August is when dynasty podcasts talk projections and ranks as if they were one object. Dynasty Zoltan's public calendar names the month "August: Projections, Rankings, Trade Recommendations and More." Those are four items on one line. They are not one question.
This is a method post for that calendar theme, not a review of his ranks — we do not have them. Historical holdout accuracy and scoring-rule rank movement are separable. The second one can move a player a hundred ranks with zero projection change.
Two questions, one report
The 2026 Fantasy Football Scoring Format Impact Report (engine 1.11.0) already splits the blur:
- Can the underlying projections order players reasonably well? That is a multi-year historical holdout.
- How much can league rules move value after those projections exist? That is a controlled synthetic-pool experiment.
Section 1 is a rank-correlation claim against actual seasons. Section 2 holds projections still and changes a roster dial. Mixing them is how "this player is a projection" becomes "this player is a rank" without anyone noticing the replacement line moved.
Historical holdout: the mean, not the hero year
The holdout mean across four held-out seasons is offense 0.705 and IDP 0.600. That mean is the accuracy claim. The table includes the 2025 holdout; it is not the headline. Players need usable prior history and at least eight games in the target season. The full protocol, market comparison, and failed experiments are in How Accurate Were Our Projections?.
| Target | Role | History | Offense ρ | IDP ρ |
|---|---|---|---|---|
| 2018 | Tune | 2016–2017 | 0.646 | 0.548 |
| 2019 | Holdout | 2016–2018 | 0.694 | 0.600 |
| 2020 | Tune | 2017–2019 | 0.694 | 0.613 |
| 2021 | Holdout | 2018–2020 | 0.664 | 0.596 |
| 2022 | Tune | 2019–2021 | 0.691 | 0.601 |
| 2023 | Holdout | 2020–2022 | 0.689 | 0.585 |
| 2024 | Tune | 2021–2023 | 0.719 | 0.571 |
| 2025 | Holdout | 2022–2024 | 0.771 | 0.618 |
The 2025 holdout is the latest and strongest offense result in this run (0.771). It is useful current detail. It is not the headline.
Limitation: rookies without usable history are not represented like veterans. The holdout population requires prior history, so a first-year player is not in that mean.
Superflex: same points, different ranks
The second question does not need a new projection. Hold scoring and projections still. Add one SUPERFLEX starter. That is a controlled synthetic-pool experiment. The synthetic format results are engine behavior, not 2026 player recommendations.
| Metric | 1QB to superflex result |
|---|---|
| QBs in the comparison | 22 |
| QBs that improved in overall rank | 22 |
| Mean overall-rank gain | 107.09 places |
| Mean VORP gain | 257.5 points |
| Change in projected player points | 0 |
Every quarterback in a 22-player comparison improved overall rank — 22 of 22, mean +107.09 places, projected points change 0. Replacement dropped when the second starter-equivalent slot appeared. Same production, different job.
The write-up is The Superflex QB Premium, Quantified. The public superflex rankings run the same replacement-aware math on a disclosed board.
Across half-PPR, full PPR, TE premium, and IDP-heavy, the same report's largest rank move on a fixed pool was 52 places. Still engine behavior. Still not a 2026 player rec.
The ranking step is the league, not a second model
A projection is an event-rate guess. A ranking is that guess, scored under this league's rules, then ordered by surplus over replacement. Value-based drafting is that second half. Re-projecting the same player for a superflex room is the wrong move if the rates did not change. Rebuilding replacement is the right one.
A connected-league board is that ranking step. It is not a second projection model. It reads the starter slots and the scoring, then ranks the surplus the league actually pays.
Connect the league you actually play
August's calendar is projections, rankings, and trades. Rankings are the scoring-and-roster question. Projections sit still.
Connect your league — free. LeagueFrenzy reads the scoring and the starter slots, then rebuilds replacement for the league on your screen.
Secondary reading: the scoring-format impact report.