THE BOARD
Every player in the game under the same dials as your XI. P start is a column, not a footnote — it is the term that decides everything, and where a projection is worth distrusting it is almost always that number.
Evidence is how much Premier League history sits behind the number.
NO PRIOR means an empty history_past — a positional median and a flat
0.50 start prior, an admission of ignorance rather than a projection.
PART is some PL history but under a full season.
Sort by it: a projected 4.2 built from a fallback should never read the same as a 4.2 built from 38 games.
xMins is expected minutes for the next gameweek, from this season's per-gameweek history.
Blend w is how much of a player's rates come from this season rather than last:
w = min(1, minutes / 500), so a nailed starter is near 0.7 by GW5 and a rotation
player stays anchored to his prior — which is right, because we have not learned much about him.
Form, price movement and transfers are display columns and are not in the projection. Form is realised points over 30 days, so it carries exactly the finishing luck the xG terms exist to remove, and it double-counts what expected minutes already measures. Price movement predicts price; transfers predict price and ownership. Those are real questions — they are just not "how many points will he score". The reasoning in full.