The market is a controlled dynamic system. - page 272

 
avtomat:
Prival:


Note the statement "

.

Letov-Kalman is inapplicable to non-linear objects of the third order... "

I will translate into simple language with sufficient accuracy for practice, it is possible to solve it to the third order, i.e. the first and second derivatives..... and sufficient accuracy for practice - nobody needs to know the movement of currency quotes to a hundredth of a tick, enough to a tick )))

)

You can solve it, go ahead)

)
And here I do not agree with you. It has nothing to do with ticks. Considering large TFs we deal with movement that contains both slow and fast components (small parameter in the higher derivative). One may give up on it and "ignore" them - but it will inevitably lead to a lag in the model response and a decrease in the accuracy of reproduction. So again we come to the question of model accuracy. And it's not about "
hundredths of a tick", but about large TFs - month, week, day.


this is where your mistakes lie

1 - week, month, day.... What are you talking about, the longer the interval, the harder the prediction (the error is directly related to time). I showed this maths somewhere, even drew a picture with an example. Briefly compare the model error of a simple straight line in two conditions, the error is one degree in the slope angle y=a*t+b even if the process is strictly linear, in a month the error is absolute + - three paws to the right of the sun )))), in 10-15 minutes will be within acceptable limits

2. I am reminded of a school course. If you remember from school, there are no slow and fast components in the movement, that's what traders of science think. Open the textbook any movement is expressed through the derivatives first, second, etc. ( and their change in time). Everyone went to school and hopefully remembers that this is velocity acceleration, etc. These are the coordinates in which the problem has to be solved

Z.U. So reminded, I have always been for ticks, I personally do not care what the rate will be in a day, month, year. Totally don't give a damn. It's enough to know what it will be in 1 minute, 5 minutes, 15 minutes.... the ACF is long ago posted herehttps://www.mql5.com/ru/code/8295 it clearly shows the time period over which you can predict the price series....

 
Prival:
to manage not the price, but your account by controlling actions (buy/sell)

Like this :

https://www.mql5.com/ru/blogs/post/753

 
Prival:


that's your mistake

1 - week, month, day.... What are you talking about, the longer the interval, the harder the prediction (the error is directly related to time). I showed this maths somewhere, even drew a picture with an example. Briefly compare the model error of a simple line in two conditions, the error is one degree in the slope angle y=a*t+b even if the process is strictly linear, in a month the error is absolute + - three paws to the right of the sun )))), in 10-15 minutes will be in the acceptable limits

2. I am reminded of a school course. If you remember from school, there are no slow and fast components in the movement, that's what traders of science think. Open the textbook any movement is expressed through the derivatives first, second, etc. ( and their change in time). Everyone went to school and hopefully remembers that this is velocity acceleration, etc. It is in these coordinates that the problem must be solved

Z.U. So reminded, I have always been for ticks, I personally do not care what the rate will be in a day, month, year. Totally don't give a damn. It's enough to know what it will be in 1 minute, 5 minutes, 15 minutes.... the ACF was posted here a long time agohttps://www.mql5.com/ru/code/8295 it clearly shows for which period of time you can predict the price series....


No, those are your mistakes.

1) Each TF has its own range.

2) Decompose the motion into its components, and you get different significant components of the motion, with frequencies spaced one or two decades apart, i.e. 10 to 100 times apart (remember spectral analysis) -- the same order of magnitude have differences in velocity and acceleration. (I remind you that frequencies and velocities are two sides of the same coin). With a high school course -- that's you overdoing it...

3) I have never been interested in tics.

And for the umpteenth time :I do NOT predict the price range !!!

 
avtomat:

And once again :I am NOT predicting the price range!!!

Buggahaha )) and you don't use TA, eh
 
avtomat:


No, these are your mistakes.

1) Each TF has its own range.

2) Decompose motion into its components, and you get different significant components of motion, with frequencies spaced one or two decades apart, i.e. 10 to 100 times, (recall spectral analysis) -- the same order of magnitude are the differences in velocities and accelerations. (I remind you that frequencies and velocities are two sides of the same coin). With a high school course -- that's you overdoing it...

3) I have never been interested in tics.

And for the umpteenth time :I do NOT predict the price range !!!


Yes two sides of the same coin and spectrum decomposes into components. But a spectrum is a spectrum, it says nothing about the movement. The model of movement, has always been given in derivatives, path, speed acceleration .....

1. This phrase is completely incomprehensible "Each TF has its own range." range of what. A set of clever phrases.... about nothing. If the range of movement, then I can personally show a daily movement (total of 3 ticks) and a minute candle from which any weekly one will faint ))), I can do the opposite...

2. I decompose and decompose and obtain the components of movement, which I describe in derivatives.... and the TF has absolutely nothing to do with it. Once again I go back to school textbook, there is motion, you are talking about it, and if there is MOTION it has first and second derivatives ...

3. And you shouldn't be interested, there is such a theory of measurement. Not it's probably complicated. Let me try a simpler example, you're in charge. You drive your car along a mountain road and measure your motion parameters discretely 1 time per hour or day (which candle you like best) + of course you drive at the same moment of time, for you do not know what is happening inside an hour (you drive blindfolded :-))). I think the result is clear, who has a chance to come to the finish line, the one who once an hour touches the wheel and opens his eyes, or the one who at any given time does it(every tick is looking at the road)

Z.U. Well don't predict, I predict. That is the reference trajectory. With quality forecasting, everything else is elementary.

 
TheXpert:
Boogahaha )) and you don't use TA, eh


TA as it is commonly known -- I do NOT use it. And you booger louder, because to you TA is perfection itself.
 
Prival:



Z.U. Well don't predict, I predict. That is the reference trajectory. With a quality forecast, the rest is elementary.


Let's leave it at that...
 

I have a couple more posts in theTrading Systems section of the blogs.

The market will open in 5 minutes - there will be new data, and with it the corresponding control system signals.

 
avtomat:
TA in common parlance -- I do NOT use it. And you boogeyman is louder, because to you TA is perfection itself.
Do you even know anything about me, you fool? )))
 
TheXpert:
Do you know anything about me, you fool? )))


And I don't want to know, "smarty-pants".