|
PureBytes Links
Trading Reference Links
|
You've got two problems. System testing is of limited value in that it
doesn't really do what most people think it does which is give them
absolute answers on which is the best system, and it is good only for
relative comparisons of one system to another. If one is 10% more
profitable with half as many trades and it's out of the market half
the time, it's better than one the one that earns 10% less and is in
the market all the time. However, in live trading you don't know which
one will actually perform the best.
In additon all systems tests are based on previous history. Depending
on how much curve fit is in your system, the systems tests may be
meaningless for the future.
All TA programs have so so systems testers. You can't compare one
systems test run on one program to the same test run on another
program because they all use different assumptions in their programs.
The best use of testing is to check the robustness of your system, to
figure out what kinds of market conditions it functions well in and to
figure out what filters help the system perform reasonably well.
If you want to understand the issues of market conditions better, read
Roy's newsletter, www.metastocktips.co.nz
When you start trading anything live, including the so called
mechanical systems, they're going to produce different results than
the ones you got out of the tester.
I think using tradesim is a good idea depending on what you're trying
to do. I sounds like you want to reinvent the wheel, which is okay,
but it does take a lot of time and money.
You find the market conditions have the most dramatic outcome on the
performance of any system you choose to trade so get some
understanding of that. Like a said Roy's newsletter is a good place to
start.
A lot of the backtested systems you see in books, magazines and on the
net have never been traded live, only bactested. You know what that's
worth?
Stridsman's books have hundreds of backtest results in them so you can
see who different things work in general, but again Stridsman didn't
test the system in uptrends, downtrends, sideways markets and sideways
markets with rotation. If he had he would have gotten very different
results for each condition.
When you run 7000 stocks through a tester and it says look your system
WORKED great on these 200, that has nothing to do with what is going
to work great on that same 200 during the next three years.
I've read maybe 20 systems development books and another 200 trading
books (Roy's newsletter is still the best of all of those.) Here's
what I've concluded after testing over 1000 systems.
Find a good basic system, and I mean basic. Learn how it works in each
of those four market conditions. Learn how to identify those
conditions. Find a strong list of prescreened stocks that have a
excellent trade record of beating the market. Test your system on the
list, tweak it and then apply when the market conditions are right.
You'll make the most money that way. IBD Value Line S&P ,etc all have
those lists. You may need a different system for different conditions,
aw well as a different strategy, or be willing to sit it out when you
know the conditions are appropriate for losing money. There is no
prescreened list or system, that can consistently make money in a
downtrend except a system designed for shorting. Of course it won't
work in an uptrend.
Have fun shaping that wheel.
--- In Metastockusers@xxxxxxxxxxxxxxx, "GaryNielson" <gary@xxxx> wrote:
> I'm not trying to pick a fight here, as I am extremely impressed
> with Metastock EOD v. 9.1. However, I bought it in search of a
> strong backtesting program after reading Robert Pardo's
> classic, "Design, Testing and Optimization of Trading Systems." I've
> read the manual and from what I can tell, Metastock treats each
> security independently and does not give you a summary report on how
> a trading system would have actually worked with a basket of stocks.
>
> I have tested an add-on to TC2005, TC-Companion, and you can run a
> trading system through 7,000 plus stocks and get a summary of how
> your system would have worked on the entire stock market and then
> run Money Management to see how you would have actually done (within
> reason of course) because you obviously can't buy every stock!
>
> Am I missing something here? Am I on the right track or am I
> misunderstanding what backtesting is and is it really necessarily to
> review the performance of a strategy on hundreds, if not thousands,
> of stocks. Right now, I am thinking that Metastock's backtesting
> program is guilty of overfitting, where you find a strategy that
> fits past performance of a particular stock and trade that stock
> that way. Has anyone traded this way and does it give you the
> results you want? Maybe I am agonizing over nothing.
>
> Help me out here.
------------------------ Yahoo! Groups Sponsor --------------------~-->
Try Online Currency Trading with GFT. Free 50K Demo. Trade
24 Hours. Commission-Free.
http://us.click.yahoo.com/DldnlA/9M2KAA/U1CZAA/zMEolB/TM
--------------------------------------------------------------------~->
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/Metastockusers/
<*> To unsubscribe from this group, send an email to:
Metastockusers-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|