As far as I know, the OptimalF calculation is not exposed with a function. So OptimalF cannot be dynamically calculated in a backtest, although this would make sense. In live trading you could use the Retrain function for this.

I've made a notice for possibly implementing a dedicated optiomalF function.