How the 2026 Rankings Were Built

Most rankings you'll read this week were made the same way: someone who watches a lot of football sat down with last year's numbers and a feel for the offseason, and ordered players until the list looked right. That's not a knock. It's how I did it for twelve years in my own league, and it works about as well as anything.

This year I wanted to be able to show my work. Not because the numbers know better than a good eye — they don't — but because when a player moves nine spots in October, I want to be able to point at the reason. So the 2026 preseason rankings were built in a specific order, and every step of it is below. If you only read one section, read the last one, because that's where the rankings stopped being a spreadsheet.

Step one: what every team actually did last year

Everything starts with the team, not the player. Before I thought about a single running back, I pulled every offense's 2025 totals: how many times they ran, how many times they threw, how many yards and touchdowns came from each.

Those totals are the ceiling. An offense that ran the ball 450 times last year is not going to run it 600 times this year because you like their new back. Starting from the team keeps you honest about that. It also catches things you'd otherwise miss — a receiver on a team that barely passed can look fine by his own stats and terrible once you see how small the pie was.

One adjustment matters here: last year's numbers get pulled partway toward the league average before I use them. A team that threw for the most yards in the league in 2025 will probably throw for a lot in 2026, but probably not the most. The same goes in the other direction. The pull is stronger for touchdowns than for yards, because touchdowns swing wildly year to year and yards mostly don't.

Step two: what Vegas thinks happens next

Last year's totals tell you what an offense was. They don't tell you whether it's about to get better or worse. For that I used the one number the betting market publishes for every team before the season: the win total.

The logic is simple. Teams that win more games score more, and teams that score more produce more fantasy points across the whole roster. So every team's projected touchdowns get nudged up or down based on how far its win total sits from the middle of the league. A team the books have at 11.5 wins gets a bump. A team at 4.5 gets a haircut. Yards move too, but only slightly — good teams don't gain dramatically more yards than bad ones, they just finish drives.

I don't treat Vegas as gospel. I disagreed with a couple of the lines and changed them, and those changes are noted in the model. But as a starting point for "which offenses are the environment you want your players in," it beats any single person's guess, including mine.

Step three: deciding who gets the ball

This is the part no formula can do, and it's where most of the preseason work went.

Once you know a team is projected for, say, 460 carries and 550 targets, you still have to decide how those get divided. For a returning starter who didn't change teams, last year's share is a reasonable starting point and I mostly left those alone. For everyone else — the 73 players who changed teams and the 42 rookies who've never taken an NFL snap — there's no last year to lean on. Those got a share by hand, one at a time.

That's about 115 decisions, and they're the ones that actually shape the rankings. Whether Kenneth Walker gets 55 or 62 percent of Kansas City's carries matters more than any metric I could apply to him afterward. Some of these I feel good about. Some I'll be wrong about by Week 3, and when I am, the share changes and the rank moves, and the change column will show it.

Every player's share and the team total behind it sits in the model. When I say a player projects for 1,150 rushing yards, that number is a share times a total, and both are visible to me. Nothing is typed in directly.

Step four: the metrics

I built four of them. You'll see them referenced in the weekly write-ups, so here's what each one does in plain terms.

Opportunity Trend tracks whether a player's share of his team's touches is growing or shrinking over the last month. It's the one that catches a role change before the box score does.

Role Clarity is for running backs only. It measures not how much a back is used, but where — early downs, third downs, the red zone, the two-minute drill. A back who owns all four is a different asset from one who only sees early downs, even if their carry counts match.

Scheme Fit compares a receiver's profile to the offense he's in. It matters most right after a trade or a coordinator change, which is exactly when everyone's guessing.

Passing Game Health is for quarterbacks. Is the line holding up, are the weapons on the field, is the team playing fast and throwing more than the score suggests it should.

Here's the thing about the preseason, though: most of these are quiet right now. Opportunity Trend needs a month of 2026 games before it means anything. Scheme Fit is switched off until I'm happier with it. Role Clarity and Passing Game Health are running on last year's data at half strength. So the preseason baseline is really a test of steps one through three. The metrics start pulling their weight from Week 3, and I'll say so when they do.

What I'll tell you about how they're weighted: each one adjusts a specific stat — carries, touchdowns, yards per target — and never the rank directly. What I won't tell you is the exact weight on each. The inputs are public. The weights are not.

Step five: the board

By the end of step four, every player has a projected stat line and a points-per-game number. That is not a ranking yet, for two reasons.

The first is that points per game across positions is misleading. Quarterbacks score the most, so a raw list would be twenty quarterbacks and then everyone else. The fix is the same one experienced drafters use by instinct: measure each player against the best option you could get for free at his position. A running back who scores four points a game more than the last starting-caliber back is worth more than a quarterback who scores four more than the twelfth quarterback. The model does that math, and it's why the top of the board looks like a draft board and not a scoring leaderboard.

The second reason is that the model doesn't watch film, doesn't know a coach's tendencies, and doesn't know that a player looked different in August. I do, sometimes. So the final step is a board I moved by hand. Every player started where the model put him. I dragged the ones I disagreed with, drew the tier lines where I felt the cliffs were, and started writing a note on everyone I moved more than a few spots. Not all of those notes are in yet — the biggest moves have them, the rest land over the first week. Where the note is missing, you'll see a dash, not a guess.

On the rankings page, you'll see both numbers: where the model had him and where I put him. When they match, I trusted the projection. When they don't, I overrode it, and the note says why. That column will be more interesting in November than it is now, but it's there from day one because the point of this site is that you can check.

What happens next

Next Wednesday the first real change column appears. Every player will have a number next to his name showing how far he moved from this list and a reason if it's more than a few spots. This preseason file never changes — it's the thing the rest of the season gets measured against.

I'll be wrong about some of this. Probably the shares more than anything else. When I am, you'll see it here first, because the alternative — quietly fixing the number and pretending — is the thing this whole site exists to not do.

The rankings are here. The one rookie I'm staking the year on is here.