Market move opening → now (primary bookmaker): AH line and price, in ppt of win probability. Generated ….
Moved — the market moved ≥ 4 ppt opening → now (line and price together; about one quarter-line step). ≥ 8 ppt is marked "big move". Trigger: absences (the side it moved against has more new absences), manager change (last 14 days), else unknown.
Stable, something amiss — moved < 4 ppt, but a registered bias with a past edge (|z| ≥ 2) fires, or one side has 2+ more new absences and the line hasn't reacted.
Stable — moved < 4 ppt, nothing amiss. Not open — no odds yet (scraped from 5 days before kickoff).
Position — only from tested rules with strong evidence (now: 4+ without a win while the market stays calm, past return about +2%, z ≈ 3), at the current line and price. "Edge priced in" = the market already moved towards that team. Paper positions only: they are logged ~1 h before kickoff to build the forward record.
Flags — a popular belief that applies to the team. Past return = backing such teams at the closing handicap; every bet averages about −1.8% (the margin), so "vs the average bet" is the edge. Outlined = strong evidence (|z| ≥ 2).
Line track: hollow ring = opening, filled dot = now, ±1 goal around the opening, ticks every quarter line; home handicap, left = home stronger. † = found in the data (only the forward test counts). New absence = not out in the team's previous match.