Re: CL_Evaluating & Optimizing systems-Oct97,Oct98 WATCH OUT

["Phil Lane" <patterntrader@xxxxxxxxxx>]

----- Original Message -----
From: M. Simms <prosys@xxxxxxxxxxxxxxxx>
To: Phil Lane <patterntrader@xxxxxxxxxx>; <clayburg@xxxxxxxxxx>;
<fritz@xxxxxxxx>
Cc: <omega-list@xxxxxxxxxx>; Code List <code-list@xxxxxxxxxxxxx>
Sent: Tuesday, November 23, 1999 5:29 AM
Subject: RE: CL_Evaluating & Optimizing systems-Oct97,Oct98 WATCH OUT


> Good point....
> I am having trouble (who isn't ?) determining optimal contract size or the
> usual "size-of-bet" problem as it applies to trading.
>
> I like the volatility-based approach....Any ideas on exact implementation
?
>

here's what I use - dolswing is the desired dollar swing per ATR. This is
something you can relate to the account size - which you can also specify.
On a long-term SP simulation it has you trading about 20 contracts back in
the 80's for every 1 now. If you do this and your system is working well the
profits and losses should be about the same for the duration of the test
(instead of getting bigger and BIGGER!!!!)

if currentbar>maxbarsback then begin
value38=Xaverage(TrueRange,50);

if compound=1 then
Base_Cnt=(startact+netprofit)/startact*dolswing/(value38*Bigpointvalue)
else
Base_Cnt=dolswing/(value38*Bigpointvalue);

if whatever then buy base_cnt contracts close;