1. Slate setup
V8.4 Automated Data Research · Salary feed pending
Checking published schedule…
Selecting a matchup does not create salaries or projections. Import the correct DraftKings contest CSV first; filtering only removes players from other games. Team abbreviations in the CSV must match the schedule.
CSV minimum headers: name,team,position,salary. Original DraftKings Showdown lineup-template CSV is also supported. No projection, ceiling, or ownership columns are required. Salary is FLEX/base salary in both modes. In Showdown, captain costs 1.5× and earns 1.5× points. Classic salary cap $50,000; Showdown cap $50,000.
V8.1 · Availability and projection diagnostics
Optional: import a manually verified availability CSV with columns name,team,status,opportunity. Status accepts available, questionable, out, unknown. This is not a live feed; confirm the source and timestamp yourself. Unmatched records are never applied to another player.
No availability file imported.
V8.1 · Verified usage review (optional)
Import independently checked usage information. Columns: name,team,snaps,routes,targets,carries,opportunity. Counts are recorded for audit only; opportunity is a manually justified percentage (0–100) applied to the projection. The app does NOT infer playing time from raw counts, since they need game and team context. Do not enter future-game statistics.
No verified usage data imported. Historical fantasy-point averages are not a substitute for playing-time data.
Missing projections and manual overrides
Players with no reliable model data remain excluded. You may import projections from a source you trust using name,team,projection,ceiling. This does not create or verify NFL statistics. Both numbers are required and must be nonnegative; ceiling must be at least projection. Overrides apply to this slate only.
No projection overrides imported.
Show players missing projections and reasons
Diagnostics appear after applying current-season estimates.
Lineup audit will appear after importing players or generating lineups.
V8 · Pre-entry quality gate
Historical testing found substantial projection errors and unreliable ranking. The model has no verified live availability, usage or ownership feed. A lineup with unreviewed players is a research scenario, not a vetted contest recommendation.
Import salaries and projections to see the review gate.
No slate loaded.
Projection model: sample-size-adjusted recent-form blend with historical baseline; ceiling from 2026 player-game upper-tail distribution, with limited-sample shrinkage and opponent adjustment. Offense is PPR-based; K/DST use play-by-play scoring estimates (DST points allowed approximated); manual availability/opportunity adjustments; no live injury or ownership feed.
2. Player pool
V7.0: Status and opportunity are manual inputs, not live verified reports. Choose Out to exclude a player, or set expected opportunity below 100% to reduce projection and ceiling. Unreviewed players remain eligible unless you turn on strict review below. Confirm official inactives before submitting.
| Player | Team | Pos | Salary | Projection | Ceiling | Own % | Status (manual) | Opportunity % | Confidence | Role review | Active |
|---|
Confidence measures historical sample/data completeness only. It does not establish that a player is starting or active. Role review is manual and is not an injury feed.
3. Optimizer results
Cash ranks projected points with a volatility penalty. GPP uses a revised experimental ranking emphasizing P90, simulation mean, and projection instead of noisy P99; includes Captain diversification; ownership leverage is used only when real ownership inputs are provided. These are relative heuristic rankings, not calibrated cash or first-place probabilities.
Monte Carlo estimates use shared team/game factors and position-sensitive outcomes; these are illustrative, not calibrated win probabilities. Rankings are heuristic and do not promise profit.
Best alternative for each Captain
Independent Captain comparisons, ranked by the same GPP score. These alternatives do not use lineup exposure limits.
4. Scoring checks and retrospective comparison (V8.0)
Built-in tests cover offensive, kicker and defense scoring examples. These check our calculation functions, not the accuracy of historical K/DST estimates or the current official contest rules; confirm scoring against the DraftKings contest rules before entry. The comparison CSV is diagnostic only: a true backtest requires projections saved before kickoff, not today's projections compared against past games.
Scoring checks have not run.
No actual-score comparison imported.
Small-sample caution: Players with fewer than four modeled games are flagged below. Confirm their role and availability before trusting lineups; this model cannot infer starter status from statistics alone.
5. Historical backtesting (V8.0)
Before kickoff, generate lineups and save a dated JSON snapshot. After the game, import that exact snapshot plus a verified actual-score CSV. This compares forecasts and the saved lineups only. A snapshot made after kickoff is NOT a valid prospective backtest. Files stay on your device.
No historical snapshot loaded. Generate lineups, then save the snapshot before kickoff.
6. Release status
V8.0 research beta (NOT validated as profitable): salary/roster validation, configurable Showdown simulation, player-confidence warnings, manual role reviews, scoring-unit checks and pregame snapshot comparison are implemented. Not independently verified: official current DraftKings scoring for every contest, live injuries/starting roles, historical kicker/defense point estimates, actual ownership, prediction calibration or tournament-winning probabilities. Keep a dated pregame snapshot and validate actual results before treating projections as dependable.