93 solo rounds · 3,043 GPS swings · every chart below recomputes for the years you select
The whole story in one table (all years, unaffected by the filter). Scoring collapsed because the blow-ups collapsed — birdie/par rates barely moved.
Selected years vs Shot Scope's published averages by handicap.
Green-hit rate on genuine attempts (club could reach; par-5 layups excluded). Context so the odds land right: the distance where you convert half your attempts is about 75 yards. For a 20-handicap it's ~92y; for a 10-handicap ~129y (Arccos data). Even a 10-handicap hits fewer than half his greens from 150. Honesty note: the mid-iron rates (125y+) rest on small samples and a green-center estimate that's fuzzy to ~10-15 yards, so the true figure is a range — roughly 1-in-6 to 1-in-5 from 150-175, not a precise 8%. The direction (odds fall off a cliff past 125y) is solid; the exact number is not.
Each card is a green seen from above, drawn to scale (shaded disc = 15-yard radius green). The flag marks the green's estimated CENTER — the actual day's pin isn't in the data, so "long/short" means past or shy of the middle of the green. Day-to-day pin positions scatter evenly front and back, so they blur individual dots but can't fake a one-sided pattern over 100+ shots (and your own tap-in stats agree: 425 misses short vs 202 long). Treat differences under ~10y as noise.
Same verified holes, split by the club in hand. Cells show green-hit % (attempts). Blank = fewer than 6 attempts. This is the "which club do I actually trust" table.
The 18Birdies export never included lie or surface. These four panels recover it by testing every GPS point against the real course geometry (fairway, green, bunker and water polygons) that OpenStreetMap has mapped for 9 of your 10 most-played courses. A call counts as confident only when the ball sits clearly inside a surface (beyond GPS error of a boundary); boundary-zone balls are shown as "edge" and kept out of headline rates. Unmapped courses (Compass Pointe + some one-offs) are excluded. Data caveat: every shot is logged in the app, but the export silently drops ~38% of swings — mostly whole holes at a time (the holes present are 93% complete chains), and the vanished holes average slightly better scores, so rates here skew a touch pessimistic if anything. The app's own stats remain ground truth for overall rates; the value here is conditional: what happens given the lie.
Genuine green attempts only (club-can-reach). Does the rough — not just under-clubbing — explain the short miss?
First logged shot on par-4/5 holes (par from OSM where mapped, else long openers). The app gives one fairway number per round — this splits it by the club in your hand. "Edge" = within GPS error of a boundary, could be either side.
Logged shots that finished in (or flew over) mapped water. Penalties aren't in the export, so this is a floor, not a total — and "crossed" includes safe carries over hazards.
The dashboard's "sequence method" infers a green hit from stroke arithmetic; the polygon method reads the map. Where both speak, do they agree?
Every logged opening shot on a par 4/5, plotted against the straight tee→green line (up the page = downrange; doglegs blur single holes but not the pattern). Dots inside the shaded band finished within 20 yards of the line. The stats card includes what the club leaves you into the green — distance-to-target is only half the question; the other half is whether the leave sits in your make-rate sweet spot (see "Approach school" above for your odds by distance).
Par 4s and 5s you've played enough to pattern (3+ logged tee shots, selected years). Median leave = your typical approach distance into that green after the tee ball. Holes where the leave lands in the 125–175y dead zone (your 8–16% make range) are flagged — those are the holes where a different tee club or a deliberate layup spot could buy strokes. (Courses where OSM lacks hole/par data show "—" for par and can include the odd long par-3 — iron rows with tiny leaves are those.)
Practice pays where you're both often and weak. Counts genuine green attempts only (club-can-reach test); "layup share" shows how much of that distance you already, sensibly, don't attempt.
Each updates with the year filter; targets are the next level up, not tour golf.
Every club: how often you hit it, and its true distance (median, middle 50% shaded). Toggle years above to watch the bag evolve.
How this is measured: for each drive the GPS knows the tee, where the ball finished, and the green. We draw a straight tee→green "target line" and measure how many yards left or right of it the drive ended. So "left/right" means of that line, not in the rough — it's a scatter measure (doglegs can fool it on single holes; the pattern over dozens of drives is what counts). Actual fairways hit comes from the app's own stat. Dogleg check (Nick's): holes whose played path runs ≥18% longer than the straight line (32 of 161 classifiable) were excluded in a re-test — the scatter barely moved (35→37% within 20y), so the two-way miss is real, not a dogleg artifact.