Most sports analysis research often moves from performance to motion: starting with the outcomes and working backward to explain them. But for my research, the outcomes were noisy and was affected by many factors, such as slope, grain, or tiny disturbances. A flawed stroke could even go in (although it would lack consistency).
By focusing on motion patterns, I was able to ask the meaningful questions after. Rather than attributing “good strokes” or “bad strokes” (or success of failure) to isolated variables, I was able to spot patterns in groups.
This perspective also shaped what “skill” might mean in putting; it’s not just about having a good putt once, but it’s about reliably creating a certain type of motion.
What’s important here is that this approach doesn’t assume an “ideal stroke.” Different motion patterns can coexist as an optimal motion, each with their strengths and weaknesses. I thought forcing them into a binary judgment would not fit the purpose of this research.
The project is more of computer-vision motion science now, and putting has become a case study.