Algorithm Optimisation Championship. - page 95
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I was helped to understand it to a sufficient extent by a text problem.
Some time ago I described my understanding of optimization in the Open Championship, and as there was no criticism or any corrections from the topicstarter, I decided that I understood the optimization correctly. Here is my understanding of the problem: https://www.mql5.com/ru/forum/89230
Your approach and that of Yury are both "lacking" a very specific view of optimization problems, it lacks abstractness and universality. This may lead to unsatisfactory results when FA is unknown, which is 99% of optimization problems in trading.
Well you're wrong there. My too specific view (that's when the function is known) is only for warming up and visual verification of user code. Then the algorithm is trained in "shenerize FF" mode randomly.
The user algorithm does not know
- how many parameters the function has, maybe 3 or maybe 500
- the function: is it a formula, an array of numbers or is it just a set of letters to compose a sentence
Here's the code for the panel. It's very simple. I can add my own buttons and fields in MT4.
Please note that I wrote it for myself and in no way pretend that it serves as a model. This is just a sample design of a competition panel with nothing but graphs and buttons.
Due to some perturbations in the preparation of the participants for the championship and the general discussion, it has become necessary to outline the actual procedure for the championship.
In a few days I will provide the final versions of the source code for the FF library, the FF key generator and the executing script. I need time to think over the nuances, so please be patient and wait a little longer.
Next, the actual championship will take place in the following sequence:
1. Publication of the source code of the FF library, the FF key generator and the executing script. This is to show how everything works, to ensure transparency and objectivity in comparing participants' algorithms, and to make sure that participants cannot adapt their algorithms to the FF problem.
2. Reconciliation. Discussion and answering questions about the FF.
3. Publication of algorithms by participants in this thread. I recommend to apply algorithm protection in the form of 7 days validity limit.
4. Immediately after p.3 referee in the person ofAnatoli Kazharski (I sent him a message with a request) will download the source codes of the FF and the key generator, generate a key and insert it in the FF code, and then compile the library. After 3 days (to ensure that participants can no longer edit algorithms) from the date of publication of the algorithm by the last contestant, the referee will post the library in this thread.
5. After clause 4, everyone who wishes can download participants' algorithms and FF library, and conduct tests by himself. In the meantime I will also conduct tests and post the table with the results.
That is all.
Suppose we start with only two participants, but maybe people will get interested and then it will be possible and more interesting and spectacular to repeat the steps of the championship again, but for a larger number of participants.
Let's give it a try. Let me know in a private message when everything is ready.
Let's give it a try. Let me know in private messages when it's ready.
Vasiliy Sokolov
Pull up. The action begins.
Pull up. The action's starting.
I'm here.
Can you suggest functions like f(x1, x2) to be included in the FF along with mine?
I don't quite understand you.
If I suggest FF functions along with yours and we include them in the FF, it turns out that we will know a formula that we should not know...
Isn't the referee supposed to make up the FF function?
Please clarify.
I don't quite understand you.
...
Explain, please.
Again?
I don't quite understand you.
If I suggest FF functions along with yours and we include them in the FF, it turns out that we will know a formula that we should not know...
Isn't the referee supposed to make up the FF function?
Please clarify.
Of course the functions have to be provided by the participants, the thread says so. I have announced to collect functions from participants, but so no one has provided any.
I will select 3-4 different functions of the form f(x1, x2), from them will be compiled FF for the contest. The referee has to generate a sequence of parameters and a sequence of calls to these functions, obtain unique sequences. Even though I know the functions, and the contestants will know them, no one will know the final total function. This was discussed earlier in this thread.
There is now a last opportunity for you to add one or more of your own functions to those I have compiled. If you don't remember what we are talking about, search for the keyword "f(x1, x2)" in this thread.
Again?