Machine learning in trading: theory, models, practice and algo-trading - page 1643
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
)))), at first it seemed to be worth it, but in fact it is automation of strategy tester, somewhere on the forum about 7-9 years ago, there was an article on the subject. Such a system would be terribly energy-consuming.
The result will still be probabilistic :)You don't get it... like everyone else... This is probably a good thing.
You don't get it... and neither do the rest of us... That's probably a good thing.
I guess you describe one thing and mean another, in which case it would be really hard to understand.
You don't get it... and neither do the rest of us... I guess that's even a good thing.
You're suggesting that instead of filter parameters, the meta-parameters of the adaptation algorithm will be meta-parameters.
You propose to change an awful lot - instead of filter parameters there will be meta-parameters of the adaptation algorithm.
I'm tired of explaining, if you all really do not see the difference ... that's bad, for all of you.
...
There is an indicator (it can be anything) (the man wanted ZZ)
The indicator has parameters that are not constant, while everyone trades constant parameters
1) We take and find out what parameters are adequate on the history at every moment of time, get a series of values
2) we learn to forecast this series
3) at the current moment we substitute the forecasted parameter in the indicator and become adequate to the market
4) trace the error between the forecasted parameter and the real one; if the error differs, then your "the moment, when the system stopped working".
not a word about the price here!!!!!
...
Similar principle I tried to develop in the topic "Algorithmic centrifuge"https://www.mql5.com/ru/forum/329078
With the use of neural networks the problem can probably be solved.
ZS. I read that thread and remembered the MAIN problem, which put me in a stupor:
Even on the history we could not find the ideal entry/exit points to match them with indicators. This is ridiculous.))) I was advised to use ZigZag, but if you think about it - it's the biggest stupidity, because it absorbs all the dynamics inside itself and instead of a strategy it creates complete nonsense, rather than a well-established trading concept...
Show me what you got in practice.
I will show you in a couple of days, I write a lot of code that is not typical for me, and I come up with non-standard solutions, it is long and takes a lot of effort
I tried to develop a similar principle in the topic "Algorithmic centrifuge"https://www.mql5.com/ru/forum/329078
Sorry, I couldn't focus and understand what it was about
Sorry, I couldn't focus and understand what I was talking about.
In fact, you suggested choosing and adjusting indicator parameters on the fly, and optimizing their values. The problem is similar in that topic - you should automatically select the best parameters (from the prepared sample) for the signal, checking their values in the "ideal points" on the history. If the values in the parameters are repeated at the points, it means that they are good for the TS.
The problem: the criteria of "ideal points" were not found. Therefore it is impossible to find these points on the history and therefore no "adequate" parameters for the TS signal.
Deadlock.
I'm tired of explaining, if you all really don't see the difference... that's bad, for all of you.
There is a difference. "There are no free lunches," but paying for them can be quite different.
...
The problem: the criteria of "ideal points" were not found. Therefore - it is impossible to find these points on the history and therefore - not to select "adequate" parameters for the TS signal.
...
I didn't put it correctly. To find "optimal points" for entry/exit (on the history) you can But this requires defining the criteria of a "good deal" and writing a special algorithm that analyzes the history in the context of the dynamics. It will select suitable trade areas and optimal points of entry/exit. Then, on these points we will test the parameters of the indicators, and if their values in the points are repeated, it means they really "follow" the market and can make a profit. Then include these parameters in the TS signal.