A redraft ranking answers what a player is worth this season. A dynasty ranking has to answer something harder: what he returns across every season you hold him, above what the waiver wire hands you for free. Those are different questions, and they often give opposite answers.
Jahmyr Gibbs and Josh Allen project within ten points of each other for 2026, at 347 and 357. By 2033 the model has Gibbs at 14 points and Allen at 157. Same starting line, and one of them is still a starter eight years later. That gap is what a dynasty ranking is supposed to price, and an age penalty bolted onto a redraft list will not find it.
So every player here is projected forward eight seasons on a decline path measured for his own position and age, because the positions age nothing alike. A running back aged 21 to 23 returns 3.5 times his current season over the next eight years. A receiver the same age returns 5.6. A quarterback returns about 5 whether he is 24 or 29.
Value is the discounted sum of future points over replacement: for each of the next eight seasons, what a player is expected to score above the last startable player at his position, discounted 10% a year because a championship now beats one in 2033.
That surplus is an expectation, not a subtraction. Season outcomes are wildly spread: grouped by their prior level, real seasons land with a coefficient of variation of 1.16 at 60 points and 0.45 at 260. Simply subtracting replacement from a projection throws all of that away and hands a zero to everyone below the bar, which on a 210-player board meant 118 players scoring exactly nothing and tying for 93rd. A receiver projected at 150 against a 150 bar is not worth nothing; he beats it about a third of the time. So each season is priced as an option payoff on a distribution rather than a difference of two numbers.
The replacement bar takes its shape from history and its scale from our projections, because neither source is right alone. Historical finishes have the correct shape: a replacement-level season is often a partial one, since the 22nd-best quarterback season usually belongs to somebody who started eight games, and that is genuinely what you fall back on. But those are realised totals, and the bar gets subtracted from projections, which are compressed expectations. Rank for rank against 2019 to 2025, our backs, receivers and tight ends sit at a steady 0.87 to 0.91 of the historical finish all the way down. Quarterback inverts, at 0.89 for the best but 1.16 by QB22, because we carry 29 quarterbacks and every one is a starter, so our pool contains none of the half-seasons that define replacement. The bar is therefore the historical value at each slot, scaled by the compression measured over the range where our pool is complete.
The decline paths are measured directly, not compounded from a yearly rate. The first version of this model fitted a one-year retention ratio and raised it to the k-th power, which broke badly: pooled across all ages and quality levels, quarterback retention came out at 0.75 a year, implying a 320-point passer scores 76 points in year five. Everyone fell below replacement by year four, and extending the horizon from five years to twelve did not move a single rank.
Measured properly, quarterbacks who cleared 200 points at age 24 or younger went 280, 261, 235, 201, 194, 199 over the following five seasons. That is 0.934 a year, not 0.75, and 59% were still 200-point quarterbacks five years on. So each cell here is the observed ratio of a cohort's points in year k to its points today, with a zero recorded for every season a player spent out of the league. Counting those zeros is the point: averaging by age alone is survivorship, because at 31 you only see the backs still good enough to have a job.
Summed over eight years, as a multiple of the current season: a receiver aged 21 to 23 returns 5.6, a quarterback aged 28 to 29 returns 5.3, a quarterback aged 24 to 25 returns 4.7, a back aged 21 to 23 returns 3.5, a back aged 26 to 27 returns 2.0. A young running back is worth less than a 30-year-old quarterback, and that is not a preference, it is the measurement.
The format toggle is one number. A 12-team superflex league starts roughly 1.85 quarterbacks a team, so about 22 are startable rather than 12. That moves the bar a quarterback must clear from our 12th-ranked projection at 262 points to our 22nd at 222, worth about 40 points a season to every startable quarterback and nothing to anyone else. Eleven quarterbacks make the superflex top thirty here; two make the 1QB top thirty.
Rookies do not follow the decline path, because a first-year backup is not a stable base to decay from. Career years one through three follow the measured ascent path for that position and draft round, and those differ enormously: a first-round receiver grows 25% into year two, a third-round tight end nearly doubles by year three, a first-round back grows 0% because backs arrive finished. Those cohorts are small, so each rate is shrunk toward no growth by its own sample size.
On draft capital against college production, which is the usual dynasty argument: for receivers drafted 2021 to 2024, draft position alone explains 43% of the variance in their best season through year three. Target-per-route rate adds a little on top and age at declaration adds a little more, while yards per route run adds nothing once draft capital is in. Round is the anchor, not the college box score.
One more correction worth naming: expected games for 2026 are a one-season injury call, but they were also anchoring every forward year, so a single availability discount got counted eight times. George Kittle's Achilles override alone cost him 31 points of baseline. Forward years now sit at the position's median availability instead, and the decay curves supply attrition on their own because they were measured on real seasons. This turns out to barely matter, which is as it should be: a 2026 injury is not an input to an eight-year valuation.
A handful of quarterbacks carry a named judgement adjustment on the years after this one, because the model has one input per player and then decays it on a cohort curve. That curve knows the average quarterback eventually loses his job; it cannot know which one, and it cannot know that a rate is temporarily depressed. Patrick Mahomes is projected at 17.99 points a game while recovering from a combined ACL and LCL tear, which reads as QB13 and is not who he is in 2028. Kyler Murray is on a one-year deal. Malik Willis is holding a job rather than owning one. Those are adjusted up or down and each one is recorded with its reason.
Worth being straight about: this is judgement, not measurement. I tested whether a track record predicts better forward retention and it does not. Holding age constant, quarterbacks with four or more prior startable seasons returned 0.652 of their current level over the next five years against 0.703 for those with none, and for receivers the effect is significantly negative. There is no measured provenness bonus, so nobody gets one automatically.
What this still cannot see: holdouts, coaching changes and the trade market. It prices football production, not leverage. It prices a cohort, so a genuinely generational player beats his own curve. And it values each player independently, so it cannot know that your third good quarterback is worth far less to you than your first.
The position buttons break this into dynasty RB rankings and dynasty WR rankings on their own, and the age filters isolate the players young enough to still be appreciating. Every column sorts.
| # | Player | Pos | Age | Value | vs redraft |
|---|
In a dynasty league you keep your whole roster from one season to the next. There is no annual redraft of everybody. You draft once when the league starts, then hold, trade and add incoming rookies every year after, so the only new players entering the pool are the rookie class and whoever gets cut.
That changes what a player is worth. In redraft you are buying one season, so a 30-year-old coming off a big year and a 23-year-old projected for the same total are worth roughly the same thing. In dynasty you are buying every season you hold him, and those two are nowhere close.
How does dynasty fantasy football work in practice? Mostly through trades and the rookie draft, which is why it is not the same as a keeper league. Keeper leagues redraft most of the roster every year and let you protect a handful of players. Dynasty keeps everybody, which is why trades matter so much more and why a rebuild can take two seasons rather than two weeks.
What is superflex, first: it is a league that adds a flex slot which a quarterback may fill, so you can start two. In practice a 12-team superflex league starts about 1.85 quarterbacks per team, which means roughly 22 quarterbacks are startable in a given week instead of 12.
That single rule change is worth more than any opinion about the players. It moves the bar a quarterback has to clear from the 12th-best projected passer at 262 points to the 22nd at 222, so every startable quarterback gains about 40 points a season and nobody at any other position moves at all.
Flip the toggle above and watch it happen. Eleven quarterbacks make the superflex top thirty and only two make the 1QB top thirty. Josh Allen goes from third overall to ninth, Drake Maye from eighth to well outside it, Geno Smith from 88th to 121st. Same players, same projections, same ages; only the replacement level moved.
Whether that makes the format better is taste, but it does make it deeper, because the position with the least turnover becomes the position you cannot punt.
It depends on the format, and the gap between them is large enough that a single ranking cannot serve both.
In superflex, quarterbacks take eleven of the top thirty spots on this board. Not because they score the most, but because their production barely decays: a quarterback aged 28 to 29 returns 5.3 times his current season over the next eight years, which is more than a quarterback aged 24 to 25 at 4.7. Age is close to irrelevant for them until 33.
In 1QB, two quarterbacks make the top thirty, and young receivers lead instead. A receiver aged 21 to 23 returns 5.6 times his current season across eight years, the longest tail of any position and age band we measured. Receivers hold their level until about 30 and only then fall away.
Running backs are last in both formats at equivalent age, which is the one thing the two boards agree on. Filter to the dynasty QB rankings or the dynasty TE rankings on their own and the ordering inside each is nearly identical between formats; it is only where they sit against everyone else that moves.
Earlier than feels comfortable, and the numbers are blunter than most people expect.
A startable back aged 21 to 23 returns 3.5 times his current season over the next eight years. By 26 to 27 that has dropped to 2.0, and by 30 to 32 it is 1.6. Compare that with a receiver, who is still at 3.0 at 26 to 27, or a quarterback, who is above 5 at almost any age.
The trap is that the decline in dynasty value starts well before the decline in production. Jahmyr Gibbs projects for 347 points in 2026, more than any other player on this board, and 14 points in 2033. Ashton Jeanty projects for 100 fewer points next season and more than four times as many in 2033. A back can post two near-identical redraft seasons while quietly losing half his dynasty value in between, and the market usually notices a year after the model does.