Ten Avalanche, Eight Lightning

The Drafters board ranks 224 players and they are spread across all thirty-two NHL teams, an average of seven apiece. Carolina has the most with eleven. Vancouver and Calgary have three each.

I wanted to know what that spread was worth. The intuition it invites is obvious: teams with more draftable players are better places to shop, so know which ones are stocked and lean that way when two players look close. It is the kind of heuristic that sounds like knowledge.

It is worth nothing, and the reason is worth a closer look.

Counting a team's players is counting the wrong thing

I compared each team's number of ranked players against where those players actually sit on the board. The correlation is 0.022, which is essentially zero. Knowing a team has ten ranked players instead of six tells you nothing about whether any of them go early.

The cleanest illustration is at the top:

Colorado has more players on this board than Tampa Bay does, and Tampa's typical player goes a hundred picks earlier. Both facts are true at once, and only one of them is about your draft.

The Avalanche list explains itself as soon as you read it in order: MacKinnon at 1.6, Makar at 5.6, Necas at 16.7, and then nothing for eighty-three picks, until a goalie at 99.9. Five of their ten ranked players go after pick 148. Tampa's eight arrive at 3.3, 28.9, 40.3, 46.6, 50.8, 64.8, 104.4 and 167.7. Seven of the eight are gone before the deadline, on a near-identical player count.

Count to your deadline instead

Keep counting, but stop at 148.

Exactly 148 of the 224 ranked players sit at or before that pick, so the average team supplies about 4.6 players you can actually plan around rather than seven. Recount the league that way and it rearranges itself:

San Jose drops twenty-two places. Philadelphia drops twenty-one. Both of them look like well-stocked teams by the count everybody does in their head, and both of them hand you three players before the board goes quiet.

St. Louis moves the other way for the opposite reason. Five ranked players is among the fewest in the league, and all five arrive before 148. The roster is not deep; all of it simply sits in front of you.

The teams that sell you exactly one thing

The pattern underneath the reshuffle is that some teams are a star and a void.

Buffalo is the extreme case. Dahlin goes at 17.3, Tage Thompson at 23.0, and the next Sabre on the board is Mattias Samuelsson at 156.1, a 133-pick hole and the largest inside any team. San Jose runs Celebrini at 3.8 and then nothing until pick 97.4. Colorado's gap is 83 picks, Edmonton's 80.

This is where the counting error costs you. Taking Celebrini fourth overall buys you one player, because the rest of San Jose does not arrive until round ten. The two picks that feel like they should follow (the linemate, the defenceman you half-remember) are ninety picks away, and by then you will have forgotten why you wanted them.

Everyone has somebody, which is the real trap

What kills the heuristic outright is that every one of the thirty-two teams has at least one ranked player. There is no team to write off.

But they arrive at very different times. Through pick 25 you have touched fifteen teams. Through 50, twenty-five. It takes until 100 to reach twenty-nine, and you do not touch all thirty-two until pick 148, the deadline itself.

Seven teams have nobody at all in the first fifty picks: St. Louis, Carolina, Seattle, Calgary, Pittsburgh, Philadelphia and Vancouver.

Carolina is on that list while leading the entire league in ranked players. Eleven Hurricanes carry a draft position and the first one is Sebastian Aho at 66.5. Whatever "Carolina is deep" means, it does not mean there is a Hurricane worth a top-fifty pick.

What to do with this

Not much. Do not use team affiliation as a tiebreaker. The board has already priced every player on it individually, and the team column adds no information the ADP does not already carry. When two players look close, a deeper sweater is no reason to prefer one.

The one place it earns its keep is planning the shape of your middle rounds. If you are counting on a run of players from somewhere between picks 100 and 148, check that the somewhere actually has them. Buffalo has nobody in that stretch at all; San Jose and Colorado have one apiece. Carolina has four, Toronto and Washington three each.

That reverses part of the argument above. Carolina, the team with no top-fifty player and the one whose eleven ranked names flatter it most, is the best supplier on the board in exactly the window where you are trying to fill a hole. So the count was never useless; it was answering a question I had not asked.

What would make me wrong

• If your league is not ten teams. Everything here hangs on 148 being the deadline, and that number comes from a 180-pick draft. In a twelve-team league the deadline sits elsewhere and every supply count shifts with it. The method transfers; the table does not.
• If you have a real read on a roster. I am measuring what the draft room thinks, not what is true. If you have watched enough San Jose hockey to know which of those six late names is about to matter, the void I am describing is an opportunity and I have just told you to ignore it.
• If depth pays off in-season rather than at the draft. A team with eleven ranked players plausibly gives you more waiver targets in January than one with three. That is a real edge and this piece does not measure it, because I am only looking at draft day.
• If ADP is compressing the late names artificially. Half of my "void" claims rest on players sitting past 148, where the board's ordering is thin. If those players are more distinct than their numbers suggest, the barbell teams look worse on paper than they are.

The short version

How many ranked players an NHL team has is uncorrelated with when those players go: the number is 0.022, which is another way of writing "no relationship at all." Colorado has ten and Tampa has eight, and Tampa's median player goes a hundred picks sooner.

Count to your deadline instead of counting the roster. Every team has someone, but seven teams have nobody in the first fifty picks, and seven hand you three players or fewer before the board stops working.