Finishing

Step 01 · Every shot

180 shots on goal, each with a price

Finishing is the one place in net goals where the puck going in counts. The on-ice components price chances and never look at the result; Finishing books the difference. Each of Geekie’s 180 shots on goal and goals in 2025-26 was priced by the same expected-goals model (location, shot type, rebound and rush flags, side of the ice, season and period calibration; empty-net shots excluded). Missed shots are left out here: they never test the goalie, so the shots-on-goal version of the model predicts held-out goals best.

Step 02 · Where it came from

Beating the price shot by shot

Sorted by how dangerous the shot was, his conversion against the league’s on the same kind of shot:

His five least likely goals each added almost a full goal over its price; his best chances that did not go in each cost their price:

Step 03 · Goals above expected

GAx

That ranked 2 of 933 shooters in 2025-26. League-wide the 72,554 shots on goal produced 7,578 goals on 7,582.6 expected: the model is calibrated so the two match every season, which is what lets a positive number mean better than an average shooter on the same shots.

Step 04 · Into goals

Pricing a goal above expected

One adjustment first. A goal in regular-season overtime ends the game, so it is worth what the shooter’s team gains in the standings, not a regulation goal’s third of a point: each overtime shot’s goals above expected is weighted by that value (about 1.5 regulation goals at 3-on-3). Geekie had 4 overtime shots on goal:

The same standings regression that prices on-ice play gives 25.33 points per goal above expected per game; divided by 27.33 points per goal per game, that is 0.9265, essentially face value. A goal above expected is a goal, and only the shooter gets it here.

Step 05 · The assist share

Half of it belongs to the passers

A goal is rarely the shooter’s alone. Half of every even-strength and power-play shot’s goals above expected is handed to the teammates on the ice for it (weighted 4 : 2 : 1 for the primary assist, the secondary and everyone else), and half of its expected-goal cost is shared the same way. That credit lands in their Playmaking; the shooter pays it out of Finishing. On Geekie’s 177 EV and PP shots on goal (37 goals, 22.86 xG):

The file’s debit is −6.6752. Like every reallocated column it is then re-centred on the league (−0.000282 per game):

Step 06 · Position

Measured against the average forward

After every other step, each component is re-centred once more, separately for forwards and for defencemen, so the average forward in 2025-26 is exactly zero in every column. The games-weighted mean finishing of all forwards that season was −0.00059 net goals per game, so 81 games of it come off:

Step 07 · Result

Per 84 games

Ratings are quoted per 84 games, one full schedule, so a part season and a full one sit on the same scale.

✓ Recomputed +7.52; his 2025-26 row on the player card and the Players table shows +7.5. They match.

Step 08 · Context

How unusual is this?

Skaters with 40 or more games in 2025-26 (612 players). Morgan Geekie ranks 2 of 612.

How much of a season like this is skill? The model estimates, from how much players’ own rates bounce between seasons versus how much players differ, that a career rate of finishing is half signal and half noise after 616 shots on goal. Season rows like this one are shown exactly as they happened; careers and the card’s Regressed view are pulled toward average by that amount.

For Finishing the half-signal point is counted in shots on goal rather than games: several seasons for most shooters. Shooting luck is loud.