-dependent. We use the underlying signal where we have it.
- Pretend the model is the player. A good prior plus a small sample is still a small sample. The numbers are inputs to judgment, not substitutes for it.
How to read the site
The Players and Teams pages are sortable tables with presets — a few curated views that we think are worth your time, instead of a hundred columns you have to assemble yourself. The Shift Data tab is where the newer work lives: shift-level success, implied purpose, and the residual metrics derived from them. The Glossary & Methods tab carries definitions and long-form writeups for every metric we publish.
Sources
Play-by-play, shift charts, and boxscores are scraped from the NHL's public APIs. Roughly four thousand games of recent NHL history feed the models. Where we lean on prior public work — Dom Luszczyszyn's Net Rating, Evolving-Hockey's xG framing, the chip-era shot-location discussion led by CJ Turtoro — we say so in the relevant writeup.
Contact
Mistakes, disagreements, and pull-no-punches feedback are all welcome. The fastest route is the email address listed on the Glossary page.