Genetic algorithm optimization?
Genetic algorithm optimization?
What is the advantage of using the Genetic optimization compared to the exhaustive one? And also are the default settings in the GA the best ones to use? or can someone please explain what each one means so it can be set to recieve the best results....Thankyou
forget about the GA , just use exhaustive, see on difference.
Most important..learn how to read a chart, if u are a daytrader...
I combined bullish chart recognition and follow the signal, happened twice to me today...after i got into position of YM couple of seconds later the signal alert visual and audio follows
eventually i'll let MC do the trading...after 2.1 version maybe.
Most important..learn how to read a chart, if u are a daytrader...
I combined bullish chart recognition and follow the signal, happened twice to me today...after i got into position of YM couple of seconds later the signal alert visual and audio follows
eventually i'll let MC do the trading...after 2.1 version maybe.
- Attachments
-
- picture1.png
- YM (rising three methods) bullish continuation patterns
- (54.91 KiB) Downloaded 932 times
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Genetic Optimization vs. Exhaustive (Brute Force) Optimization.
1. Genetic Optimization is hundreds of times faster. Calculations that might take years and even centuries in exhaustive search would only take days in genetic. Exhaustive optimization goes through all the possible combinations which usually takes a lot of time (especially if you are adapting many rules and parameters). Genetic optimization, on the other hand, only focuses on improtant areas of solutions space. It quickly narrows the number of combinations to be calculated by focusing on the areas that are most profitable and most stable.
2. Genetic optimization helps to avoid the curve-fitting trap. Exhaustive search yield the optimum results under very specific conditions which might not necessarily occur again. Genetic optimization provides more universally applicable solutions.
Genetic algorithm is a widely used search technique and there are a lot of sources covering the technique. You could research this issue further using the multitude of sources.
As for the best parameters for genetic research in MultiCharts - reading the relevent section of MultiCharts Help could be a good start.
1. Genetic Optimization is hundreds of times faster. Calculations that might take years and even centuries in exhaustive search would only take days in genetic. Exhaustive optimization goes through all the possible combinations which usually takes a lot of time (especially if you are adapting many rules and parameters). Genetic optimization, on the other hand, only focuses on improtant areas of solutions space. It quickly narrows the number of combinations to be calculated by focusing on the areas that are most profitable and most stable.
2. Genetic optimization helps to avoid the curve-fitting trap. Exhaustive search yield the optimum results under very specific conditions which might not necessarily occur again. Genetic optimization provides more universally applicable solutions.
Genetic algorithm is a widely used search technique and there are a lot of sources covering the technique. You could research this issue further using the multitude of sources.
As for the best parameters for genetic research in MultiCharts - reading the relevent section of MultiCharts Help could be a good start.
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Hi,Genetic optimization, on the other hand, only focuses on improtant areas of solutions space. It quickly narrows the number of combinations to be calculated by focusing on the areas that are most profitable and most stable.
Could you possibly tell me/illustrate how the important areas of solution space are determined for currency or equity markets? This seems to me to be the critical component of your method because if your "subset of solution space" is, in spite of what you perceive, non-critical, then the data generated by the search, although correct, will not be useful.
TIA
lj
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Hello,
If you are talking about the basis for finding the best solution area it is determined by the user (e.g. net profit etc).
If your question was about how the algorithm employed by the genetic optimization finds the best solution areas please refer to the sources mentioned in the previous posts. Genetic optimization is not exactly 'our' method. It is based on a widely known and long-used algorithm. Please refer to the previosly mentioned sources for more detailed info.
If you are talking about the basis for finding the best solution area it is determined by the user (e.g. net profit etc).
If your question was about how the algorithm employed by the genetic optimization finds the best solution areas please refer to the sources mentioned in the previous posts. Genetic optimization is not exactly 'our' method. It is based on a widely known and long-used algorithm. Please refer to the previosly mentioned sources for more detailed info.
Last edited by Marina Pashkova on 13 Aug 2007, edited 1 time in total.
GO
It was the second possibilty I was asking about and was requesting some feedback on how you use GO in your protocol. I do appreciate that the request may have crossed the border into "proprietary info" and if so, so be it. I am a firm believer in the "garbage in - garbage out" paradigm and hence the question. I am not, of course, suggesting that what you do is garbage because I don't know what it is that you are doing with GO. Search optimization has been a hot topic for some years now and it will be interesting to see how well this particular "cross-cultural" methodology works. The data sets are a priori rather different and are modified (updated) in different ways. A critical question, IMO, is how compleat your base "financial" set is or put another way, what is the universe of data on which you perform your GO algorithm?
Thank you for your considered response,
Sincerely,
lj
Thank you for your considered response,
Sincerely,
lj
- Marina Pashkova
- Posts: 2758
- Joined: 27 Jul 2007
Hello,
Your question did not cross into the borders of 'proprietary information'.
There is nothing special about what we 'do' with Genetic Algorithm. In our program it is employed exactly the way it would be elsewhere.
You could look up information on GA/GO on the web you will get more details on how it works and what its advantages over Exhaustive Optimization are - this info would be more detailed than I could provide here within the framework of this forum.
I could also suggest a simple test: apply a strategy to a chart and first run Brute Force (Exhaustive) Optimization and Genetic Optimization and compare the results and how long each of them has taken.
Your question did not cross into the borders of 'proprietary information'.
There is nothing special about what we 'do' with Genetic Algorithm. In our program it is employed exactly the way it would be elsewhere.
You could look up information on GA/GO on the web you will get more details on how it works and what its advantages over Exhaustive Optimization are - this info would be more detailed than I could provide here within the framework of this forum.
I could also suggest a simple test: apply a strategy to a chart and first run Brute Force (Exhaustive) Optimization and Genetic Optimization and compare the results and how long each of them has taken.