# Launch Monitor Import > Canonical: https://pinflag.io/features/launch-monitor-import PinFlag turns launch monitor sessions into per-club dispersion truth: imported, fit robustly with outliers flagged, and stamped as measured data — the top-priority source behind your on-course numbers. Sources are never blended, and every number says where it came from. *The PinFlag iOS app is in pre-release. Leave your email and we’ll tell you the moment it’s available.* Launch monitor sessions become durable, per-club dispersion truth. Each import is fit robustly — cold-start shrinkage, outliers flagged rather than silently deleted — and stamped as measured launch-monitor data, the top-priority dispersion source the strategy engine uses. Sources are never blended: a number comes from your launch monitor, your tracked rounds, or a model, and it says which. ## Why it matters - Your real carry numbers and spread live in the launch monitor; decisions happen on the course. This is the bridge. - Averages hide the pattern — a robust fit of the whole session doesn’t. - Provenance matters: measured data should outrank a model, and it should say so. ## How it works 1. **Export from your monitor** — Trackman, Garmin, or Rapsodo files — or any monitor via a mapped CSV with explicit units. 2. **Import a session** — Sessions keep device, ball, and environment metadata, up to 5,000 shots. 3. **Get honest fits** — Each club’s dispersion is fit with cold-start shrinkage; outliers are flagged, never silently deleted. 4. **Play with it** — The fits become Measured dispersion — the source your live strokes-gained numbers prefer above everything else. ## Everything Launch Monitor Import does - Trackman, Garmin, and Rapsodo file support, plus mapped generic CSV - Yards/meters and mph/kph handled explicitly at import - Robust per-club dispersion fit with 3σ outlier flagging — flagged, never deleted - Durable sessions with device, ball, and environment metadata - Measured provenance: launch-monitor data outranks on-course and model sources - Never blended: exactly one dispersion source behind any number ## A worked example — real numbers **What tighter dispersion is worth on the green.** Proximity is where dispersion cashes out. PinFlag’s putting baselines price the difference between the putts a wide pattern leaves and the putts a tight one leaves. - 25 ft left: Expected putts 1.931 - 15 ft left: Expected putts 1.780 - 8 ft left: Expected putts 1.500 (0.431 strokes better than 25 ft — per approach.) Moving your typical leave from 25 feet to 8 feet is worth 0.431 strokes every time it happens. Knowing your real pattern is how you find aims that produce it. *Numbers computed with PinFlag’s engine (frozen SG engine, Broadie-verified baselines v2026.4, app lineage c21ee22).* ## Who it’s for Launch monitor owners — simulator players, range-session regulars, fitters’ customers — who want their measured game to be the one the strategy numbers use. ## What’s different - Never blended: launch monitor, on-course, or model — exactly one source behind any number, disclosed. - A sample-size floor (10 shots per club) before a source is trusted at all. - Outliers are flagged for you to judge, not deleted to flatter the fit. - Import is architecturally walled off from the scoring engine. ## Frequently asked questions ### Which launch monitors does PinFlag support? Trackman, Garmin, and Rapsodo session files are supported directly, and any monitor works through a column-mapped CSV import with explicit unit handling. ### Does PinFlag mix launch monitor data with course data? Never. Dispersion comes from exactly one source at a time — launch monitor, on-course, or model — with launch monitor data ranked first. Blending is structurally impossible and tested. ### What happens to outliers in a session? They are flagged at three standard deviations and kept visible. You decide what a topped 5-iron means; the fit doesn’t hide it. ### Are meters and km/h handled? Yes — distance and speed units are declared at import (yards/meters, mph/kph) so a European export can’t silently corrupt your numbers. ### How many shots does a session support? Up to 5,000 shots per imported session, with device, ball, and environment metadata kept alongside.