MACKENZIE · the bag · ball flight · —
A shot that finishes 15 yards right either started there or bent there, and the fix is not the same. The launch monitor records both: where the ball set off, and where it finished. The gap between them is curvature. Split every shot that way and the bag turns out to hold two opposite faults at once.
Across the page is where the ball started relative to the target line. Up the page is how much it bent in the air. A ball in the centre box started on line and flew straight; every other box is a named miss. One dot per shot — nothing is averaged away.
Every trusted shot · scale 1:1 both ways
The same total miss, split into the part the ball started with and the part it bent. These stack: start + curve is exactly the offline distance, because the export's own geometry says so — the two bars reconstruct it to within 0.02 yd. Read left of the line as left of target.
Median per club, yards
The same nine boxes, one panel per club, all on a single shared frame so the panels compare. The long irons sit high (bending right, face open to path); the short irons sit low (bending left, face closed to path). Those are different faults, not one fault at two sizes.
Face-to-path is the mechanism, and it is only on 209 of 255 shots — two sessions tracked the ball but not the club. It correlates with curvature at r: directionally right, not deterministic, since curve also depends on speed and spin.
Curve is not a mystery: it comes from where the face pointed relative to the path it was swung on. Each diagram below is one club seen from above, at the moment of impact, drawn to scale — the arrow is the swing path, the bar across it is the face. When the two disagree, the ball bends; the wider the wedge between them, the harder.
Only clubs the monitor tracked appear here. The Gap Wedge is the case worth studying: its arrow and its face point the same way, so the strike is clean — they are simply both aimed a long way left. Compare the 6 Iron, where the face sits open to a path swung across the ball, and the ball leaves bending. The 5 Iron is the honest exception: its face and path are square to a fifth of a degree and it still bends 12.9 yd right, which is the clearest sign on this page that face-to-path is not the whole story.
Face-to-path against curvature · one dot per tracked shot
A median off one afternoon describes an afternoon. Each club's curvature is plotted per session below, and the answer is uncomfortable: the clubs that look like they slice are exactly the ones with the least corroboration. The 6 Iron produced a 4 yd draw one day and a 20 yd slice another. The only per-club curves seen twice and agreeing both bend left.
Median curve per session, yards · each mark is one session
Nothing here is modelled. The export carries launch_direction (where the ball set off) and carry_deviation_angle (where it finished), both in degrees, positive right. Curvature is the difference. In yards at the shot's own carry:
start = carry × sin(launch_direction)
curve = carry × ( sin(carry_deviation_angle) − sin(launch_direction) )
Two checks before trusting it, computed from the same data this page draws. Curvature must track the spin axis — —. Start line must track the club face — —. And the two parts reconstruct the export's own offline column to —, so the split is arithmetic rather than interpretation.