An automatic average can be mathematically correct and still be the wrong number to use for club selection.
Understand what entered the average
A club history may contain mishits, partial shots, punch-outs, unusual elevation, strong wind or incorrectly recorded start and finish points. If those attempts are treated as stock swings, the learned distance can drift away from your real capability.
A recommendation engine cannot recover missing context that was never recorded.
Separate stock, partial and recovery shots
Maintain a stock carry estimate from representative strikes. Label deliberate partial shots and recoveries separately. Total distance can vary with ground conditions, so carry is often the cleaner starting point for approach decisions.
Do not allow a handful of abnormal outcomes to replace a distance established through reliable measurement.
Audit obvious outliers
Review club histories periodically and remove or exclude records that are demonstrably wrong: a missed GPS point, duplicate shot, unrecorded stroke or distance that belongs to a different club. Preserve genuine poor outcomes in round analysis without treating them as stock capability.
This distinction lets the golfer learn from mishits without teaching the caddie that a mishit is the intended number.
Use recommendations as evidence, not authority
Confirm the lie, wind, elevation, landing requirement and your current strike before accepting a club suggestion. A device can support a decision; it should not replace the golfer’s awareness of the shot.
When a recommendation looks implausible, investigate the underlying club record rather than adapting your entire game to the output.
Keep stock distances separate from partial, recovery and corrupted records. A poor shot may belong in the round diagnosis without belonging in the club recommendation.
Use the free golf scorecard to record your round, or explore swing video analysis for a movement question.
STROKES interprets the scoring evidence you record. It does not diagnose technical swing mechanics or biomechanical conditions from outcome tracking alone.
