Machine learning in trading: theory, models, practice and algo-trading - page 2153

 
Renat Akhtyamov:

I don't read it and I'm not going to download it.

Since.

no one here in this thread has managed to earn

like I do, today and always:

//that's why it's too early for me to tell you what to do.

// asked for more details, I did it.

// that's all there is to it

what do you always say?

Show me for the last two months, you don't have to show me for always.

 
mytarmailS:

What do you always say?

show me the last two months, don't always.

i showed you how fnc is trading

the second system has a stack unlike any other.

it's always on the plus side, no drawdown

the strategy is described in general terms in response to Maxim's post

it has volumes of buys and sells and their prices

i have never shown the monitor screen to absolutely no one, let alone the state

in the early stages of strategy development I posted some screens and stats on the forum, starting in 2014

If you're interested, you can find

 
Renat Akhtyamov:

showed how fnc is traded


Describe intelligently FNF, what is Y ? Where did you get this formula (link) from? Or an author's formula?

 
Vladimir Perervenko:

Describe intelligently ELF, what is Y ? Where did you get this formula (link) from? Or the author's formula?

X - input signal (price chart), in which we are not satisfied with the presence of high-frequency noise

Y - the output, not lagging the signal we are interested in, without noise

I attached the link where it's all from

it is very wise written there, i wrote easier above and only what i need to get the indicator in MQL

I will add

For the first calculation ( n=Bars) let's take (since the farthest bar is not as important as the current one)

Y(n-1)=iClose(n-1)

and then Y(n) to the current bar according to the formula

again

http://we.easyelectronics.ru/Theory/chestno-prostoy-cifrovoy-filtr.html

Честно простой цифровой фильтр / Теория, измерения и расчеты / Сообщество EasyElectronics.ru
  • we.easyelectronics.ru
Вы работаете с АЦП. Получаете результаты преобразования, один за одним. И замечаете, что эти результаты «скачут». А хотелось бы, чтобы стояли, как… Ну, короче, чтобы стояли! Есть много причин, почему отсчеты АЦП могут быть нестабильны. В своей заметке я не говорю об этих причинах. Я говорю о том, как успокоить показания, получая их AS IS. И как...
 
Maxim Dmitrievsky:

one more thing...

I almost always regret, when I ask a question here

the spread is formed by external factors, just like increments. what is the point of accounting for it. i have a queue after me))))

 
Valeriy Yastremskiy:

the spread is formed by external factors, just like increments. what is the point of accounting for it. follow me in line))))

What factors?
 
Renat Akhtyamov:

X - input signal (price chart), in which we are not satisfied with the presence of high-frequency noise

Y - the output, not lagging the signal we are interested in, without noise

I attached the link where it's all from

it is very smartly written, i wrote above in an easier way and only what i need to get the indicator in MQL

I will add

For the first calculation ( n=Bars) let's take (since the farthest bar is not as important as the current one)

Y(n-1)=iClose(n-1)

and then Y(n) to the current bar according to the formula

again

http://we.easyelectronics.ru/Theory/chestno-prostoy-cifrovoy-filtr.html

it works, but not ALWAYS.

 
Valeriy Yastremskiy:

works, but not EVERYWHERE.

agree

I usually don't post my systems if I have a better one

but in principle, even this one is not bad
 

I'll say it again, for the stupid.

There are points in the feature space. Some are to buy, others are to sell.

Suppose these points can move in such a way that the buy-sell sequence is not observed, i.e. the information about the spread in the dataset is lost.

the spread can be equated to the Euclidean distance between points, or between two classes of points

How to add this information. FNF, acceleration, and other stuff you can shove in your s. This is for clarity of perception, so to speak.

 
Maxim Dmitrievsky:

I'll say it again, for the stupid.

There are points in the feature space. Some are to buy, others are to sell.

Suppose these points can be mixed up in such a way that the buy-sell sequence is not observed, i.e. the information about the spread in the dataset is lost.

how to add this information. You can shove the LPF, acceleration, etc. up your ass. This is for clarity of perception, so to speak.

yep

And in this space there are only two of them:

1. purchases

2. sales

but given your sense of tact, you should only take the information from where you advise to shove it

and you're right.

It really isn't.

That's why I've been fiddling with it since November 2018.

the last so to speak hint, screenshots and a bunch of posts were, including the formula

However, the way is wrong.

right = use what you have

;)

Reason: