Slate Board
Every player the model priced this round: a research board, not the betting record. Only rows marked with a tier were given out as picks; guard-held candidates stay visible with a reason, and everything else is NOT A PICK. Tagger, opposition and injury-boost context is layered on each row.
Edge = model projection - line. L5 = last-5 disposal avg, Vol = last-10 std dev, both over games actually played before this round. Where a player has fewer games than the label, the real count is shown beside the number (“25.7 3g”); under 4 games we show n/a rather than a figure. E/Vol = |edge| / vol (risk-adjusted edge). L5 vs line = games in that same sample that hit the pick side. vs Opp = share of games by players in the same position bucket, against this opponent, that went over the line this season; a * means that sample was thin so all positions are pooled, which is a rough guide only. ! = the call is contradicted by every game in its own shown sample. TAG = likely tagged (model suppression on hover). BOOST = usage lift if the named teammate is out. HELD = a candidate the guardrail held this round; the reason shows on hover so nothing silently vanishes. Green = over, red = under. Click a player for their card.