WS-25 · The Divergence Note · 2026 Draft Season

Where the board and the evidence disagree

Your top-100 is the market's best guess. This is where thirteen seasons of tested evidence say the guess is wrong. Bring whatever rankings you trust. This note does not replace them. It marks the three places the draft market and the evidence pull in opposite directions, so you know which of your own rankings to trust hardest and which to hold loosely.

Last summer we ran every plausible way to forecast a fantasy season a year out against thirteen seasons of results. Two things survived the tests: what a player produced the year before, and how good his offense is. That is the whole list. The signals the market falls in love with every July, a splashy new address, a projected target share, an offseason chemistry story, did not survive.

The draft board is built largely on those signals that failed. That is not a knock on the people making the boards; it is what happens when a market prices hope. It is also where the value hides. Below are the three patterns, each with named examples from the current board, and the one thing to do about each from your seat.

1. The market pays last year's price for this year's coin flip

Pattern: the relocated-star tax

NEW SITUATION When a proven player changes teams, almost nothing you can know about the new fit tells you what happens next. We tested it directly: line up everything about a relocated player, how good he was, how good the new offense is, how the old pairing produced, and the best single clue explains about three percent of the result. Knowing everything about the ingredients tells you almost nothing about the dish. Yet the board keeps pricing these players near what they did at their old address.

Kenneth Walker III, now in Kansas City, going in the second round. New scheme, new backfield, a workload nobody has seen yet. A proven back dropped into a brand-new system at a second-round price is exactly the coin flip the evidence warns against paying full freight for.

A.J. Brown, now in New England, priced as a top-twenty receiver. This one is a two-sided flag. The Patriots offense graded well last year, and offense quality is one of the two signals that actually predicts, so that part is real. But Brown and Drake Maye have never played a competitive snap together, and a first-year pairing is precisely what our test measured as close to unforecastable. You are paying a proven receiver's price for an unproven duo.

What to do from your seat: none of these are avoid. They are do not pay the certainty price. Let the room pay July prices for coin flips. If one of them slides a round or two past his ADP, that discount is the moment the uncertainty turns into an edge instead of a tax.

Why: across thirteen seasons, a relocated player's new-situation details explained about three percent of what he did next, the weakest link in the study. See the study.

CROSS-REFERENCE: THE EXCEPTION THAT PROVES IT The tax is about the price, not the move. D.J. Moore also changed teams, to Buffalo, but he is going near the fifth-round turn, where the discount already pays you for the uncertainty, and Buffalo finished as a top-twelve offense, which is the signal that survives. Same kind of move, priced honestly. That is the version you take.

2. The market pays for the storyline, then acts surprised

Pattern: the narrative premium

Two of the market's favorite reasons to reach, a projected target share and an offseason breakout story, are the two reasons that failed our tests outright. Target share flunked. So did how in sync a pairing looked in the spring. The board reaches on both every year, and nowhere harder than on players with no track record at all to anchor the price.

The rookie running back going in the third round. Jeremiyah Love was a top-of-the-draft rookie this spring and the board already has him inside the first three rounds. A rookie is the purest form of paying for projection over production, because there is no production yet to test at all. The talent may be real. The price assumes a floor no rookie has earned.

The proven talent on a stalled offense. Garrett Wilson is going in the third on receiver talent alone, while attached to one of the least efficient offenses in football last year. The one signal that survives our tests, offense quality, is the one flashing red on him. Talent that good often finds a way, but you are paying up while the one measurable signal points the other direction.

The camp-hype risers. The names climbing boards twenty spots this month on training-camp momentum are the target-share story unfolding in real time. Some will hit. The point is that the reason they are rising, a projected role, is the reason our tests said to hold loosely, not the reason to reach.

STRONG OFFENSE The corrective is the signal that did survive. When two players feel like a coin flip, weight the one attached to the better offense, not the one with the louder summer. A mid-round receiver on a top-eight offense beat the same-priced receiver on a bad one, persistently, in our data.

Why: prior production and offense quality were the only predictors that survived thirteen seasons of testing. Target share and pairing chemistry did not. See the study.

3. The production the market forgot is where the board is soft

Pattern: the discounted incumbent

STRONG OFFENSE Flip the two winning signals around and you get a buy list the market keeps underpricing: players who already produced, on offenses that are already good, going below their evidence profile because they are not this summer's story. Both predictors green, and a discount anyway. That is the cleanest value the study points to.

Kyren Williams, the lead back on one of the two best offenses in football, going in the third. The Rams finished at the very top of the league in offensive efficiency last year and return the core of it. The knock is the usual workload questions, but they are workload questions on a top-two offense, priced like an afterthought.

Josh Jacobs on a top-five offense, with an ADP that slides into the fourth in some rooms. Established production and a stable, efficient offense in Green Bay, at a price set by the rookies and new-address names drafted around him rather than by what he has actually done.

Jameson Williams on a top-ten offense, going around the fifth. The market cooled on him over a coaching-staff change, which is a narrative, not a collapse in production or offense quality. That is the market discounting a survivor over a story.

Why: last year's output was the single strongest predictor we tested, and offense quality was second. Both point the same direction on these names. See the study.

How to run this at your desk

You do not need to agree with every name to use the note. Use the three patterns as a filter on your own board. NEW SITUATION When your rankings have a relocated star inside the first five rounds, hold that ranking loosely and let him come to you. STRONG OFFENSE When you are torn between two similar players, break the tie toward the better offense every time. And when the board is reaching for a summer story ahead of a proven producer on a great offense, that gap is your pick.

One more, for the formats that reward ceilings. STACK WINDOW Pairing a quarterback with his own top receiver added roughly one and a half points of weekly swing in our study, in both directions, and it worked for close to nine of every ten pairs we measured. Favor pairs with a real track record of producing together over pairs the market likes on this summer's hype. Same lesson as the rest of the note: last year's output travels, the storyline does not.

The offense-quality reference

2025 offensive EPA per play · all 32 teams

This is the tiebreaker, made concrete. The number below is offensive EPA per play in neutral game situations: the points a team added, on average, on a neutral-script offensive snap last season. Higher is better. When two players sit close together on your board, this is the column that breaks the tie, lean to the one whose offense sits higher here. Green is the top eight, red the bottom eight.

Rk	Team	EPA/play
1	Patriots	+0.20
2	Rams	+0.16
3	Cowboys	+0.16
4	Colts	+0.15
5	Packers	+0.13
6	Eagles	+0.12
7	Cardinals	+0.10
8	Dolphins	+0.10
9	Jaguars	+0.09
10	Lions	+0.09
11	Bills	+0.09
12	Seahawks	+0.08
13	49ers	+0.07
14	Bengals	+0.05
15	Broncos	+0.04
16	Ravens	+0.03
17	Falcons	+0.03
18	Bears	+0.02
19	Steelers	+0.02
20	Buccaneers	+0.01
21	Chiefs	+0.01
22	Giants	+0.01
23	Commanders	0.00
24	Texans	0.00
25	Chargers	-0.08
26	Panthers	-0.09
27	Browns	-0.13
28	Titans	-0.16
29	Vikings	-0.17
30	Saints	-0.23
31	Jets	-0.23
32	Raiders	-0.29

Computed by Fanvatic from nflverse 2025 regular-season play-by-play using the same neutral-situation filter as the WS-24 stack study (win probability 0.20 to 0.80, quarters 1 to 3, pass and run). That filter isolates offense quality from garbage time, so this ranking can differ from all-plays EPA published elsewhere. Values are rounded to a hundredth of a point, so teams that close, the Rams and Cowboys at plus point-one-six, for instance, are effectively tied.

CROSS-REFERENCE: FANVATIC DRAFT DESK These three patterns are wired into the Fanvatic Draft Desk, the free round-by-round strategy planner. Enter your league settings and the plan flags the relocated-star tax, the offense-quality tiebreak, and the stack window round by round, from your exact draft slot.

Rankings tell you who. This tells you when, and why.

The Divergence Note is the evidence engine behind the Draft Desk, applied to this year's board. Build your plan from your own rankings and let the research tell you where the market is guessing.

Build your draft plan

The Divergence Note reads the public draft market, Average Draft Position from open consensus sources, against Fanvatic's thirteen-season predictor study of publicly available production and efficiency data. It names players to illustrate patterns in how the market prices evidence; it is not a set of scored predictions and it is not draft advice for any individual pick. Research and education only. No picks.