Projections combined from independent forecasters, and draft position from real drafts. Every row shows how many independent sources stand behind it and how far apart they are, because a number two people agree on is a different thing from one six people agree on, and you should be able to tell which you are looking at.

Field guide

How to read this

Sources counts independent forecasters, not feeds. Several websites republish the same company's projections, and one forecaster syndicated three ways is still one opinion; counting it three times would manufacture agreement that does not exist. We count the originator. Source range is how far apart those sources are. A wide range means the forecasters genuinely disagree about a player, which in hockey is usually a question about ice time or a goalie's share of the starts, and it is not a promise that his real total will land inside it.

Under three sources we say so plainly. An average of two opinions is an average of two opinions: treat a 2-source row as a hint, not a consensus, and don't reach for that player on the strength of it.

Read the rest

Two tables, one pool of players

The first view is the consensus projection: for every skater and goalie, the stat line the forecasters agree on and the points it is worth in your scoring. The second view is average draft position, the same players in the order real drafts are taking them, with the number of drafts behind each figure. The two are kept apart because they answer different questions. A projection says what a player should produce; a draft position says what a room is willing to pay. The draft kit is where the two are set side by side and the difference is measured.

How a consensus line is built

Each outside forecaster counts once, and the vote is taken stat by stat: the median goals across the origins that projected goals, the median assists across the origins that projected assists, and so on down the line. A median rather than a mean, so one forecaster who is far out on a player cannot drag the figure toward him. Our own house model never votes in that median. It is the fallback for a player no outside forecaster has published, and when it stands alone the row says one source, so nobody mistakes our lone estimate for an agreement. The consensus line is then scored in the format you picked, and only then does it become the points column.

Why a categories league still shows a number

In a points league the scoring is arithmetic: Yahoo's default weights give a goal six, an assist four, a shot 0.9, a block one, a win five and a goal against minus three. A categories league has no total to add up, because you win a category by beating the other manager in it, not by piling up points. So the number here is the sum of the player's z-scores over the categories in play: how many standard deviations above or below the pool he sits in goals, in assists, in hits, in blocks, each on its own scale. Two rate stats are handled with care. A save percentage on forty shots says less than one on twelve hundred, so a goalie's save percentage is weighted by his shots against and his goals-against average by his starts, and a two-start goalie cannot swing a category the way a sixty-start one can. Lower-is-better categories are flipped so that a higher number always means a better player.

Formats are never mixed

Where players go in a 10-team points league and where they go in a categories league are different questions, and averaging them would describe a league nobody plays in. Switching the scoring switches which drafts you are looking at, and the sample size shown is that format's. Today the one market sampled is a categories league default, so a points reader sees those drafts with the line above the table saying so until a points market lands. A goalie's draft position moves most between formats, because a categories league that counts wins, goals-against average, save percentage and shutouts weighs him four ways where a points league weighs him once. Hits and blocks work the other way: a checking-line winger with two hundred hits goes rounds earlier in a categories draft than his points total could ever justify.

Where hockey forecasters disagree, and why

The widest ranges on this table cluster in three places. Goalies, because the forecasters are really disagreeing about the share of starts a tandem will split, and a fifteen-start swing is the whole argument. Second-line forwards whose power-play unit is not settled, because a spot on the first unit is worth a dozen points on its own and camp decides it late. Rookies and players changing clubs, because there is no last season to anchor to. A wide range on a first-line centre with a fixed role is rarer and worth a second look at the news before the draft.

What to do with a disagreement

Take it to the draft kit, which sets the projection against the room's price and flags the gap, or to a player's own page, which lists every source by name. Then decide which side of the range you believe and draft accordingly: the range is the forecasters' spread, not the odds of any one outcome. The arithmetic, and what we refuse to publish, is in the methodology.

Consensus