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

Re: Sys Test--Summary Report Question



PureBytes Links

Trading Reference Links

Walter and Larry,

Thanks to you both for the information.

What I have been doing is using the system optimizer on various combinations of
indicators.  I'm not doing this to fine tune a system, but rather I'm
intentionally curve fitting it.  I first run a gross high level test to get a
feel for the data set, then I will run a very detailed and lengthy optimization,
and using the suggestions from you and Larry, I will export my results into
either Excel or JMP to do some analysis.  I can then look at the data across
different variables in a bar chart or histogram (depending on whether I'm
looking at "hits" or values).  For me, the visual presentation works much
better.

Here's my starting point as an example:  I'm taking five minute data for a
market going back one to two years (yes, many data points).  I'm running
optimizations of various systems, say a simple MA cross.  I want to evaluate the
results in graphical form to see if there is say a bimodal or multi(?)modal
result that might suggest harmonics of a frequency (a cycle and multiples of
that cycle).  e.g.  on an optimization using twenty days of data, the results
show a strong clustering of successes around the 3/10 MA values as well as
several lesser success around the 7/21 MA then strong success around the 27/93
MA, etc. How do these optimized values vary in the next twenty days?  I want to
break the database up by weeks, months, quarters, and years to compare the
optimization results for each.  I don't know where this leads, depending on the
how the results line up, there may be some commonalities to investigate
further.  Another result may be that there are some recurring ST, very ST, IT,
or very IT cycles/trends that I can take advantage of.

I would like to take the results, depending on how the preliminary information
looks, and try to correlate this with other broader market indicators (TRIN, MA
of TICK, INDICIES, etc.) to determine if there is anything worth pursuing.
Perhaps changes in these can be used to help predict the change in MA cross
periods.

These are the sorts of investigations that I want to do.  Getting the results
out of the system tester summary was my first step.  Thanks again.  Also, if you
have a handy example of a VBA sub to open each, or you can point me to some
resources to improve/brush up my Visual Basic skills, I'd really appreciate it.
It looks like I'll probably really benefit from using it to help manage the
analysis.

Thanks for the suggestions!

Dave Nadeau

Walter Lake wrote:

> Hi Dave
>
> In System Tester --> Reports --> Reports ... Click the "Trades" tab -->
> Right click ---> save to file
>
> Open into Excel. If more than one file, use a looping VBA sub to open each
> text file as a separate sheet in your workbook. Use "!" in your formulas or
> VBA code to link it to a data or results sheet.
>
> "... can evaluate histograms for harmonics or other repeating patterns. ..."
>
> Do you mean histogram type indicators or data analysis histograms? Could you
> say a little more about your project or post some code to explain what you
> mean.
>
> Best regards
>
> Walter
>
> ----- Original Message -----
> From: "Dave Nadeau" <dave.nadeau@xxxxxxxx>
> To: "MetaStock Listserver" <metastock@xxxxxxxxxxxxx>
> Sent: Wednesday, May 31, 2000 11:58 AM
> Subject: Sys Test--Summary Report Question
>
> | This is probably repetitious because I think I remember a
> | recent posting describing the answer to this...my apologies
> | if this is so.
> |
> | Does anyone know how to export the results of a system test,
> | as shown in the summary report dialog box?  I've tried
> | printing to a file without success in being able to read the
> | results.  Ultimately, I would like to get this into MS Excel
> | so that I can evaluate histograms for harmonics or other
> | repeating patterns.
> |
> | Also, is there a way to capture all of the test results from
> | a very high iteration system test?  I realize that the
> | summary report captures only a select few of them.
> |
> | Thanks in advance for any help!
> |
> | Dave Nadeau
> |