John, keep in mind I'm not proposing rebalancing by itself. Nor, am I trying to find the best set of parameters. I'm trying to think of tests that will inform me on basic questions I have about the nature of things. Once I am convinced about the nature of things, then I will act. And, after I have come to whatever conclusion, I won't then optimize to find the 'right' parameters.
Regarding your test: you're right about RYAIX not being the true inverse. Also, I'm doing a test now that shows the results of rebalancing through the various stages between positive and negative correlation.
Matt