The eleven seats: where your players' games actually go

A twelve-team daily-lineup league starting 2C, 2LW, 2RW, 4D and one UTIL. Everything is measured three times over, for the 3-, 4- and 5-man benches real leagues carry, on the real 2026-27 schedule.

Where the games go

Every skater you draft belongs to a club playing 84 games, and you will start nowhere near all of them. What you collect is the subset that lands on a night when you have an open seat of the right kind. The rest score nothing at all - a bench point is not a point.

So we counted. Eleven starting seats plus a bench, 300 legal rosters dealt at random from the pool a twelve-team draft actually empties, and the best possible lineup set on every night of the season.

Each bench spot you add brings 84 more player-games and throws most of them on the floor. Going from 3 to 5 adds 168 games to what you own and 88 to what you start; the pile you never touch grows by 80. Which is the first sign that a deep bench is a worse deal than it looks.

It isn't congestion

The natural explanation is that your good players crowd each other out on a busy Saturday. It isn't, or at least not mostly.

A club plays 84 games across 185 dates, so on a typical night fewer than half the league - about 45% - is on the ice. The calendar is lumpy in both directions: 61 nights carry four games or fewer, 61 carry ten or more, and 63 sit in between. A 15-man roster has just under seven men playing on an average night, and eleven seats waiting for them.

Empty seats, in other words, are not the problem. The problem is which seats. Four centres playing on a night with two C slots and a UTIL means one of them sits while seven other seats go unused on the same page. Everything below follows from that, including why it gets worse the deeper your bench runs.

Lever one: a second position

Same league, same roster shape, same everything - chase multi-eligible forwards wherever the board allows, against avoiding them entirely:

No talent moved in either direction. And the gap grows with the bench, 30 games at 3 against 49 at 5, because a deeper bench is simply more men queueing for the same eleven seats, and a second position is what lets one of them jump the queue.

Per pick, trading one single-position forward for a multi-eligible one:

Notice that this number doesn't climb the way the one above it does - +2.7 at a 3-man bench, +2.4 at 5. A single swap is worth about the same wherever you make it. What a deeper bench gives you is more picks to spend on the tiebreak, which is why the end-to-end gap grows while the per-pick figure holds still. The spread around it is wide because the value of any one swap depends entirely on what else you already own.

Here it is on your own draft. The slider trades your picks one at a time for the nearest comparable player on the board carrying a second position - same seat, projecting within a rounding error of a game, so eligibility is the only thing really changing. Switch the bench depth and the draft re-runs at that setting. Benched games score nothing, which is the point.

It costs nothing to act on, because the tag is already everywhere. 41 of the top 100 skaters on our board carry more than one position:

Leon Draisaitl (C/L), Jack Hughes (C/L), Mikko Rantanen (L/R), Jason Robertson (L/R), Clayton Keller (L/R), Tim Stützle (C/L), William Nylander (C/R), Brandon Hagel (L/R), Matt Boldy (L/R), Wyatt Johnston (C/R), Mitch Marner (C/L/R), Jake Guentzel (L/R), and others.

That isn't a list of good players - that's most of the first four rounds either way. The useful version is narrower: when two men sit level on your board, the one with the second position is worth more than the tie suggests, and nothing on the draft page tells you how much more.

A third position is a fair question, and the answer is that it is worth a little more than a second. 10 men in this pool carry C/L/R rather than two of the three - Cutter Gauthier, Mitch Marner, J.T. Miller, Rickard Rakell among them. Swapping a single-position forward for one of THEM is worth +5.0 games (n=122, sd 8.4) against +2.6 for any multi-eligible man. The reason the third tag adds so little is the UTIL seat: two positions already let a forward take either of two slots plus the flex, and the night where only the third one would have helped is rare.

Lever two: the club's calendar

Clubs don't play on the same nights as each other, and the league calendar is lumpy. A player whose club draws quiet Tuesdays walks into an open lineup. One who draws crowded Saturdays finds every seat filled. Counting games played on nights with 6 league games or fewer:

Build one roster entirely from the eight quietest-calendar clubs and another from the eight busiest. Both own the same number of clubs, so what's being measured is the calendar and not the concentration:

This gap grows with the bench as well, 56 games at 3 against 91 at 5, for the same reason as the first lever. The per-pick column is one seat taken from a quiet-calendar club instead of a busy one.

Where it turns on you

If quiet-calendar clubs are good, stacking them should be better. It isn't:

Concentration costs 41 games. Players from the same club play the same nights and therefore compete for the same seats. Two Avalanche skaters are an edge; six are six men fighting over eleven seats on thirty nights and idle on the rest.

Measured directly, replacing a second player from a club you already own with an equivalent from a club you don't:

And it gets larger the more doubled up you already are.

The second slider does exactly what the idea tells you to do - swap your picks, one at a time, for a comparable Duck or Av. The first swap helps, because both are quiet-calendar clubs. Keep going and the curve rolls over, because by then you're drafting men who play the same nights as men you already have. It peaks at 1 swap at every bench depth, so the trap doesn't get deeper or shallower with roster size. It's just always there, immediately after the first one.

So the rule isn't "draft Ducks". Spread across clubs, and break ties toward the quiet calendar.

How deep a bench is worth carrying

Each extra bench skater brings another 84 games to your roster, and you can still only start eleven a night. What does he actually bank?

An eleven-man roster starts nearly everything it owns and owns very little. The third bench skater banks 64% of his season, the fourth 55%, the fifth 50% - and the half he does bank is the half that lands on quiet nights.

Which is the real case for streaming over hoarding. A bench player's value isn't his projection; it's his projection multiplied by the share of his games you can seat. In a 5-bench league the last man on your roster is collecting about 50% of what his projection says, and that's worth knowing before you hold him through a cold month.

All of it, by bench depth

One table, since it's the question this article gets asked:

Read the column your league actually plays, and read the two kinds of row differently.

End to end, a deeper bench makes every effect bigger. All the whole-roster gaps grow from 3 to 5, because every extra man is another competitor for the same eleven seats and another pick you can spend on a tiebreak.

Per pick, they don't. One swap toward a second position is worth +2.7 games at 3 and +2.4 at 5; one de-clumping swap +2.5 and +2.0. Only the calendar swap genuinely pays more on a deep bench, +3.1 rising to +5.5, because a quiet calendar keeps working for the eleventh and twelfth man on a night when everything else is full.

One number in that table needs explaining: pulling both levers at once returns less than adding them together. 39 plus 75 is 114, and the combined figure is 86. That's expected - both levers attack the same congestion, so once a multi-eligible forward has solved your Tuesday, a quiet-calendar teammate has nothing left to solve. You can't collect both in full.

The practical answer to "does my bench change how I draft" is that a 5-bench league hands you more chances at the same edge, not a bigger edge per chance. Break the same ties. You simply get to break more of them.

What it's worth in production

Extra games accrue at the back of your roster, so price them there. The UTIL and bench skaters in this pool average 0.71 points, 2.08 shots, 0.89 hits and 0.50 blocks a game.

That's the width of the court, not your likely winnings. Neither extreme is draftable - you cannot take only multi-eligible forwards from only quiet clubs and also draft well, and trying costs more in talent than it collects in games. The realistic version is the per-pick column. Break four or five ties across a draft and you collect somewhere around ten to twenty extra games: free, useful, and nowhere near enough to win a league on its own.