logo
Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Options
Go to last post Go to first unread
darkod  
#1 Posted : Friday, January 2, 2009 3:46:21 PM(UTC)
darkod

Rank: Newbie

Groups: Registered, Registered Users
Joined: 2/26/2008(UTC)
Posts: 9

Hello all. I hope someone with more experience or a Equis person will answer this question ASAP.

Initially I thought the Enhanced System Tester is the perfect tool to compare the 26 built-in systems and select the "best" one for a particular stock. However, after carefully examining the test results lately I have found numerous errors. Here is the latest obvious example:

Lets take Alpha Natural Resources on NYSE, symbol ANR. Select the last three years, 01/01/2006 to 12/31/2008. For the other options of the test I leave them mainly at default except using $15 commision per trade and set it to invest $10,000 each time (position). Also, selecting just Long trades.

When the test is done with all 26 systems order the results by Net Profit. Towards the end is listed Fractal Trading 1 with information that according to the test there were only 4 trades, all of them loss trades.

However, when loading the ANR chart with the last 750 periods and displaying the last 250 periods in the screen just in those last 250 periods (approx a year) you can clearly see more than 4 trade signals just in 2008, and few of them are profit trades. There is clear mismatch between the trade signals the expert is showing and the trades actually recorded in the Positions section of the test.

So this raises the question about the validity of the test results. The profit while applying Fractal 1 for the last three years is surely bigger than the stated -34% in the test.

And this also raises doubts about other test results too. What I liked about the EST was the ability to quickly test a lot of systems on a lot of stocks. If I doubt the results and do the math manually it becomes hell of a job...

Any thoughts please? Thanks.

EDIT: Changing some of the settings (button More) while setting up the test I managed to get better result. I changed trade execution to close prices and without any wait (0 bars). But still the Positions section does not show all the trade signals and that means the end results are not completely correct as if following the trade signals. Am I doing anything wrong?

EDIT2: I have done more tests and came to the following conclusion. Something is "wrong" with 2006 test. If testing single years, 2008 and 2007 correspond to the trade signals the expert is giving. Combining 2007-2008 together the results are still OK. But testing both single 2006 and as part of a 2006-2008 test, the results no longer correspond to the trade signals. Anyone that can comment on this?
mstt  
#2 Posted : Saturday, January 3, 2009 9:33:19 PM(UTC)
mstt

Rank: Advanced Member

Groups: Registered, Registered Users, Subscribers
Joined: 7/25/2005(UTC)
Posts: 1,042

Was thanked: 57 time(s) in 54 post(s)

Hi darkod

There are several reasons why trades are not accepted. Probably the most common cause of trade failure stems from coding that creates an entry signal on one bar but doesn’t attempt to execute until the following bar (usually the OPEN of the following bar). Using EST generated delays rather than coded delays is not a good way to go when 100% of capital is committed to each trade. Why? Because the EST calculates the number of shares it can afford to buy on the signal bar, not the execution bar. When these trade stages are not on the same bar, any upward price movement tends to create a situation where the EST cannot afford to buy the previously calculated number of shares. This results in the trade being abandoned.

Check out the various reports that the EST creates and you’ll surely find a bunch of them that are considered but not executed. As I’ve already suggested, one solution is to insert any required trade delay into the code and remove the strategic or bar delay from the EST. Another solution is to double the trading capital but allocate only 50% to each trade. What happens then is that the EST uses 50% of available capital to calculate the number of shares to purchase, but if the price to pay on the execution bar is higher it essentially still has 50% of additional capital to call on.

Don’t give up on the EST – it’s still a useful tool. However, to get the best from it you need to understand how it works and what its frailties are. I like to set the EST up such that it mimics my Trade Equity test tool. Not only do I then end up with some more useful portfolio results, but I can also check that the EST and TE results are in the same ballpark. When gross results are within a few cents of each other I can be pretty sure that both test methods are executing the same trades at the same time and generating the same results from individual securities.

Regards

Roy

www.metastocktips.co.nz

darkod  
#3 Posted : Sunday, January 4, 2009 5:51:02 AM(UTC)
darkod

Rank: Newbie

Groups: Registered, Registered Users
Joined: 2/26/2008(UTC)
Posts: 9

Thanks for the reply.

As stated in my first EDIT I already found out how to cancel the bar delay (set it to 0) and I also set the execute prices to Close prices on the same bar that the trade signal is generated. That sorted part of the problem.

I managed to sort the main problem with raising the equity and your post just confirms that for me. Without too much thinking about it, I was executing the tests with $10,000 initial equity and with $10,000 per trade. This obviously creates issues. After I raised the initial equity to $50,000 the results started to look as expected.

I also have to say that the manuals don't go too deep into these different settings of the system tester and one small setting can have a huge impact. Even going trough the Mastering Metastock pdf does not give you detailed info.

As far as coding is concerned, I am still learning my "basics" in trading and I really can't take on coding right now. Just sounds like too much to me.

EDIT: The only thing still confusing me are the Margin Requirements in the Broker tab in System Testing Options window. When I set 10,000 per transaction I want the transaction to be 10,000 regardless whether on margin or not, long or short. Does that mean I have to set Long Initial and Long Maintenance to 100% and Short Initial and Short Maintenance to 200% (it requests value between 101 and 200)? Currently I am using the default values 100, 0, 150, 200 and on the Positions tab in the results it looks like the transactions are 10,000 (or similar). Should I keep the margin values like this to get precise results? I would appreciate if you could advise me on this.

Cheers.
Users browsing this topic
Guest (Hidden)
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.