Roster Advice
Model → real add/drop/read decisions for a posted roster (composes with Sleeper sync).
Intelligence engines
How we read it
Vacated touches: what the role leaves behind
When a player is OUT, his trailing per-game targets and carries become available volume. We quantify exactly how much opportunity the role vacates.
Beneficiary: who inherits it
From the depth chart and trailing usage, the player most likely to absorb the vacated touches: the waiver predictive core, before the box score confirms it.
Confidence: how clean the read is
High when the depth chart and usage agree on one beneficiary; low when the touches scatter across a committee.
Vacated-role transfers · 2025 week 22
Vacated-role opportunity transfer: when a starter is OUT, the targets + carries his role commanded transfer through shrunk Markov role-state transitions to the next men up. No injuries means no transfer rows. We never fabricate a cascade. Context, not a point projection.
| # | |||||||
|---|---|---|---|---|---|---|---|
No rows in the source window. | |||||||
Vacated-role opportunity transfer: when a starter is OUT, the targets + carries his role commanded transfer through shrunk Markov role-state transitions to the next men up. No injuries means no transfer rows. We never fabricate a cascade. Context, not a point projection.
Data via nflverse (nflverse-data), licensed CC BY 4.0. How we source data
Model → real add/drop/read decisions for a posted roster (composes with Sleeper sync).
Composes the model + xFP + team environment (real schemeFit from neutral-script offensive EPA) + QB-forward passing signal into a real graded pool that drives every fantasy tool when the founder enables it.
ESPN QBR (results) vs Next Gen CPOE (accuracy), triangulated. Disagreement surfaced, not averaged.
RYOE/att vs volume with stacked-box context: bell-cow / buy-low / volume-dependent.
Air-yards & target share → WOPR, with opportunity-vs-production buy/sell.