Quote from jcl:
I've read his latest work. He extends OptimalF to portfolio management by calculating the individual reinvestment factors as parameters in an optimization process.
Unfortunately, this opens the whole can of worms connected to optimization, such as ending up on an unstable peak in parameter space. Therefore I do not like this approach, although it indeed theoretically generates the optimal result.
At the moment, I use instead two different factors for portfolio management: Vince's OptimalF for reinvesting, and Thorpe's covariance matrix for distributing the capital among strategy components. This has the advantage of being easy to calculate, and seems to give fine results so far.