Strategy discovery program

[rsidivergence]

Has anyone tried Stratasearch? It's supposed to do strategy discovery based on a set of indicators and trading rules. Any luck with it?

 

[dmilong]

Nice idea in principle but It runs forever if you choose too many indicators because it is based on rule permutation, which in my humble opinion is a very naIve way of doing searches.

It has many analytical tools though including Monte Carlo simulation and forward testing analysis.

 

[taowave]

StrataSearch is a great program,and I have used just about everyone out there.The poster is correct that it can take a very long time to scan if you use hundred or thousands of rules.

With that aside,I have been "enlightened" and many of the concepts and I have read or thought were correct were proved to be the exact opposite of what I should have been doing.I am a semi discretionary trader and use SS as a research assistant.Its obviously unbiased,and has no preconceived notions as to what works and what doesnt.I can assure you that if nothing else,you will discover things that you would not with any other program.

The program is not a black box in any way,and every rule is disclosed.In addition,the user can program any rule they can program,and rules can be turned on or off.I would strobly reccomend the demo.The developer is a great guy and responds to questions the same day.

Any questions,feel fre to ask..

Has anyone tried Stratasearch? It's supposed to do strategy discovery based on a set of indicators and trading rules. Any luck with it?

 

[dmilong]

I can assure you that if nothing else,you will discover things that you would not with any other program.

Any questions,feel fre to ask..

Sorry to say this...but you sound like sales or tech support...

If you know of a program that can make money why run around advertising it?

 

[taowave]

Sorry to say this,but you sound like you just started trading yesterday if not later....

Sorry to say this...but you sound like sales or tech support...

If you know of a program that can make money why run around advertising it?

 

[dmilong]

Sorry to say this,but you sound like you just started trading yesterday if not later....

I am still trading, what about you?

 

[HowardCohodas]

Nice idea in principle but It runs forever if you choose too many indicators because it is based on rule permutation, which in my humble opinion is a very naIve way of doing searches.

It has many analytical tools though including Monte Carlo simulation and forward testing analysis.

I'm not familiar with this program in particular, but not all rule permutation or parameter setting permutation programs run exceedingly long. About seven years ago I built a genetic algorithm based tool kit and applied it to finding rules and parameters for trading purposes. Genetic algorithms devise methods of sampling the search space rather than testing all nodes in the search space, to find "good" solution rather than best solutions using search techniques based on the biology of how genes split and recombine at genesis.

It was a great exercise and I discovered some interesting rule sets. However, I never really solved the rule complexity issue with my approach. The more complex the rule set the harder it is to find opportunities. I did, however, validate the solutions by applying the rules to data the generator had never seen before.

 

[Black Swan]

I am still trading, what about you?

LOL pwned...

 

[Albert Carey]

I'm not familiar with this program in particular, but not all rule permutation or parameter setting permutation programs run exceedingly long. About seven years ago I built a genetic algorithm based tool kit and applied it to finding rules and parameters for trading purposes. Genetic algorithms devise methods of sampling the search space rather than testing all nodes in the search space, to find "good" solution rather than best solutions using search techniques based on the biology of how genes split and recombine at genesis.

It was a great exercise and I discovered some interesting rule sets. However, I never really solved the rule complexity issue with my approach. The more complex the rule set the harder it is to find opportunities. I did, however, validate the solutions by applying the rules to data the generator had never seen before.

Quite interesting work you have done. Did you use an algorithm in the public domain? I'm not very good with programming but I would like to try it.

 

[HowardCohodas]

Quite interesting work you have done. Did you use an algorithm in the public domain? I'm not very good with programming but I would like to try it.

I developed the tool kit in VB using concepts I learned from text books and research papers I found on the internet.

I used the tool kit to optimize parameters on commonly available strategies. Some of my favorite come from Dr. Alexander's Books.

I also used the tool kit for rule discovery. This is a more challenging problem to solve as you try to remove all prejudice on rules that you already know. This led to interesting rule sets that were often too large, and impractical without robotic automation. This, however, violated one of my trading principles. I need to understand the rules and I need to have my brain and fingers review and implement the recommendation.

Trust but verify. --Ronald Regan

Had I the millions that Google has to put into the AI that is going into the driver-less vehicle, I might have more confidence in my work.

A man's got to know his limitations. --Harry Callahan (Magnum Force)