| Author |
Message |
michal.kreslik
rank: 500+ posts
Age: 33
Joined: 13 May 2006
Posts: 918
Location: Monte Carlo, Monaco
|
|
|
Last edited by michal.kreslik on Thu Nov 23, 2006 11:31 pm; edited 5 times in total
|
|
   |
     |
|
JPT
rank: <50 posts
Joined: 22 May 2006
Posts: 49
Location: Monterey, CA USA
|
|
| |
|
|
michal.kreslik
rank: 500+ posts
Age: 33
Joined: 13 May 2006
Posts: 918
Location: Monte Carlo, Monaco
|
|
Great & free Visual Studio learning resources:
http://www.appdev.com
After registering, one can download for free these learning videos:
ASP.NET using Visual Basic .NET 45 MB
ASP.NET Using Visual Basic 2005 113 MB
ASP.NET using Visual C# .NET 44 MB
ASP.NET Using Visual C# 2005 109 MB
Developing Applications in Visual C# .NET 49 MB
Exploring ASP.NET "Atlas" and Web 2.0 87 MB
Exploring ASP.NET 2.0 Using Visual Basic 2005 115 MB
Exploring ASP.NET 2.0 Using Visual C# 2005 115 MB
Exploring BizTalk Server 2006 52 MB
Exploring SQL Server 2005 94 MB
Exploring Visual Basic 2005 351 MB
Exploring Visual C# 2005 341 MB
Managing and Maintaining Windows Server 2003 (70-290) 80 MB
Microsoft SQL Server 2000 40 MB
Microsoft SQL Server 2005 95 MB
Programming in Visual Basic .NET 55 MB
Programming in Visual C# .NET 62 MB
Visual Basic 2005: Developing Applications 103 MB
Visual C# 2005: Developing Applications 105 MB
Visual Studio 2005 Tools for Microsoft Office 93 MB
|
|
|
|
|
|
| |
|
|
JPT
rank: <50 posts
Joined: 22 May 2006
Posts: 49
Location: Monterey, CA USA
|
|
Michal:
I have not had time to spend on this since I have been trying to get my Grail to give me more than 1 trade utilizing the fitness function 2. A month ago, it gave me a great rendition of trades, and I did not change the program.
My feeling was always that the Grail genetic optimizer was not stable and I am thinking that more and more. It seems to go off to these instability regions and then stays there for a while.
If I can't fix it, I guess I will sell my license.
Thanks
|
|
|
|
|
|
| |
|
|
michal.kreslik
rank: 500+ posts
Age: 33
Joined: 13 May 2006
Posts: 918
Location: Monte Carlo, Monaco
|
|
| I have already sold my license. Have you contacted the Grail tech support? They should still provide support for the existing users.
|
|
|
|
|
|
| |
|
|
eudamonia
rank: 150+ posts
Age: 33
Joined: 15 Jun 2006
Posts: 471
Location: Rocklin, CA
|
|
Michal,
Just curiously are you working on building your own GAs in C# or are you going to use the optimizer in NeoTicker?
Edward
|
|
|
|
_________________
Eudaimonia (pron.: you-die-moan-e-a) (Greek: εὐδαιμονία) is a classical Greek word commonly translated as 'happiness'. The less subjective "human flourishing" is often preferred as a translation.
|
|
  |
|
|
michal.kreslik
rank: 500+ posts
Age: 33
Joined: 13 May 2006
Posts: 918
Location: Monte Carlo, Monaco
|
|
Both 
|
|
|
|
|
|
| |
|
|
eudamonia
rank: 150+ posts
Age: 33
Joined: 15 Jun 2006
Posts: 471
Location: Rocklin, CA
|
|
Good answer. Are you using multi-objective GAs? Right now I'm trying to get more specific code examples (even Java or C++) for multi-objective GAs. Are there any links you can recommend?
Also, do you happen to know if the Grail operates any differently than a "standard" GA?
Edward
|
|
|
|
_________________
Eudaimonia (pron.: you-die-moan-e-a) (Greek: εὐδαιμονία) is a classical Greek word commonly translated as 'happiness'. The less subjective "human flourishing" is often preferred as a translation.
|
|
| |
|
|
michal.kreslik
rank: 500+ posts
Age: 33
Joined: 13 May 2006
Posts: 918
Location: Monte Carlo, Monaco
|
|
Look up the "genetic algorithm" keywords on sourceforge.net. Lots of good free stuff there - libraries and code, too.
Certainly you could do no harm by running
genetic algorithm library
There is no "standard" GA. But certainly Grail doesn't utilize many concepts in the solution search that could be utilized.
One of the concepts omitted in GGO is the issue of seed. How do you determine the first seed of the candidate solutions? In GGO, you do this randomly. That is ok as long as you have absolutely no clue about what the solution might be. If you already know from your prior trading experience, that one of the best MA2 length is 40, this information is useless for you if you run the optimization with GGO since there's no way telling the GGO to seed this good value in to the first generation.
The GGO then might, or might not find that 40 is actually a good value in the process. Remember, GA only SPEEDS UP the process of optimization, so if there is a very high number of possible candidate solutions, you might help the evolution a littlebit by telling the GO to use some kind of working solution as a member of the first generation.
This is like telling the nature that the genes of a dolphin would be a better way to go than the genes of a dinosaurus right off the bat.
|
|
|
|
|
|
| |
|
|
JPT
rank: <50 posts
Joined: 22 May 2006
Posts: 49
Location: Monterey, CA USA
|
|
Michal:
I want to let you know that I have been using Genetic optimizers for the last 3 1/2 years and I think that Grail has an instability. doing similar optimizations in Grail and Optimax, I get widly changing results. In addition, when I re-run the same run on Grail, just add another 1/24 of the data (one month more out of two years of data on a 3M ER2) I get dramatically different results. when I do the same in Optimax, I get very similar parameter spaces.
Personally I think the Optimax optimizer is much more stable. Unfortunately, no support and I can't get into the source code. Wish I could.
I am not so sure that one needs to GA. Most of my "robust" indicators have a few parameters anyways. What I really wanted it for is to do the Prado walk forward analysis.
Are you considering including in your C# optimizer a walk forward component?
|
|
|
|
|
|
| |
|
|
|
|
Calendar
