[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: System Tester accuracy q (from 23 yrs in mkts)



PureBytes Links

Trading Reference Links

Thanks Dave. As MS advertises, the point is to spend time figuring out
trades, not figuring out the program. I'll try to figure out a few things
myself, but knowledgeable, and gracious, users such as yourself help cut the
wasted time to a minimum.

Irv

----- Original Message -----
From: "Dave Nadeau" <dave_nadeau@xxxxxxxxx>
To: <metastock@xxxxxxxxxxxxx>
Sent: Friday, June 01, 2001 12:07 AM
Subject: RE: System Tester accuracy q (from 23 yrs in mkts)


> Irv,
>
> I run both Metastock Pro 7.03 and Tradestation 4, Build 27.  I've done
> comparisons on the System Testing capability of both and found them to be
> very close.  Not exact, but very close.  I set up tests that generated
many
> trades (>2000 on 13000 bars--TS4's limit) against the same datafile and
> compared the results.
>
> The real challenge is to make sure that your conditions are exactly the
same
> (e.g. Metastock Enter & Exit on Open, Delay = 1 is the equivalent of Buy
> Market or Sell Market in Easylanguage).
>
> Your system tests are going to zero very quickly because you are using $$$
> amounts.  This method implicity assumes that you are fully reinvesting
100%
> (or 200% on margin) with each trade...not likely in real trading.  I
prefer
> to use "Points Only" tests to give me an idea of how many points per
> contract or dollars per share the system would have generated or
ultimately
> lost.  You could then export the trades into Excel and apply any sort of
> position sizing algorithm (e.g. Optimal F) to your trades to create a
> dollarized equity line.
>
>
> Dave Nadeau
> Fort Collins, CO
>
> -----Original Message-----
> From: owner-metastock@xxxxxxxxxxxxx
[mailto:owner-metastock@xxxxxxxxxxxxx]On
> Behalf Of Irv
> Sent: Thursday, May 31, 2001 3:30 PM
> To: metastock
> Subject: System Tester accuracy q (from 23 yrs in mkts)
>
>
> Hi,
>
> I know y'all went through excruciating agony over the single- vs.
> double-precision integer deal earlier, and I can appreciate the problems
it
> could cause strictly mechanical systems, but I would like to know if
anyone
> has doublechecked the MS ST results against other programs, or self-coded
> routines? There are just SO many STests that go to zero so quickly
(trading
> longs and shorts), it makes me wonder.
>
> In case this strikes some listers as a dunce's question, over 95% of my
> income has come from the markets for the past 23 years. But yes, I'm not
> that familiar with MS.
>
> Irv
>
>
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
>