Subsystem "Asset Management" - page 4
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
Nah, not hers. ^_^
Its her :), meant the task, not Markowitz :)
Nah, not hers. ^_^
Hi Sergei !
What did you write about? Markowitz and the Nobel Prize. Markowitz is him, the Nobel Prize is her. And the right one is me. :-)
to sol
Почему-бы не применить генетический алгоритм поиска оптимального решения?
Of course, you can apply a genetic algorithm, but I haven't decided which chromosome you want. Just in case - just kidding, otherwise they will take it literally :o) Honestly, creating this topic I was expecting more help than a simple listing of fancy words. We all know a lot of different words - why remember them "in vain". Unfortunately, I am very superficially acquainted with this kind of optimization technology, and if you can help me to set the task correctly and explain at least the basic idea of using a genetic algorithm for my specific task - I would be very grateful. Note, I am not asking to solve it for me (but I will not refuse it either), I am asking to help little more than write the phrase "genetic algorithm". :о)
I believe it is only the beginning - and you will soon outline the main approaches to solve the problem using these very genetic algorithms.
to thecore
Yes, you got it right. You're looking for differences MA[i+1]-MA[i] like MA[2]-MA[1] or even MA[1]-MA[0]. As I said before, the use of MA is not obligatory. What matters is the principle of prediction.
Well, yes, this is what is called frequency analysis, a very interesting and useful thing. Advanced implementations of this kind of analysis have been based on using NS for a long time, and it's one of few cases when using NS is really justified and useful (strangely enough NS is, roughly speaking, much better at searching patterns than predicting :o).
I strongly suspect, given the type of distribution, that such "clouds" for each unique value will not be particularly informative on the bulk of observations, i.e. stable patterns of behaviour after a particular value are unlikely to be found. Although, perhaps for quantities lying closer to the tails of the distribution such clouds may already have a more meaningful appearance. One more "though" - perhaps there is a certain dependence of the series themselves, i.e. for example "more likely now what happened yesterday will happen again".
I'll have to have a look. Thanks for the description of an interesting type of prognosis, it is quite possible that its elements will be implemented in my system - extra "brain" won't hurt, and the model is not so foreign in relation to mine. I think I need a good classification. If you don't mind - I will specify some details here.
:о)
to Neutron
Hi Sergey.
You must be joking - if I remember, Markowitz got the Nobel Prize for this problem (the problem of optimal portfolio)! Do you want to find him on our forum?
Seryoga - Hi! Glad to see you again.
No - I'm not joking, Markowitz was solving a more complicated problem, estimating the total risk of the whole portfolio, and taking into account a lot of complex features. My task is much simpler. True, Markov chains are unlikely to work here, rather they can only be used within the framework of a single tool. But I think linear programming can be tried. Now I have built one third of the model, I understand how to do another third - I just haven't had enough time, and I simply don't understand how to formalize the remaining part yet.
Markowitz did not do the formulation of the problem himself (I think he was solving an already formulated, classic one), while I know this part of the problem perfectly, in the sense that I can always change it (formulation) :o))))
Besides, Serega, from an esoteric point of view you are not very correct. A man will never invent perpetual motion machine, knowing that it cannot be invented! But that's just philosophy. Take an example from Prival, he solved this problem simply and elegantly, keeping the solution to one small paragraph. And you are just joking.
to anubis
i>- Thanks, very valuable info on AR MA models! what's so difficult about it when you have an estimate of TP risk and time to reach it? plus you can try to plan future trades based on cumulative or current trades and it will give you a more or less objective picture of how many lots to open and which ones to keep for future ones.)
I am glad i could help. Only I did not manage to specify one more feature. Increasing the order of a model usually leads to increasing error. But taking into account how these models predict, as well as the practical impossibility of their good identification on price series, one can not bother with such subtleties.
As for the problem - join and it will become clear at once.
to Yurixx
What did you write about? About Markowitz and the Nobel Prize. Markowitz is him, the Nobel Prize is her. I am right. :-)
Yuri hi. Frankly speaking - I was confused too, but you cleared up everything in time with simple and understandable - "I am right". :о) Maybe you at your leisure, though, can help with a simple task? Cause Seryoga's frightening me with some kind of "Schnobel" :o(
:о)
Truth be told, grasn, I also expect something more from you than just indications of some kind of wave analysis to predict the zigzag.
I can suggest you look at this link here:
http://www.aridolan.com/ga/gaa/gaa.html
I dare say your problem can be formalized as https://en.wikipedia.org/wiki/List_of_knapsack_problems
An example of source code for solving this class of problems can be found at the first link.
to sol
Of course, you can apply a genetic algorithm, but I haven't decided which chromosome you want to use yet. Just in case - just kidding, otherwise they will take it literally :o) Honestly, creating this topic I was expecting more help than a simple listing of fancy words. We all know a lot of different words - why remember them "in vain". Unfortunately, I am very superficially acquainted with this kind of optimization technology, and if you can help me to set the task correctly and explain at least the basic idea of using a genetic algorithm for my specific task - I would be very grateful. Note, I am not asking to solve it for me (but I will not refuse it either), I am asking to help little more than write the phrase "genetic algorithm". :о)
Genetic algorithms are 99% of the time inferior to more narrowly focused algorithms.
In any case, in order to apply them, as in solving an optimization problem proper, you need a target function.
You mentioned constraints, that's clear, but not a single specific word was said about the target function.
And what help do you want without specifying the problem?
A simple statement -- profit seeks to maximize -- won't suffice, because your model, like any other, will be inaccurate -- which will manifest itself in a difference in behavior inside and outside the sample.
The target function must also include drawdown and percentage, at least.
to sol
По правде говоря, grasn, от Вас я тоже ожидаю чего-то большего, чем просто указаний на некий волновой анализ, который позволяет прогнозировать зигзаг.
Where did you ask about this "more"? All you wrote was one word - "Interesting", I couldn't find anything else. As for my forecasting model, I'm really not ready to discuss it openly now, I've already discussed it in a closed forum. But I've described some of my alternative approaches and thoughts about forecasting, if something else comes to mind or comes to my mind I will tell you about it, and I'll tell you about it in details, as always.
I don't have any objections to it, if it was abrupt, I apologize for that, I didn't want to offend anyone. Thank you very much for the links, I will certainly look them up.
to TheXpertYes, I haven't got to the mathematical formalism yet, as I wrote above - now I'm preparing the second iteration, more detailed, just in terms of linear programming. But if you read "What I'm asking about" in the first post, you'll see, I'm asking about everything. :о)
A simple statement -- profits tend to the maximum -- isn't enough, because your model, like any other, will be inaccurate -- which will manifest itself in a difference in behaviour inside and outside the sample.
Nonsense, the target function is stated loud and clear! "Increased profits" is ABSOLUTELY a self-sufficient target function. Everything else is just a constraint. All that you have "screwed" to increase profits, any interest, drawdowns, etc. and what I wrote about are (once again) just constraints, nothing to do with the target function. They must not be included into the target function in any case. Some chosen optimization system (be it linear programming, genetic algorithms, ...) will find the best solution under the conditions that are limited by you (allowed for you personally percentage, drawdowns, ..., etc., for me they may be different) and your brokerage company (number of trades, lot increment, maximum lot size, etc.).
Limitations you mentioned, that's understandable, but about the target function not a single specific word was said. And what help do you want without specifying the task?
It was said in LP terms, - the target function has already been defined and written (and don't go to the doctor), it's time to start helping, if you want to help :o)))))))))))))) Just kidding, because everyone is so super serious.
PS: Listen, how do you make breaks between paragraphs? I can't do it, it all collapses. Teach me, please.
Nastily, the target function is voiced loud and clear! "Increased profits" is ABSOLUTELY a self-sufficient target function. Everything else is just a constraint.
Good luck. I have a pretty good idea of what I'm talking about.
PS: input.
PPS: I'm on the chanterelle.
That was an overreaction. All you have to do is to choose the percentage of the deposit, at which the maximum acceptable results are achieved.
It should be chosen on the basis of known parameters of the system when it works without MM. It is not like I want to win a Nobel Prize. At least in this simple variant.
I'm not touchy at all ;)
And the closed forum is off limits to me. I am a spy and an enemy of the tester.
Good luck. I have a pretty good idea of what I'm talking about.
PS: input.
PPS: I'm on foxy.
I have no doubt, in many ways we are the same - we know exactly what we are talking about. :о)
similarly - good luck
I'm not touchy at all ;)
And the closed forum is closed to me. I am a spy and an enemy of the tester.
And who is this "tester"? Apparently, according to your attitude towards him, this tester is a big poo-poo. :о)