First version · uncalibratedProjected ownership is the share of simulated field lineups containing a player. It is a rough behavioral estimate. No historical contest results have been fitted.
All eligible players need a DraftKings FP projection, salary, position, team, and opponent. Mark each goalie’s Starter field as 1 or 0. Line / PP role and implied team totals are optional. Partial samples cannot produce meaningful ownership.
How the estimate works
The model favors projected points, salary value, PP1 roles, and shared forward lines. It samples nine-player lineups under $50,000: 2 C, 3 W, 2 D, 1 G, and 1 skater UTIL. A three-team minimum and the salary floor are construction assumptions. Profiles change the emphasis on projections and stacking; they are not fitted contest types. Goalies marked as nonstarters and out / IR players are excluded. DTD players remain eligible. Opposition to a selected goalie receives a soft penalty.
6,000 sampled lineup states from 40 chains are reproducible for the same inputs and settings. They are not 6,000 independent observed entries. Changing field assumptions shows model sensitivity, not a statistical confidence interval. Import new news / projections and rerun before using a new slate.
Premium feeds and validation
Import RG, Stokastic, or other permitted exports and choose Imported estimates to compare. Both sports open with a 50% model / 50% imported-ownership blend. Players without a RotoWire ownership row use their model estimate; imported-only view leaves missing values blank. Other source imports require both values for intermediate blends. Model weight is adjustable; it is not fitted to contest outcomes. Actual contest ownership is separate and never enters the forecast. The built-in snapshots contain the full published provider pool; the official contest roster remains unverified. Live news refresh is not connected. A matched-player error report appears after you import it.
DraftKings point scoring
RotoWire NHL optimizer FPTS supplies the published DraftKings projection when available. Unmatched players retain the earlier stat conversion. NHL ceiling and floor exports describe individual stat ranges; they appear in player details and do not enter the central FP or ownership blend. They are not joint fantasy-point percentiles. Earlier NHL stat exports are converted to DraftKings FP. Threshold bonuses use Poisson count approximations; reported goalie win, shutout and OT-loss expectations are retained. Unreported shootout goals and goalie scoring stats are assumed zero. These are uncalibrated assumptions, not RotoWire bonus probabilities. Our FP blends available forecasts (90%) with recent fantasy-point averages (10%), adjusted slightly for role and implied team total. NFL usage can adjust only the recent-form component by at most ±3%, using offensive snap share, recent target changes, red-zone share and TE routes / plays. Its ownership model uses the same small usage signal. These weights are hand-set and uncalibrated; season stat totals are never treated as another weekly forecast. Receiver alignment and defensive roster totals are context only. Zero forecast roles stay zero. Out / IR players and nonstarting NHL goalies receive zero. News overrides apply availability and usage factors. NFL DNP / limited practice uses a hand-set 0.97 / 0.99 factor, not a predicted playing probability. Full practice uses 1.00. No replacement-player usage is invented. These weights have not been calibrated. You can also import expected scoring stats using My model. Bonus inputs are probabilities from 0 to 1; they are required because a bonus cannot be inferred by applying a threshold to a projected average.