Several years ago I knew a guy that tried to develop a cvt for the micro-midgets. He ran a 250cc 2 stroke class and used the rotax snowmobile engines. The issues he had were with the clutch cycling back down. In other words in a straight line acceleration like on the starts it was real killer. But when he would have to lift in a corner like to make an adjustments to his line. There was a lag after the throttle was re-applied, where the engine would have to get back up to rpm's then the CVT would have to re set itself to the correct gear ratio, and then after all that it would start to pull. I'm not talking like a second long delay or anything, but just enough that it messed up the control of the chassis and cost an extra car length whenever letting off the throttle. He worked him self silly modifying the ramps, and springs in the clutch, and after a full season of frustration he went back to more conventional set-ups. I agree with the above comments, a V belt consumes a lot of energy, and extra weight in the drive system costs a lot of acceleration.