The Math. No black box.
Exactly how the SFB16 board is built — sources, weights, the value model, and every strategy formula.
The live blend is eight sources, normalized.
Every source in the board blend contributes a declared share. The weight math is simple and visible.
WyattFFB
14.7%
SFB16 projected points
4for4
14.9%
SFB projected points + Bonus Reliance + live draft ADP
FTN
14.0%
public SFB16 overall rank
RotoWire
12.6%
SFB16 overall rank + ADP
War Room Editorial
12.2%
SFB-tuned editorial rank
Fantasy Alarm
11.5%
public SFB16 player rankings table
Fantasy Six Pack
10.5%
SFB16 projected score + ADP
SFB ADP Market
9.6%
average draft position blended from ingested ADP sources
Consensus, but source-aware.
The board does not raw-average ranks from mismatched pools.
- 1. Take each source's rank for the player. Points-ordered sources (WyattFFB, 4for4) enter as a 50/50 blend of their positional-slot mapping and raw points rank — floored at the player's curated (non-points) source median, so a raw QB over-grade can't promote him above what the curated sources believe.
- 2. A source that omits the player still counts, depth-fairly: median(covered) + 40 with 4+ covering sources, + 80 with 3, else poolSize + 25 (capped at 400) with 2 or fewer. Then take the weighted mean of those ranks — the rank itself is never z-scored.
- 3. With 4+ real inputs the composite is clamped to the worst real input + 12 (the envelope), so one outlier source can't drag the consensus off the pack.
- 4. Tier on local score-gap breaks across the top 300 (untiered beyond). Value = ADP − composite rank (positive = the market lets him slide); disagreement is the within-source z spread, used only for the contested / dark-horse signal — never for the rank.
Consensus base, VOR context.
The base board is the trusted consensus rank on a smooth curve; projected-points VOR sizes positional context.
- 1. Blend raw SFB points from Wyatt, 4for4, and Fantasy Six Pack at 37% / 37% / 26% when covered.
- 2. Set positional replacement by projected points: QB20, RB36, WR54, TE13.
- 3. VOR = projected points − positional replacement; negative values are allowed below replacement.
- 4. Base value = 600·e^-((consensus rank − 1)/60), so the base sort exactly matches consensus.
On a player page, the per-source ranks for the points-based sources (WyattFFB, 4for4, Fantasy Six Pack) are shown as dRank — each source's covered players re-ordered by projected points above positional replacement (QB20 / RB36 / WR54 / TE13), i.e. a within-position VOR rank, not that source's raw points rank. Those chips are tagged "(VOR)"; hover one for the raw rank.
What the model does on real players.
These values come from the generated board JSON used by the app.
| Player | Value | Board slot | Read |
|---|---|---|---|
| Puka Nacua | 600.0 | #1 · WR1 | consensus curve top slot |
| Josh Allen | 552.0 | #6 · QB1 | elite QB held by consensus, not raw points |
| Bowers / McBride | 533.9 / 542.9 | #8 · TE2 · #7 · TE1 | TE-premium elite tier |
| Colston Loveland | 415.8 | #23 · TE3 | later TE follows consensus unless a lens compresses TE |
A strategy lens moves players only through its demand schedule and strategy-VBD blend, inside hard depth caps — the consensus baseline itself never changes.
Every badge on the board, defined.
The board layers quick-read signals over the rankings. Each one is a fixed, public rule on the same data — nothing editorial.
The 8-source blend disagrees hard in a player's favor: at least 4 sources rank him 24+ slots above his consensus rank, or a single source has him 80+ slots higher. Hover the badge for the exact split. It marks upside the market average is hiding — not a recommendation by itself.
The Val column's color encodes ADP − our rank: green = the market drafts him later than we rank him (discount), red = earlier (reach). An amber tint layers on when source disagreement exceeds 0.45 — read those numbers with more caution.
The sticky position pills count players still available (not drafted, not yours) at each position; with a position selected, the tier note shows the shallowest open tier's remaining supply — the scarcity read behind position runs.
The War Room warns when three or more of your players share a bye week — weighted worse when they cluster at the same position or at QB/TE, where SFB16 lineups are least flexible.
A draft plan plus VBD, bounded in rank space.
Each lens is a per-position demand schedule (how a real practitioner spends picks across the draft) blended with strategy-shifted VBD on the SFB16 scoring, then capped so promotions shrink with depth and the 200+ tail stays on consensus.
The control board: the blended SFB16 consensus rank on the shared value curve. It is a literal passthrough so every strategy lens compares against one stable baseline.
Suppresses early RB demand, front-loads WR/TE and superflex-safe QB value, then reopens RB demand in the pick 60-140 harvest window with VBD and depth caps keeping the board disciplined.
Keeps one elite RB anchor, pivots into WR/TE/QB demand, and reintroduces RB2/RB3 value after the opening instead of blindly burying every quality back.
Raises demand for the scarce TE-premium tier and uses deeper TE replacement in VBD, while risk and depth caps keep thin TE tails from flooding premium picks.
Elevates WR demand through the top 120 and rewards receiving-bonus VBD, while elite QB/RB/TE profiles remain viable and deep WR tails stay capped.
Delays the early QB tax without punting superflex safety, then clusters discounted pocket and volume passers into the QB value pocket.
Blends ADP discounts with shifted VBD and risk gates, so market fallers rise only when projection quality and data coverage support the discount.
Rewards spike-week paths with real bonus-reliance data, especially WR/RB/TE explosive profiles, and penalizes null bonus data instead of treating it as hidden upside.
Rewards stable weekly scoring through QB/TE demand, low bonus reliance, positive VBD, and agreement across sources, while volatile spike-only profiles slide.