2026-27 Fantasy Hockey Rankings

Why a z-score and not a points total

In a points league the projection is one number: the stat line scored by your league's weights. Under the Yahoo default a goal is worth six, an assist four, a power-play point two, each unit of plus-minus two, a shot on goal 0.9 and a blocked shot one; a goalie earns five per win, 0.6 per save, five per shutout and gives three back per goal against. That number adds up, and the Points format on this board ranks by it. A categories league has no such total. Your matchup is decided column by column, and a player who wins you hits by a mile and loses you goals by a hair is worth more than the raw counts suggest. The honest way to compare him with a pure scorer is to ask, in each category, how far he sits from the middle of the pool, measured in standard deviations, and to add those distances up. A player at plus two in shots and plus one in hits is worth the same three units as a player at plus three in goals, and the board treats them so. The pool is every projected skater for the skater categories and every projected goalie for the goalie categories, so a goalie is never measured against a winger.

Two rate categories need care. A save percentage on 300 shots says less than one on 1,600, and a goals-against average over 20 starts says less than one over 60. The z-score for save percentage is weighted by projected shots against and the one for goals-against average by projected starts, so a backup with a pretty number on a small sample cannot out-rank a starter with a plainer one on a full season. Goals-against average is a lower-is-better category and is flipped before it is summed, so a higher figure always means a better player in every column.

What the tinted columns tell you

The single value hides the thing a category manager most needs to know: where it comes from. Two players at a z-sum of 6.0 can be built completely differently, one at plus one across six categories and the other at plus three in two categories and nothing anywhere else. The columns show that. A deep green cell is a player 1.5 standard deviations or more above the pool in that category, the kind of edge that wins the column most weeks on its own; the lighter green is 0.5 to 1.5 above, a real contribution; the quiet middle is a player near the mean, which is neither help nor harm; the gold and red cells are below the mean, and a red cell on a player you are about to draft is a hole you will have to fill elsewhere. A blank cell means his side does not play the category: a goalie has no shots on goal and a skater has no save percentage. Read the row, not the number, when two players are close, and draft the one whose colours fit what your roster is missing.

Punting: the toggle that most changes a draft

Punting is the deliberate decision to lose a category every week in exchange for winning the others more often. A team that gives up on hits can spend every pick on skill, and the players it should want are the ones who were being held back on the full board by a weak hits column. The punt toggle beside each checked category does exactly that: the column stays on the board, greyed and struck through so you do not forget it exists, but it is taken out of the sum, and the board re-sorts. Try it before you decide. Punting plus-minus barely moves the order, because plus-minus is noisy and few players sit far from the middle; punting hits or blocks moves it a great deal, because those two categories are where the grinders and the stay-at-home defencemen earn most of their number. Punting a goalie category is a different game, since only four goalie columns exist and each one is a quarter of a goalie's worth. Your punts are written into the address of this page, so a link carries them.

Two position rules, one board

The two big platforms disagree about what a forward is. One lists every skater as a centre, a left wing or a right wing, starts two of each, and changes a player's eligibility during the season as he moves between wings; the other pools the three into a single forward slot, starts nine forwards and adds a utility seat any skater can fill. The position rule toggle switches the board between the two. Under the pooled rule the pills read F, D and G, the filter offers those three, and the value is priced on the roster with the utility seat; under the split rule the pills carry C, LW and RW and the filter follows. The z-scores themselves do not change, because a category is a category whichever slot a player fills, but the replacement level does, which is why a fourth-line centre can look slightly different on each.

Goalies on the same board

A goalie's line is wins, goals-against average, save percentage and shutouts, and most of it depends on the club in front of him and on how many nights he gets the net. Under the default set those four columns are four of the ten categories, so a workhorse starter on a good club carries as much category weight as a first-line winger, and the board seats him accordingly. Where the sources disagree most is usually the start count, not the save percentage; a projected 60-start goalie who becomes a 40-start goalie in a tandem loses a third of his wins and a third of his shutouts at once. The board reads season totals, so it cannot see this week's starts. Read the goalie tandem guide for who actually gets the crease.

Points and roto

The Points format scores the stat line with the Yahoo default weights above and ranks by the total, which is what most public leagues do. The Roto format is a fixed nine-category set, six skater and three goalie categories, scored the same z-score way as the picker but without the picker, because roto pools tend to run the same categories all season and compare cumulative totals rather than weekly matchups. On a roto board the tinted columns matter even more: a roto standing is won by being solidly above average in every column, not by dominating two, so a row with no red cells is worth more there than its sum suggests.

Where the projections come from

Every outside origin gets one vote, and the stat line the z-scores are built from is the median, stat by stat, across the origins that published that stat. Our own model never votes in that median. It is the fallback: a player nobody outside projects takes the house line whole and shows a source count of one, which is why the deep end of the board is mostly ones. The outside sources stop a few hundred players deep. The draft position beside the value is the market's price for the player, so a high z-sum with a late draft position is a bargain the room has not noticed, and the reverse is a player the room likes more than the categories do.

What this board is not

This board prices players for the season, in redraft. It knows nothing about this week's calendar: a club with two games and a club with four carry the same value here, which is right for a draft and wrong for a Monday lineup. The schedule planner answers the games question and the goalie tandem guide answers the crease question. The same tiers print one column per position as the printable cheat sheet, and for pick-by-pick help in a live draft the draft companion crosses players off as they go. Details in the methodology.