Ok, to be a little constructive here, if you feel you have to adjust the allocation among the algos (please read "avoid allocating all capital to a single algo"), then you should add constraint to LSPM algorithm.

Or more specifically - while finding the global extreme for all combinations of individual optimal Fs, you can constrain the matrix to avoid individual Fs i.e. > 0.4.

OK: 0.3, 0.3, 0.4
OK: 0.0, 0.2, 0.2
NOK: 0.1, 0.4, 0.5

This way you can find the global extreme under constrains.

Disclaimer: I pulled this of the top of my head, I didn't yet read Vince's books till the end.


Edited by nemozny (07/10/18 16:46)