If rocket kills are easy, you need to find more skilled dodgers. I'll admit a fair percent of my rocket kills are pretty easy, and do I consider prediction a skill? Yes. A lot of times though, people think it's hard to test because they don't assess each individiual opponent as exactly that, an individual. This is why I don't force models. I aproach, and assess each player as an individual, with an individual style, and don't simply lean on one particular approach, just because it seems to work on a larger percentage of my opponents.
I've played tons of people who just dodge when they know you can fire again, and also dodge the same direction every time (probably because of "handedness"), I've played people who just constantly dance, in a pattern they they think makes them as hard to hit as possible, and I've played players who wait you out, and actually react to your shots. It takes one time engaging a person to tell usualy how they are, 2 max, and when you start dealing with them as individuals, it changes your game. That level of adaptation is the kind of "skill" I like to practice. Hence, when able to predict something because you studied it, becomes more of a skill then just blind prediction. That is where your misconception lies.
It's the difference between betting on a horse race according to the names listed in the race, and actually reading up about the jokey, and the horse, and history behind both of them. Yeah, one isn't really a skill, it's just blind luck. The other, is ones ability to asses what has happened, and how that will affect what will happen.
My favorites? The ones who do react. They are the most satisfying to kill. You gotta time your rockets fast, and make them dodge in a way, where they force themselves into getting hit. Get them to move, move them again, but this time where you want them, then get some damage and push done, and finally use the last push to know where the next rocket goes for the kill, all the while making sure they don't kill you in the process.