|
PureBytes Links
Trading Reference Links
|
Scott wrote:
> Optimal f is an awesome tool. However, it's applicability is
> directly proportional to the stability of the distribution of
> wins and losses of your system. If your win / loss patterns
> are stable optimal f *will* return incredible equity growth.
> Its just math. As I have said many times to many people,
> the bulk of system design effort should be spend on
> characterizing the stability of the distribution of returns
> from the system, not on maximizing the absolute value of the
> systems expected return (edge). Give me a stable system with
> a *tiny* edge (like blackjack with no bet limits) and I will
> make orders of magnitude more money than a guy with a *large*
> but unstable edge.
Any hints on how to develop this type of stable systems?
Presumably, treating a diversified system as one system is one approach
(e.g., apply your m.m. to the results of a system traded over a
portfolio).
Any other ideas? Types of systems which are fairly stable?
Ways of measuring the stability of a system? Any number of minimum
trades to consider? On the order of thousands is probably preferable: so
you should either have a lot of data to test over, or else trade over
short time frames ...
What else?
Thanks!
Cab
|