And a whole lot of diagram searching
The idea is that for each stadium, I find the best diagrams I can possibly get and import them into Sketchup. If the stadium is outdoors, I just use Google Earth (in 3D when available) and I create a picture from above the stadium and import that picture into Sketchup as well.
When I have a home run to estimate, I first identify where the ball landed. I make sure I can try and find the seat or a landmark and I try to match it to it’s location in the diagram/picture in Sketchup. If successful, I then use the pencil tool from that location and I trace a line back to where homeplate is located.
Next, I try and identify the angle of descent along with the height above field level in the video. This gives me the Sp. (landing spot) of the home run. Other factors I always consider are the rate of which the ball is seemingly losing (or not losing) velocity and it’s apparent hang time (loft). I try to visually create a path in which the ball would travel in a worse case scenario given the information I have. High arcing flyball home runs are typically the easiest, while line-drives that land high up are always the hardest estimates in this regard. But once done, this gives me the Min. (minimum plausible distance). Then I reverse that thinking and I try to visualize a reasonable best case scenario to get the Max. (maximum plausible distance). Then the Est. (most likely the correct estimate) is the easiest part. You simply average out both Min. and Max. and that’s how you get Est.
As for how credible this method is… well, it depends. I actually have a system to help out with this as well. Conf. (confidence) tells you how confident I am in that particular estimate. 0 being ‘I have absolutely no idea’ and 6 being nearly perfect or exactly perfect (landing on flat ground relative to field level). As for my skills/judgment abilities, you might not believe this, but I’ve actually trained myself to do this. I first practiced with MLB home runs with estimates provided by the folks at Hittracker (Statcast had reliability issues when I was practicing). I’d make my estimates first and then compare my results to theirs to see if I was close enough. I will admit that at first, the results were too sporadic. It was still better than just making a blind guess, but not good enough to document. However, over time, I did get much better. I’m now at the point where around 90% of my estimates are within 3m / 10ft of the Hittracker estimates. This is good enough for me. Is it good enough for you? That’s for you to decide.