The concept of evolution theory as trading system

Chart Forum » Trading - Systems » The concept of evolution theory as trading system

«« Previous Next »»

Thread Start New Thread
Last Poster Posts Pages Last Post

Author

Message

pouria

Member

Post Number: 1
Registered: 08-2007

Rating: N/A
Votes: 0


Thursday, August 02, 2007 - 11:16 pm:

concept evolution

Hello all you Trading expert….
I have been for over a long time trying to develop a system that is so robust that can be used on many different markets.

We all know that an entry-signal that might work perfect for one stock is some time totally worthless for another stock or market or even another time period of that same stock.

To over come this problem I have used the concept of evolution theory in my trading system.

Basic explanation:
------------------
I have 3 entry signals (Entry3, Entry4, Entry5), each Entry signal has an entry value (Ent_3, Ent_4, Ent5). Default value for all entry values are equal to zero.

At end of each trade the, value belonging to the Entry-signal that entered the trade is :

- incremented with one if the trade is a wining trade or
- Decremented with one if the trade is a losing trade

If an Entry signals value goes below -3 (this value can be optimized) then that entry-signal is disabled for a period of 150bars (this value can be optimized).

After the 150bars disabled for a period of that Entry signals, the Entry signal is enabled by incrementing its value by one. The reason for this is that the Entry signal might work nicely for some period but be useless for other periods.

So for the first 1-2 years of the simulation the system is learning which Entry to use after that It only uses the entries that fits best to that stock, and from time to time it gives disabled entries a new chance to compete.

----------------------------------------------------------------------
Like the giraffes with long neck, the Entry-signals with high value will live and the entry signals with low value will die out ( only for period of time ( 150bars)).

Now hear is some of my questions:
1.) Has there been any similar work done earlier ?? ( I most sure it has!!!) If so was the result succesfull???
2.) Please give me your opinion on the Idea and what the negative side of a system like this would be.
4.) There is a learning period for the system, dose this mean that I have to choose a much earlier starting date in the my SCAN ???
5.) Any ideas how this could be more improved. Maybe use the same thing on the exit signal ?!?!? No sure how that would work….

Please share your thoughts with me I'm very interested to see what every one else thinks about this idée.

/Pouria

lafee

Member

Post Number: 608
Registered: 04-2003

Rating: N/A
Votes: 0


Friday, August 03, 2007 - 04:03 pm:

Pouria,

Victor Niederhoffer runs a website called www.dailyspeculations.com. He studies evolutionary biology to find concepts to apply to the financial markets.

You may find some of his ideas useful for this VERY relevent topic.

Cheers Lafee

Don't ask an academic if what he does is relevant

peter1

Member

Post Number: 163
Registered: 12-2005

Rating: N/A
Votes: 0


Monday, August 06, 2007 - 10:18 pm:

An entry signal is only part of a trading plan. A trade cannot end without an exit signal. You may need to develop Exit3, Exit4, Exit5. Now you will have 3x3 trading systems. Are any of these trading plans profitable? You will have to back test all of them to find out. Then you will have to dissect the yearly results to see which ones are robust over all market stages (Weinstein) or suited to particular market stages.

You may evolve a great entry signal but without a great exit signal and money management you will have nothing worth using.

nihonjin

Member

Post Number: 13
Registered: 05-2007

Rating: N/A
Votes: 0


Tuesday, October 02, 2007 - 10:14 am:

An interesting read is The Predictors by Thomas Bass. It tells how chaos theory was used to build predictive software for financial markets.

Presumably the programs are even better now. The book makes you realize that the playing field is not flat, that the markets are rigged so that we are punters in a casino and the banks are the house with a permanent edge.

The Prediction Company is real and owned by UBS.

The market wins because it has an edge! Traders blink first.

dark_matter

Member

Post Number: 2
Registered: 09-2007

Rating: N/A
Votes: 0


Tuesday, October 02, 2007 - 08:55 pm:

Hi all,
even tho i'm new I am reading this topic with interest. Pouria, the system actually seems more like a neural network to me. as each trade occurs the system "learns", a successful trade allows the system to repeat the trade. a continuing run of unsuccessful trades tells the network not to do it again. the larger the dataset the more easily the system can perform a successful trade.

however the entry (and exit points) are not necessarily predicted with this system. In terms of evolutionary theory these points are more readily described as selection pressure. maybe some method of quantifying this value will allow a more predictive system. possibly a relative value away from an average. by relative I mean the actual dollar value the price varies from the mean divided by the mean. as this value increases (it will be a small decimal) the "selective pressure" increases and could indicate an entry however as the value decreases there is less "selective pressure" and would indicate an exit.

randomness is also a key feature of evolutionary theory with random mutation occuring every time a cell/organism reproduces. and this is true randomness in being unpredictable. however the effects of this mutation are variable and measurable, and considerably effect the success of the organism. in trading there are many levels of randomness and the success of a trade may not necessarily be predicted. however if the success a majority of trades can be predicted then a winning system can be developed.

the one thing that most emphasises evolutionary theory (particularly the darwinian model) is adaptation. this turns out to be a function of selection pressure and mutation. if an organism cannot successfully adapt to increased selection pressure it cannot survive and reproduce. in most organisms mutation is cumulative and is heritable (passed to the offspring). but i'm a bit lost on how to apply this to the trading paradigm you have outlined. at first guess i think these ideas apply to the system itself. i'll give it some thought and try to find a way to quantify it in some applicable manner. (might even have to drag out some papers and texts on evolutionary theory).

possibly some of the thousands of existing TA tools can be substituted for the signals but you'll be as old as moses before you can test them all.

Anyway it sounds like an interesting idea and is probably no sillier than trying to pick winners with astrology of reading tea leaves (no offence meant to those that use these methods) i'd be interested to hear if you develope the system and if it can stand up over time. Good Luck!!

sittingduck

Member

Post Number: 16
Registered: 11-2004

Rating: N/A
Votes: 0


Wednesday, October 10, 2007 - 09:08 pm:

I am a believer of the school of dart board picking/scalpel exiting. I don't mean totally random picks, but aimed in a general direction based on some simple rules (ie trending up etc). It's how you exit that makes or breaks your system.

So my advice to Pouria is to start your focus on the exit strategy first, then worry about better entries.

I think that's sort of what Peter said isn't it :-)

Grab a copy of Van Tharp's Trade Your Way to Financial Freedom for a process of evaluating your strategy in an objective manner to determine how successful it should be. By following the process you will learn a lot about how to formulate a strategy and what bits you may have missed due to whatever reasons. It's good because it doesn't preach a particular style/indicator/method, it's simply a process to allow you to develop your own system based on your own beliefs of what will work.

Good luck
Dave