The methods described in the book DO work. Problem is, it's extremely difficult to learn, and even MORE difficult to PERFECT. Then when you DO master it, you end up with an edge that, in practice; is smaller than what you can get by counting cards. You'll almost have to be insane to try it in the U.S., with the double zero games. AND; because roulette is not an even money game, the variance is so bad it will make your BJ swings look like a sure thing.
Then once you decide you can handle all of these problems, you'll go in to play, and that's when Murphy's law is going to take over. Whatever can go wrong will; changing dealers, changing ball weights, etc. Imagine standing around for six months, recording 50,000 spins, hoping to find a bias in the wheel; then you come to find out that the wheel wasn't biased AFTER all! Or; you find the bias, and the night before you finally decide it's time to play, they change the whole wheel.
Small wonder he'd rather sell the system than play it himself.