2026-27 Standings & Projection

GP and Pts are the real standings (65 of 1344 games played). Proj is where the model has each team finishing: the points already banked plus 10,000 Monte Carlo runs of the rest of the schedule. Each team’s strength comes from its recent lineups (the players who dressed in its last 10 games, every one priced by the model), and a calibrated win-probability model plays the games, back-to-back rest edges included, with overtime and shootouts modelled separately (shootouts from each team’s shooters’ and goalies’ shootout records). ± is one standard deviation across sims, and it isn’t the same for everyone (hover it): game luck plus each roster’s projection uncertainty — skater and goalie track records, and how much the team would lose to injuries given its next best 3 F, 2 D and 1 G. Playoff% is the share of sims that make the playoffs. Updated daily.

What if? Got a trade you keep pitching to your group chat? Make it, sign a free agent or change a lineup, and re-run this simulation in the trade sandbox.

Playoff odds, seed by seed

Every one of the 10,000 simulated seasons above is played on through the real playoff bracket: three per division plus two wild cards, the conference’s better division winner draws the second wild card, best-of-seven with home ice to the better record. Playoff games use the same game model as the regular season (no shootouts), each sim’s team strengths carry over from its regular season, and the starter is in net every night. Division place and Wild card are where the team finishes when it makes it; Missed is its place among the conference’s eight non-playoff teams (9th = first team out). Each row adds to 100. Hover the first-round opponent for the next most likely ones and the chance of home ice. The small numbers beside Playoffs and Cup are the change since the morning of Thursday, October 8.

Big games

Who should you root for? Each game on the card is played both ways on the same 10,000 seasons: once with the home team winning, once with the visitors winning (the loser still banks its overtime point as often as the game would really go past regulation). A bar is how far a team’s odds swing between those two results, and it points at the result that helps it: rivals in the race are often rooting harder than the two teams on the ice. The numbers at the ends are each result’s change from today’s odds. Switch to Reg · OT · SO for all six results: how a game ends matters, because the loser of an overtime or a shootout keeps a point, and ties in the standings go to regulation wins first, then regulation plus overtime wins (shootout wins count last). The small percentages are how often the model sees each result. Pick your team to pin it to every game and get its rooting list.

Projected points through the season

Each point is the standings simulation’s mean projected points for the team (10,000 runs of the rest of the schedule, played games locked in) as it stood on the morning of that day’s games, from the preseason projection on; below it, the share of runs that make the playoffs. Hover a panel for that day’s numbers; hover or tap a team to pick out its line. Each team page has its own game-by-game chart with the ±1 SD band and the playoff lines. Latest: the morning of Oct 9.

Roster ratings by category

Projected is the simulation’s own forward-looking roster strength (projected net goals per 84, weighted by expected games) — the number the standings above are built on, and the one this table is sorted by. The other columns are each roster player’s career net goals added per 84 games (the same rates as the Players page), weighted the same way and regressed toward zero for small samples. Goaltending is the goalies’ GSAx per 84; Overall is the skater components plus goaltending. Teams leaning on prospects with no NHL history read low in the career columns: the model can’t price games that haven’t happened yet. Two tilts worth knowing. Every component is centered to zero over all rated players, but NHL rosters are the talented top quarter of that pool, so talent-linked columns (Finishing, Penalties, EV Off) run positive across the league — good players draw more penalties than they take. And Faceoffs run mildly negative because the value-based roster builder tends to cut low-scoring faceoff specialists in favor of skill wingers.

Rank is within the grouping shown (division, conference or league). The red line is the cut: a division’s three guaranteed playoff spots, or a conference’s eight places in projected order (wild cards can cross divisions). The roster ratings above use each team’s recent lineups and update daily.