NHL analytics powered by
real statistical modeling

What We Do

We build data-driven tools that make NHL statistics more useful and accessible. Our models pull from publicly available data and apply statistical techniques to surface insights that go beyond the box score: every player gets a rating, every contract gets a value verdict, and every tool on the site is built on those same numbers, so a player looks the same on his profile, in the Trade Machine and in GM Mode.

Our Methodology

Player ratings are league percentiles from 0 to 100, measured against players at the same position: a 90 means better than roughly 90% of them. Each skater gets an overall plus component ratings for offense, defense, play driving, finishing, power play, penalty kill and speed. The current season blends the last three, weighted toward the most recent, so one hot or cold stretch doesn't swing a rating too far. Goalies are rated on goals saved above expected (GSAx): the goals they stopped compared with what an average goalie would allow on the same shots.

Our contract valuation model is built on Wins Above Replacement (WAR), drawing inspiration from the publicly available research at Evolving-Hockey. We model expected AAV as a function of WAR and market conditions, then compare it against each player's actual cap hit to produce a value rating.

Our draft model uses CSS scouting ranks as a Bayesian prior, adjusted by NHLe (NHL equivalency) production. NHLer% represents the probability a prospect plays 164 or more NHL games. Star% represents the conditional probability of becoming a top-line contributor, given they reach the NHL.

Our expected goals (xG) model estimates the probability that any given shot becomes a goal, using shot distance, angle, shot type, and whether the shot came off a rebound or off the rush. It is deliberately blind to who took the shot, so comparing a player's actual goals against his expected goals measures how he finished the chances he generated. On a like-for-like basis, excluding empty-net situations, the model currently lands within half a percent of the league's real goal total.

The Trade Machine and GM Mode value players the same way: rating and years of team control, plus the gap between what the model expects a player to be worth and what he's paid. Draft picks are valued by round and the original team's projected finish.

All data is sourced from publicly available NHL APIs, Daily Faceoff, CSS rankings, and official transaction records. Models are updated regularly throughout the season.

New to advanced stats? Start with Stats Explained, or check the FAQ.

The Community

BluelineMetrics is built for fans who like to argue with numbers. Create a free account to get a public profile with a username, profile picture and favorite team. Publish your trades to Community Trades and your GM Mode sims for others to like, react to, discuss and load, follow other fans and see what they're up to in the activity feed, compete in Pick'em and on the puzzle leaderboards, and keep your daily puzzle streaks synced across devices. Find people to follow.

Premium

Most of the site is free. Blueliners (Premium members) also get the full daily picks board, the player and goalie card explorer with every season back to 2008-09, the live tracker's advanced analytics, unlimited GM Mode sims with saving and publishing, the daily puzzle archive and Unlimited mode, 60 BluelineBot questions a day, no ads, and a blue check next to their username.

Contact

Questions, feedback, or data inquiries. We'd love to hear from you. See the contact page for what to include.