Methodology
What the source can support today.
This is a source receipt, not a performance page. It explains the current-season data we can verify, the official baseline we show, and the gates a future MatchDay XI model must clear before it earns a label.
Live scope
The public command center reads the official Fantasy Premier League source for the current season. The verified context includes the current or planning gameweek, deadline, dynamic squad rules, teams, players, and fixture records. The fixture window can degrade separately, so missing fixtures are shown as a source limitation rather than silently substituted.
- Official contextSeason, gameweek, deadline, rules, teams, players, and availability fields.
- FixturesOfficial fixture records, including blank/double context and kickoff uncertainty when published.
- PreseasonPublic entry identity can be checked, but future picks are not public until the FPL deadline has passed.
The baseline is not a model claim
Official FPL baseline
ep_next is an official FPL field and a low-information reference point. It is not MatchDay XI xP, it is not a calibrated probability distribution, and it does not justify a confidence interval. When it is missing, malformed, or unavailable, the product says so.
Freshness and unavailable data
Every verified source record carries a status, verification time, generation, and any error or limitation. Fresh data can support the current context. A stale complete snapshot can remain useful only when it is explicitly marked stale and inside its allowed age. If a required source is missing, malformed, too old, or internally inconsistent, MatchDay XI makes no current decision claim.
This is why the command center can say “source unavailable” and still point to a browser-local draft: a useful next step does not require pretending the missing source is healthy.
Model promotion gates
No proprietary model is promoted to a user-facing MatchDay XI projection today. Any future model must have all of the following before it earns a public label:
- Leakage-safe evaluation: temporal splits and withheld periods that reflect how a manager would actually receive a forecast.
- Calibration: uncertainty that is measured, finite, and honest about where the sample is thin.
- Provenance: reproducible source snapshots, feature definitions, artifact identity, and a reviewable generation record.
- Legality: outputs constrained by the live season rules, availability, budget, club limits, and positional quotas.
- Failure behavior: malformed or missing inputs fail closed; they never become a confident-looking number.
What is not public yet
Transfer optimization, chip strategy, live matchday tracking, and mini-league intelligence are future surfaces, not claims about what this public build already does. There is no public MatchDay XI xP model, captain edge number, backtest table, or model-performance score to report yet.
Known limitations
- Preseason data has little completed-season evidence, so official baselines are especially low-information before GW1.
- FPL availability fields can change after a snapshot; recheck the source and official team news near the deadline.
- Fixture publication can be incomplete or revised. Blank, double, postponed, and TBC states stay visible rather than inferred away.
- A verified source is not a guarantee of a player outcome. It is a record of what was knowable when the source was checked.
Use the receipt
The command center shows the current source status and limitations in context. If you are preparing for GW1, build a local draft and treat the official baseline as a reference—not a promise.