The reason I didn't suggest the Matrix approach, as I said in my original response, is because it doesn't synch up properly to notes of all lengths without using some sort of automation to change the pattern or it's rate of play...which defeats the purpose of avoiding mod amount or mod wheel automation.
Suppose you have a note which is one bar long, and another note which is one half-bar long. Now, if you want the mod effect to come in, say, half-way through the note, you'd need to create a CV, Curve, or Note pattern which rises after one-quarter of a bar for the half note, and one half of a bar for that whole note. Since the Matrix repeats, the modulation would either rise and fall twice on the whole note, or not at all for the half note, depending on which note you were trying to set it up for. Am I not right on this? It gets even worse if you have a long note which isn't placed at the start of a bar.
Of course, as has been said, the mod wheel can be used instead of the Mod A/B. I would wire the output of the mod from the controlling Malstrom to the Mod Wheel input on the sounding Malstrom.
Again, I refer everyone to the demo Reason Module.
http://www.waterwave.org/Malstrom%20Delay%20Example.rns