regression fitness function - smoothing out the equity curve

trading strategy optimization using genetic algorithms

Moderator: moderators

SJTRADES
rank: <50 posts
rank: <50 posts
Posts: 14
Joined: Mon May 22, 2006 3:32 pm
Reputation: 0
Gender: Male

Postby SJTRADES » Wed Jun 07, 2006 3:13 pm

Michal,
Frankly, I can't understand the math. Could you explain how the above would be used on a simple MA crossover System of say ma2 and ma9. I have been trying to use OpenPositionProfit and 'closed' NetProfit as an added filter to reduce positions and subsequently increase positions when OpenPositionProfit recovers to > 0 until the new (higher) 'closed' Net profit is reached.
Thanks for any help.
Jay

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Wed Jun 07, 2006 3:56 pm

SJ,

unfortunately, it is not possible to employ custom fitness function in Tradestation alone without some 3rd party optimizer.

I am using GGO in this respect.

User avatar
eudamonia
rank: 500+ posts
rank: 500+ posts
Posts: 536
Joined: Thu Jun 15, 2006 9:50 pm
Reputation: 0
Location: Rocklin, CA
Real name: Edward Heming
Gender: Male
Contact:

Postby eudamonia » Thu Jun 15, 2006 10:37 pm

Michal,

I just wanted to say I read this first on the TS forum and I wanted to commend you for putting this up. This is exactly the thing that I have been investigating with the Grail (but had come up with no tenable solutions) and which all GAs need (appropriate fitness functions). FYI have you investigated true multi-objective fitness functions? I have read some great journal articles on this but haven't seen anything put into use for this type of application. Hellis' description of the combative simulataneous fitness functions described in the GA primer link on your site seems quite interesting.

Additionally, I wanted to stand up and say "nah,nah" to those close minded ninnies that dominate the TS forum. I can't stand posting or reading there anymore and am frankly am ashamed of the treatment that TRO received there. Just remember that they will reject anything that doesn't fit into their narrow mental paradigm. Along those lines, it is a joke to get any help from the so called "masters of the universe" crowd over there (i.e. Nemo, Mathematician etc. etc.). No we are not cool enough to hang with them. Sorry but I think that's a bunch of bologna.

Lastly, I wanted to thank you for putting together this forum where folks can actually discuss trading systems and putting together components in an intelligent fashion. Despite the fact that both you and TRO are several years ahead of the trading curve from the majority of the rest of us, I've never heard either of you disparage or act condecendingly to others (even newbies). You're and TRO's willingness to share even whole strategies is unheard of in the TS crowd. If nothing else your strategies have given me the incentive to achieve better with my own strategy design.

Thank you.

Edward Heming
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.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Fri Jun 16, 2006 12:01 am

eudamonia wrote:have you investigated true multi-objective fitness functions?


I am starting an academic project in a matter of months in which 3 universities will participate. The project theme is Time series prediction in financial business systems using genetical algorithms. We will have a supercomputing cluster at our disposal for the project at the Brno University of Technology running Linux. The purpose of this project is to explore thoroughly the possibilities of GA model and "basic building blocks" search in time series prediction and self-development, so stay tuned - the official pages of the project will be here at kreslik.com. I have not heard of any similar project in the world, so maybe this one will be the very first in this respect and scope.

eudamonia wrote:Hellis' description of the combative simulataneous fitness functions described in the GA primer link on your site seems quite interesting.


This will be one of the research objectives. The dueling parasites paradigm in genetic algorithms is yet to be explored in the financial time series prediction field.

eudamonia wrote:Along those lines, it is a joke to get any help from the so called "masters of the universe" crowd over there (i.e. Nemo, Mathematician etc. etc.). No we are not cool enough to hang with them. Sorry but I think that's a bunch of bologna.


Well I've got no problems communicating with Mathemagician. He even told me that anytime I would need any help, he is here to help me. I agree JJ is a littlebit excentric type of guy, but then if you're a first-class mathematician, you're allowed to be one :) One must hand it to him he is one of the best system developers on the TS forums.

eudamonia wrote:Lastly, I wanted to thank you for putting together this forum where folks can actually discuss trading systems and putting together components in an intelligent fashion.


Originally, I've put it up for TRO to be able to post without restrictions :-# and, I confess, to get rid of that ugly bunch of emails from the horrible Yahoo group stuffing my Outlook every day, too. Now I see that other people (including me) also need more space to express their ideas freely. So this is the place to express anything without any limitations.

eudamonia wrote:You're and TRO's willingness to share even whole strategies is unheard of in the TS crowd.


Speaking of myself, I share the tools that will lead you to the correct strategies rather than a specific strategy code. One has to use his own brain to crack it. Honestly, I'm not willing to share the core strategy code. That would even be dangerous to anyone who doesn't understand the underlying principles. I'd rather discuss and teach you how to fish than bring you a plate full of fish. Which one is more valuable for you?

eudamonia wrote:Thank you.


[schild=14 fontcolor=000000 shadowcolor=C0C0C0 shieldshadow=1]You're welcome, Ed![/schild]

Michal


P.S. to create the cool looking smile with a custom banner message like that above, just hit the "Smilie Creator" button in the message editor.

User avatar
eudamonia
rank: 500+ posts
rank: 500+ posts
Posts: 536
Joined: Thu Jun 15, 2006 9:50 pm
Reputation: 0
Location: Rocklin, CA
Real name: Edward Heming
Gender: Male
Contact:

Postby eudamonia » Fri Jun 16, 2006 1:04 am

michal.kreslik wrote: I am starting an academic project in a matter of months in which 3 universities will participate. The project theme is Time series prediction in financial business systems using genetical algorithms. We will have a supercomputing cluster at our disposal for the project at the Brno University of Technology running Linux. The purpose of this project is to explore thoroughly the possibilities of GA model and "basic building blocks" search in time series prediction and self-development, so stay tuned - the official pages of the project will be here at kreslik.com. I have not heard of any similar project in the world, so maybe this one will be the very first in this respect and scope.


This sounds very interesting. Please keep us posted.

michal.kreslik wrote:This will be one of the research objectives. The dueling parasites paradigm in genetic algorithms is yet to be explored in the financial time series prediction field.


Excellent. I'm very excited to hear what you come up with.

michal.kreslik wrote:Well I've got no problems communicating with Mathemagician. He even told me that anytime I would need any help, he is here to help me. I agree JJ is a littlebit excentric type of guy, but then if you're a first-class mathematician, you're allowed to be one :) One must hand it to him he is one of the best system developers on the TS forums.


Well I wasn't intending to take away from either of those guys skills, however, I do find they can be rather stand offish on the forum (I don't know either of them personally). As my mother would tell me many times, "If you don't have anything nice to say then say nothing". My experience has been they post many snide remarks and few contributions. Also I don't believe greatness excuses poor behavior. Sorry I don't buy that.

michal.kreslik wrote:Originally, I've put it up for TRO to be able to post without restrictions :-# and, I confess, to get rid of that ugly bunch of emails from the horrible Yahoo group stuffing my Outlook every day, too. Now I see that other people (including me) also need more space to express their ideas freely. So this is the place to express anything without any limitations.


I am very glad you see it that way. Personally, I believe that intelligence is best shared. If everyone in the world kept all of their best ideas to themselves the world would be in a very sorry state.

michal.kreslik wrote:Speaking of myself, I share the tools that will lead you to the correct strategies rather than a specific strategy code. One has to use his own brain to crack it. Honestly, I'm not willing to share the core strategy code. That would even be dangerous to anyone who doesn't understand the underlying principles. I'd rather discuss and teach you how to fish than bring you a plate full of fish. Which one is more valuable for you?


I agree that teaching one how to fish is more important. However, it is also important to see other strats to see how you can improve your own.

Edward

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Mon Jun 19, 2006 10:55 am

eudamonia wrote:I agree that teaching one how to fish is more important. However, it is also important to see other strats to see how you can improve your own.


The core elements of any susccessful strategy are so embarrasingly simple that a first grader could actually come up with the idea. Yet no one can "teach" anyone to come up with good ideas.

What can't be devised by a first grader though is the exact logic, the programming, the prior research. This is hard work that has to be done.

Resume: you can learn how to do proper testing, logic/algorithm modeling, programming, how to conduct a research. But still you have to come up with some idea in the first place. The hard part is 99 out of 100 ideas turn out to be a piece of trash.

User avatar
eudamonia
rank: 500+ posts
rank: 500+ posts
Posts: 536
Joined: Thu Jun 15, 2006 9:50 pm
Reputation: 0
Location: Rocklin, CA
Real name: Edward Heming
Gender: Male
Contact:

Postby eudamonia » Mon Jun 19, 2006 4:03 pm

I agree that good strats are very simple, but interestingly enough "simple" is usually the last place that a beginner will look.

Additionally, it is difficult for a beginner to distinguish between an idea that has absolutely no potential, and one that might work with just a couple of adjustments. Experienced strat designers can provide a certain amount of guidance to a beginner that I personally know was crucial for my development as a strategy designer.

Probably the biggest difference between the experienced strat developer and the beginner is not their "market knowledge" although that is also important, but rather their abilility to diagnose what exactly is causing a particular strategy to not meet expectations. I believe that this is probably the most important skill beginners are missing. Any thoughts on that?

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.

User avatar
whitmark
rank: <50 posts
rank: <50 posts
Posts: 3
Joined: Fri Jun 16, 2006 11:39 pm
Reputation: 0
Gender: None specified

Postby whitmark » Sun Sep 17, 2006 11:49 am

Hi Michal,

Have you tried to code your regression fitness function for NeoTicker as a custom statistic for inclusion in the system performance viewer potentially to be used as a Smart Search strategy for the grid optimizer? If so, would you be willing to share?

Regards,

Whitmark

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Mon Sep 18, 2006 1:33 pm

Hi Whitmark,

I have not yet coded it for NeoTicker since I'm now in the process of going thru all "undocumented functions" :) within NeoTicker to be able to start working efficiently with Neo.

As soon as I'm done, I'll post my code here.

Michal

Keys
rank: <50 posts
rank: <50 posts
Posts: 2
Joined: Mon Nov 20, 2006 4:15 pm
Reputation: 0
Gender: Male

Postby Keys » Mon Nov 20, 2006 4:27 pm

For those of us not in the know, where can we learn about these "undocumented functions"? Thanks.

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.


Return to “genetic optimization”