A points league and a category league are not the same game with different scoring — they are different player pools. This post measures how different, category by category, from the 2026-09-08 projection run: 724 skaters with enough NHL history to project (475 forwards, 249 defensemen), 84-game totals throughout.
Ranks below are league-wide among those 724 skaters, not within a position, so "top 50 in blocks" means top 50 in the league. The six columns a category league usually reorders on are goals, assists, shots, power-play points, hits and blocks; penalty minutes and faceoff wins are in the first table too. There is no plus/minus board anywhere on the site, and the reason is in that table's last two columns.
How far each category moves the board
The middle columns are overlap: of the 50 skaters the model likes most in a category, how many are also top 50 in points, and how many sit outside the points top 100 entirely. The last two are from last season's backtest — the rank correlation between the projected order and what actually happened, model against simply carrying the previous season forward. Higher is better.
Goals, assists, shots and power-play points are mostly the points board wearing a different hat: between 32 and 38 of each top 50 is also a top-50 scorer. Hits and blocks are not: one player in the hits top 50 is a top-50 scorer and one in the blocks top 50 is, and 49 and 48 of those fifty sit outside the points top 100. The blocks board is also the only one that belongs to a single position: all 50 of the top 50 are defensemen, where the goals top 50 has none.
The last two columns are the part worth sitting with. The model ranks PPP, blocks and hits more reliably than it ranks points — 0.84 for PPP, 0.83 for blocks, 0.82 for hits against 0.79 — and faceoff wins at 0.95 more reliably than anything else on the board. Counting stats that look like noise are mostly usage, and usage repeats. Plus/minus is the exception that proves it: at 0.24 against 0.29 for carry-forward, it is the one category where the method makes the order worse, which is why we do not publish it.
The skaters who fill more than one column
Fifteen skaters are top 50 in four of the six. No skater is top 50 in five.
Fourteen of the fifteen fill the same four — goals, assists, shots, power-play points — which is another way of saying they are the points board. Brady Tkachuk is the whole exception: 249 projected hits, 29th in the league, on top of 326 shots (3rd) and 27 power-play points. He is 50th in points, which in a points league makes him a useful forward and nothing more. His brother Matthew Tkachuk projects 87 points to Brady Tkachuk's 75 and fills the ordinary four; the hits column is the only thing separating them, and it is worth 137 a season.
Hits and blocks are a different league
Kiefer Sherwood leads the league at 401 projected hits and ranks 242nd in points. Alexander Romanov leads blocks at 191 and ranks 512th. Neither board asks anything of the points board, and they do not ask much of each other either: exactly one skater in the league is top 50 in both hits and blocks, Radko Gudas, 619th in points.
That is the practical size of the difference. Only 94 skaters are top 50 in at least one of the four scoring categories. Add hits and blocks and the pool that matters to somebody's lineup grows to 188 — the physical columns roughly double the number of players a league can care about, and almost none of the additions are visible on a points board.
Shots, and the stars who do not take them
Shots are where the two formats quietly disagree about forwards everyone rates. These are the forwards in the projected top 40 by points who fill two or fewer of the six columns.
Thomas is the sharpest case: 19th in projected points, and top 50 in exactly one column — 151 shots is 208th in the league, 22 power-play points is 70th. Then look at the last column, where he is 7th at 806 faceoff wins. Whether he helps you is entirely a question of which columns your league counts. Matthews is the same problem pointing the other way: 37th in points, well below the names above him here, but 5th in the league in shots and 6th in faceoffs — two columns a points league does not pay for at all.
The faceoff board is the one the model is surest about — 0.95 against 0.93 for carry-forward — and the one that reshuffles the least from season to season, because it is almost entirely a question of who takes the draws.
What this does not know
Ice time and power-play units are last season's until camp lines are synced, and power-play points move more than any other column when that lands. Hits and blocks are box-score counts recorded by rink scorers and vary by building; the projection regresses them but does not correct for the scorer. And the top 50 cutoff here is a convenience, not a roster size — the fantasy page will rank any of these categories under your own league's weights.