Unfortunately vertical spread is not always a consistently reliable input as there are factors in addition to launch angle that can influence POI.
Yes, but it all tends to be normal data. And stats lets you see the magnitude and shape of that normal/bell curve. A certain amount will perform within +/- 1 SD, and a certain amount within +/- 2 SD, and so forth, which also accounts for the fliers. The fliers simply live in the extremities of the curve. There's nothing we can do about how well the ammo is manufactured, or environmental effects, but we can measure what a given sample does in our gun(s) and get predictions from those obervations with enough of them.
I'll give an example. Here's a test I did a long while back with my Win52D's factory barrel and the mid-barrel tuner that came with the gun when I bought it. The Spearman's rank correlation coefficient between delta Y (elevation on target) and velocity for Win52D mid-barrel tuner:
test-1-without.csv is: -0.528 (p-value = 0.007)
test-2-without.csv is: -0.317 (p-value = 0.122)
test-3-with.csv is: 0.138 (p-value = 0.511)
test-4-with.csv is: 0.208 (p-value = 0.319)
With the first ammo tested in test 1 the barrel was clearly in an upswing, which is what the negative correlation value tells you, with a very trustworthy answer, which is what the p-value tells you. Lower p-value is more trustworthy. The second ammo in test 2 also exited during an upswing, but was slightly more erratic, as seen by the higher p-value. The higher that value gets, the less you can trust the correlation value because it seems to be more noisy data. Then test 3 is with ammo 1 again, but with the mid-barrel tuner installed at some particular location. The p-value is much worse, so noisier data that's much less trustworthy as a result, and now the barrel is in a downswing since the correlation value is now positive. And test 4, still with the same tuner location, shows something similar for ammo 2.
My suspicion in this case was that the barrel's motion was near one of the reversals in the motion cycle where it goes from downswing to upswing, and the installation of the mid-barrel tuner at that location slowed the overall motion down enough such that bullet exits were now on the other side of the upswing trough and into the downswing portion of the motion cycle. Possibly with some exiting very near or slightly on either side of that trough, and this is probably what's causing there to be a lot more noise in the data making it much less trustworthy. If it had been solidly on the other side of the trough rather than perhaps sometimes straddling it then I believe the p-value would be a bit lower, showing more trustworthiness that all shots were leaving during the downswing. Overall, the noise that the ammo's characteristics bring to the picture should be more or less stable, so substantially more noise in the data would tend to indicate something different on the side of the gun's behaviour.
Is that something you could determine with just shooting? Well, you would definitely be able to tell something was wrong due to the increased vertical stringing. It might be harder to tell where to go from there without this kind of data, though. In this case, I would know that the barrel was being slowed down too much because bullets were no longer exiting during an upswing and I would have to search the middle ground to find the optimal region, as it clearly goes from an upswing that's a bit too fast to a downswing between the two tests. Since it was already in an upswing, which is what you want, you wouldn't want to stray too far from that. The goal would be to get the upswing into a favourable speed range for that target distance. It gets tricky when you're in the neighbourhood of a crest or a trough since you instead want to exit during a stretch of time where the motion is always in the same upward direction. Get everything just right and that correlation value drops near zero, indicating it can't tell if muzzle velocity is making any difference in elevation on the target anymore because more and more shots just go into the same spot.
