I'm currently studying Ralph Vince (ISBN 978-0-471-75768-9) and Workshop6. I'd like to ask you for a confirmation (or denial) if I got it right ("system" means here the combination of system and market):

1) it's better to use one single account instead of creating virtual sub-accounts per system and assigning sub-equties per system. Let every system "think" it has the whole account.

2) OptimalF should be determined for each system.

3) In a portfolio, OptimalF for each system is not necesserely the same as it would be if the system were alone.

Understanding (3) is my problem right now. Ralph Vince writes that the peak should be searched in multi-dimensional space (3-dimensinal for two systems). In my spreadsheet example, however, an overall peak is always at the same OptimalF values, which I found for each system individually, although two systems are not producing 100% correlated trades. Where is my mistake?

(this one is by Raplh Vince)