|
PureBytes Links
Trading Reference Links
|
I have been bred to believe that a system should be able to stand on its
feet trading with a simple single contract. Once that is achieved, only
then apply the variations in contract sizes.
Is that just bad breeding ?
Thanks for your reply
Peter
-----Original Message-----
From: Mark Johnson [SMTP:janitor@xxxxxxxxxxxx]
Sent: Saturday, August 22, 1998 8:52 AM
To: omega-list@xxxxxxxxxx
Subject: Why I don't worry about Profit Factor
ProfitFactor (Net$Won / Net$Lost) assumes you are
trading a constant number of contracts per position
throughout the entire duration of the test. But that
isn't the way real people trade their real money
accounts.
Suppose Jane A. Trader starts out with a $50,000
account trading one-lots of TBonds. When her
account equity rises to $250K, Jane is _not_ going
to stick with one-lots. So why should she perform
testing on one-lots? And why should she employ a
figure-of-merit that assumes she will keep constant
position sizes, independent of account equity?
Instead, Jane A. Trader will employ some kind of
"anti-martingale" position sizing. (This simply
means that as her account equity gets bigger, she
puts on larger number of contracts for each position).
Since she trades this way, Jane should test this
way. WITHOUT constant position sizes. Violating
the first assumption in Profit Factor.
|