Import stats accepts Sportradar NFL JSON (Seasonal/Game Statistics or Boxscore feeds) or a CSV of player stats, and recomputes projections from it. Pull the feed with your own API key from developer.sportradar.com. CSV columns recognized: player, games, pass_yd, pass_td, int, rush_yd, rush_td, rec, rec_yd, rec_td, fum_lost (one row per player-season, or include a week column for per-game ceiling/floor).
Defaults are computed from each player's 2023–2025 regular-season box scores (weeks 1–17; preseason, Week 18, and playoffs excluded), then adjusted for 2026: each team's output is scaled by its Vegas implied team total (a low-scoring offense can't have everyone booming), and intra-team share is tilted toward 2026 roles by ADP (so new additions take share — e.g., Davante Adams behind Nacua). Average uses ≥30%-usage games, floor uses clear-starter games, ceiling is the single best game — never mixing positions or other players' data. Players tagged RANK are rookies/no-data (ADP + position ranking). All fields are editable; Wk σ is per-week standard deviation from actual game-to-game variance.
#
Player
Pos
Team
Src
Berry
UD ADP
DK ADP
Proj pts ✎
Floor ✎
Ceiling ✎
Wk σ ✎
Roster
Per-team target and rushing allocation. Targets = WR + TE + RB target share; Rushing = RB + QB rush share. The projection rule: each team leaves ≤10% unallocated on both. PASS = within rule, FAIL = exceeds 10% unallocated.
Team
WR tgt %
TE tgt %
RB tgt %
Tgt alloc %
Tgt gap
RB rush %
QB rush %
Rush alloc %
Rush gap
Rule
Draft settings
Every pick is scored by one engine: each player's floor, projection and
ceiling for all 17 weeks, matchup- and offensive-line-adjusted, converted
into the points he is actually expected to contribute to a best-ball
lineup. There are no build presets to choose between — the engine takes
the best available player for the roster you already have.
How it works
Simulates a full snake draft. The field drafts by the selected site's ADP (with positional sanity caps).
At your picks, the engine takes the most optimal available player — maximizing value over
replacement given best-ball roster construction (so it won't over-draft QB/TE early). It tracks the board closely early — taking value that falls to you while avoiding reaches — and progressively opens up later for darts and stack completion. The result is the
theoretically best roster that could realistically fall to your slot on that site.
What the projections are built from. Three seasons of real workload — target share, carry
share and snap rate — rather than name recognition. Target share is weighted more heavily than snap counts, because being on the
field isn't the same as getting the ball. Team pace is applied on top as an opportunity
multiplier and is treated as a property of the coaching staff rather than the franchise — so a
new head coach moves his team's outlook immediately rather than a year later.
Weekly outcomes are lopsided, and modeled that way. Most weeks are ordinary and a few are
enormous, because touchdowns arrive in bunches. Each player's range of weekly outcomes is shaped to his own scoring
profile — quarterbacks are comparatively steady, pass catchers carry the longest tails, and workhorse
backs are steadier than boom-or-bust wideouts. Weeks are matchup-adjusted, and a soft matchup widens a
player's whole range rather than just nudging his average.
Ceilings and floors mean something. Ceilings are held to what the position has actually produced in
recent seasons, so a ceiling means a realistic great year rather than a number no one has ever hit. Floors
are drawn from large-scale season simulation and measure the bad-but-healthy outcome rather than a generic
injury penalty, so they separate steady players from volatile ones. Total proj is not a sum
of your roster — it simulates the season week by week and counts only your best starting lineup each week
across the tournament regular season, with byes scored as zero.
Underdog uses half-PPR projections; DraftKings uses full-PPR — and DraftKings additionally scores kick and
punt return touchdowns, which are modeled for every RB and WR. Switch sites to compare optimal builds.
ZR Zero RB: WR/TE early, RB mid-to-late (still grabs value).
RR Robust RB: strong early RB core.
HF Hyper-Fragile: ~4 RBs through round 8, then leans WR.
BT Bully TE: ~2 premium TEs early to win the position.
PS Pissing (Robust WR): 5–6 WRs through the first 8 rounds.
My exposure
Underweight values by round value vs ADP · your %
Three recommended lineups —
Built for the BBM / Milly structure: cumulative points advance you through Weeks 1–14 (top 2 of 12), then Weeks 15, 16, and 17 are single-week eliminations — with most of the prize pool paid on Week 17 alone. Each lineup blends advance value with matchup-adjusted single-week ceilings (Week 17 weighted heaviest). Correlated picks are valued across all three gate weeks, not just Week 17, and each is judged by the combined implied total of that specific matchup — the ceiling of a stack is set by the game, not one side of it — with bring-backs from the opposing team counted at just over half a same-team pairing. The three lineups intentionally use different QB rooms, since single weeks are too high-variance to concentrate a portfolio on one QB.
ScoringShow top
Weekly points start from each player's projected per-game pace, then adjust for that week's opponent using Defense-vs-Position — fantasy points each defense allowed to QB/RB/WR/TE over the 2025 season (computed from play-by-play), adjusted for star-defender availability — each unit is graded on games its impact defenders (CB1s, elite pass rushers, run-stuffers) actually played, so one injury-wrecked stretch doesn't mislabel a defense. Hover an opponent to also see how much softer the matchup gets if their key defender sits. A ▲ marks a favorable matchup (defense gives up more to that position), ▼ a tough one; hover any opponent or point total for the exact adjustment. 2026 byes remove players in their bye week, and season-ending-injury players are excluded. Toggle Matchup adjust off to see raw per-game pace.