Machine learning in trading: theory, models, practice and algo-trading - page 1011
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
Mostly standard indicators.
How do you set the indicators? By default, or something else?
And how are the indicator settings taken? By default, or something else?
I do it empirically, i.e. intuitively.
Empirically, that is, intuitively.
And if we assume that out of the set of all settings, there are those that describe the market better and worse?
Maybe it is necessary to evaluate the settings from the beginning?
Mostly standard indicators.
I already asked what indicator/indicators should ideally be.
Well, except for the "chop dough" button. ))
Here Igor Makanu suggests testing on a non-random process. Maybe.
Forum on trading, automated trading systems and testing trading strategies
Is it possible to become a Ruble millionaire with 1 dollar in a cent account?
Igor Makanu, 2018.07.04 19:33
I've been working on this for a long time now.
If you don't mind, here are the NZDUSD quotes, they are not random (the generator is based on Weierstrass function - I liked it and it's easy to implement), pick up the indicators so that everything would be +.
https://cloud.mail.ru/public/Kqtf/NA3JPyr7i
ZFS: Disconnect MT4 from the server and switch the NZDUSD timeframes to mine, but I doubt that within a couple of hours you'll be able to pick up a trading strategy, even on random quotes, neuronets are not interesting
;)
On the other hand why should I invent a story if it already exists?
But to formulate the criteria of prediction ability evaluation, apparently, requires a deep understanding of the market.
For a correctly posed question is at least half the answer.
And if we assume that out of the set of all settings, there are those that describe the market better and those that describe it worse?
Maybe it is necessary to estimate the settings in some way from the beginning?
And this was also investigated) The market is generally difficult to describe, if possible. It also depends on what markets.
I do my research on forex. I have money for other markets, I will study them.
I have already asked what indicator/indicators should ideally be.
Well, except the button "chop dough". ))
Here Igor Makanu suggests testing on a non-random process. Maybe.
On the other hand, why make up a story when you already have one.
But to formulate the criteria for evaluating the predictive ability, apparently, will require a profound understanding of the market.
If you ask the right question, at least half the answer.
I don't understand this function and I don't use what I don't understand. I basically have enough of my own ideas.
And this was also investigated) The market in general is difficult to describe, if it is possible. And it depends on what markets.
I do research on forex. If I have money for other markets, I will study there.
Can you tell me how I have evaluated it? Or is it a trade secret?
I just mean that there is the ideology that thinks that any average settings will do - as an example, the MA and there are purely imperial claims, like "МА 100 is twice as fast as МА 200 and I need the average - so I will use МА 150" and there are those who will take 144 because analysts like it and therefore it is used by the masses for their purposes. I am interested in a mathematical choice, for example, МАХ divides the market most evenly in half, so I will take it or some other condition. For example, my Expert Advisors/signals with the word Raznica in their names, except for Raznica 01 and Raznica 02, work on the channels whose settings have been found exactly by analyzing the price behavior relative to the indicator (and I believe that the price and its derivatives as indicators affect each other) for the past 6 years. I know such three different approaches (there is another one - to make a primitive strategy on one indicator and look for the best results through optimization, and then take these settings for predictors), that is why I wonder which one you use.
From the series "funny pictures": the forest is 15 deep.
And what about outside of training, but like this)
Well, what's interesting...?
Are they predictors, maybe they were not. 1000 pages spelled out, and they couldn't even show such still lifes.
If anyone is interested, to the question of overtraining, here's a test with a forest of three deep.
Well this is for those who are not readers, and for the rest there were not only examples of reports, but examples of advisors...
https://www.mql5.com/ru/forum/86386/page649#comment_6508250
Can you tell me how you evaluated/tuned it? Or is it a trade secret?
I just mean that there is an ideology that says that any medium settings will do - as an example MA and there are purely imperial statements, like "MA 100 is twice as fast as MA 200 and I need the average - so I'll take MA 150", and there are those who will take 144, because analysts love it, and therefore use it for their own purposes. I am interested in a mathematical choice, for example, МАХ divides the market most evenly in half, so I will take it or some other condition. For example, my Expert Advisors/signals with the word Raznica in their names, except for Raznica 01 and Raznica 02, work on the channels whose settings have been found exactly by analyzing the price behavior relative to the indicator (and I believe that the price and its derivatives as indicators affect each other) for the past 6 years. I know about three different approaches (there is another one - to create a primitive strategy on one indicator and look for the best results through optimization, and then take these settings for predictors), that is why I want to know which one you use.
I have used main indicators from standard delivery set as predictors and that's it, I seldom changed periods, if I don't change them, the picture will not change much.
Is there any filtering in the channels for breakdown or rebound?
The main indicators from the standard delivery set as predictors and that's it, I rarely changed the periods, if you do not change them, the picture will not change much.
If I have not changed them, the picture will not change much.
They have a simple ideology, but with a gradual increase of position. It is almost a grid by the channel but in fact the grid rarely gains many positions. I use my own indicator for the channel, but its similarity will be the usual МАшки with an indentation of x points. The system is counter-trend so the first thing I do is to determine the trend that has formed and then I start to open positions. Of course there are some filters, I use a lot of them and gradually I try to use them in MO. However there are two types of filtration - for opening a position and for continuation of a position and there is also MM with profit forecasting (profits are floating, not fixes). The Expert Advisor is not simple, but the ideology is simple. About filters, there are 9 basic settings for filters that are relevant to all pairs (I use 22 of them in trading), the optimization task is to find the optimal filters (which of 9 we will use) - there are 512 variants and there is something to choose from. Why do I use optimization in general? Different pairs have different characters, some like to twitch, some like to be calm, and some prefer to roll without rolling back. And this dynamic remains on the history. And each filter struggles with certain peculiarities in the behavior of currency pairs.