EditorialsRoster construction
What are the biggest settings where every pick still matters?
A 12-team league runs out of players who start somewhere in the NFL at pick 114, which is a roster of about 9. The position that runs out is running back, not quarterback: 34 running backs play a majority of their club's offensive snaps, and a 12-team draft has taken 51 of them by the end of a 15-man roster. Every published platform limit is far above this, because a platform limit is an engineering choice and this is a fact about the NFL.
Every figure on this page is HexValue, the cross-position number HexMetrics assigns a player once it knows a league's shape. Two boards are built from the same 553-player pool for the 2026 season, identical except for the one setting under discussion, and the difference is what you see. Both are 12-team leagues. Percentages are suppressed for players below a HexValue of 5, because a ratio against a player worth almost nothing reports the size of the denominator rather than the size of the change.
The supply is set by the NFL, and nobody publishes it
ESPN allows 30 roster slots, MyFantasyLeague 90, Sleeper states no cap at all. None of those is a claim that a 90-man roster contains 90 players worth rostering. They are limits on what the software will store, and the platforms have neither the means nor the reason to say anything more.
The number that decides whether a pick means anything is a property of the league being played rather than the one storing it. There are 32 NFL clubs. Each fields one quarterback at a time, one back in most formations, three receivers and a tight end, and carries exactly one kicker and one defence. That is the whole supply, and it does not change when a commissioner adds a roster spot.
Counting it two ways, which is why the number is trustworthy
A player counts here if he is on the field for at least 50% of his club's offensive snaps. That is a description of a depth chart rather than a fantasy opinion: below it a player is a rotational piece or a handcuff, and drafting him is buying a lottery ticket rather than a starter.
The line is round and it was not chosen to produce a tidy answer, which is checkable, because a completely independent count lands in the same place. Eleven personnel across 32 clubs puts 192 skill players on the field league wide. The snap count says 212. Two methods sharing no inputs, 20 apart.
| Position | Players who start somewhere |
|---|---|
| Quarterback | 44 |
| Running back | 34 |
| Wide receiver | 94 |
| Tight end | 40 |
| Kicker | 32 |
| Defence | 32 |
| Total | 276 |
Running back runs out first, and it is not close
The total says 276 picks, which would be a 12-team league with a 23-man roster. The total is not what binds. Demand is measured here from average draft position across the public sources rather than assumed, so these are the players a real draft actually takes, in the order it takes them.
| Roster | Picks | QB | RB | WR | TE |
|---|---|---|---|---|---|
| 12 | 144 | 17 / 44 | 41 / 34 | 55 / 94 | 18 / 40 |
| 15 | 180 | 21 / 44 | 51 / 34 | 62 / 94 | 22 / 40 |
| 18 | 216 | 27 / 44 | 61 / 34 | 75 / 94 | 26 / 40 |
Quarterback, receiver and tight end are comfortable at every depth. Running back is over supply at all three, including a shallow twelve-man roster. In a standard 15-man league 51 backs come off the board and 34 of them play a majority of snaps, so roughly a third of the running backs drafted in an ordinary league start nowhere.
That is the committee era stated as a number. It is also the reason the position feels thin in a way the other three do not: the thinness is real, and it is not about talent.
The ceiling, and what it says about the leagues people actually play
Walking the same board until any position exhausts its supply gives pick 114. Every selection before it is a player who starts somewhere in the NFL. After it, the draft is selling depth that does not exist.
| Roster size | Teams |
|---|---|
| 10 | 11 |
| 12 | 9 |
| 14 | 8 |
| 15 | 7 |
| 16 | 7 |
| 18 | 6 |
| 20 | 5 |
A standard twelve-team league with a fifteen-man roster is therefore about two thirds larger than the point at which every pick still matters. It crosses the line in the tenth round.
That is not an argument that standard leagues are broken. It is an explanation of something everybody already experiences: the waiver wire matters, streaming works, and the last rounds of a draft are lottery tickets. Those are not failures of preparation. They are what happens on the far side of pick 114.
How much of this rests on the 50% line
More than any other number here, and pretending otherwise would be the dishonest way to publish this. Running back supply moves further across the threshold range than any other position, so the ceiling moves with it.
| Snap share | Backs available | Ceiling in picks | 12-team roster |
|---|---|---|---|
| 25% | 67 | 247 | 20 |
| 40% | 47 | 163 | 13 |
| 50% | 34 | 114 | 9 |
| 60% | 15 | 32 | 2 |
So the answer is a range rather than a number. A standard twelve-team, fifteen-man league sits inside the loosest reading and outside the strictest. Where you land depends on how many committee backfields you are willing to call starters, and that is a judgement rather than a measurement.
What does not move across that range is the ordering. Running back binds first at every threshold, and the other three positions never bind at all.
What this means for a draft this season
Pick 114 is the point at which a 12-team league has filled about 9 places on every roster. Every pick after it is being spent on a player who holds no starting job in the NFL. That is a fact about the supply of football players rather than about anyone's draft strategy, and it holds however well the room drafts.
The first consequence is an ordering one. Running back exhausts at 34 and receiver at 94, so two positions on the same board run out at points nearly three times apart. The last back who plays a majority of his club's snaps is gone long before the equivalent receiver, and a board that treats the two as interchangeable at similar projected points will take the receiver too early and find no back left. Where two players sit close together, the running back is the one whose replacement does not exist.
The second is about what late rounds are for. Once the pool of starters is spent, a pick can no longer buy a starter, so drafting for need past that point is shopping for something the market does not stock. What those picks can buy is the backup to a starter already on the roster, or a player whose snap share might rise. Both are bets on a job opening rather than on a job existing, and the honest way to hold them is as bets.
The third is a caution about the line. At a 25% snap threshold the ceiling is 247 picks and the tenth round is unremarkable; at 60% it is 32 and the fourth round has already passed it. The round number is therefore a fact about where "matters" is drawn rather than about football. What survives every line is the ordering: running back binds at all four thresholds and no other position binds at any of them.
For the record, the running back figure is the one my own priors most wanted to be true, which is a reason to be careful with it rather than pleased about it. 34 backs above a half-snap workload across 32 clubs is the committee, counted rather than complained about.
Where HexMetrics stops, and why
HexMetrics caps league size at 32 teams, which is above everything on this page and below MyFantasyLeague's 100. The reason is the same kind of supply limit seen from the model's side rather than the NFL's: past a certain size there are not enough priced players left for a last starter to be worth anything, so the board would render and mean nothing. The full working is in the roster-size piece.
Both limits are limits of the data. One is the NFL's supply of starters, the other is the model's supply of priced players, and neither is a decision about what the software could store.