The Cycloneshock is simply not designed for accuracy, as you indicate above. Its power, combined with the tiny "barrel" means that inaccuracy will always be a problem. The whistle is innately inaccurate, even with the improvement of two holes. Any tendencies you have observed will be due to systematic error (That is, some flaw in the design causes darts fired from that particular model to leave the barrel in a particular way more often than usual). One of the greatest mistakes made in nerf "science" is using tiny sample sizes. Try firing 10 different Cycloneshocks from 10 different stores 1000 times each (Impractical I know, but think of it as a thought experiment), then see if the same tendency is still present.
Sleightof hand is an answer, but will not work the same way for everyone, simply because the gun itself is different in every model. What you have done is get used to the quirks of your particular model, and reacted accordingly. I will be testing this method Today (possibly) to find out is the same flaw is present in my model. Bravo on finding a solution, but the suggestion that it will work for others is flawed.
I had thought of the factor of different models, but I kind of put it to the side. But nonetheless you are right. It did work for me, but it's more accurate of me to say it "may" work for others, but is not guaranteed. Thank you for that, and thank you for offering to run tests. I had some activities to do today so I couldn't get around to it myself, but if you'd like to that's very helpful and highly welcomed.