What is the optimum depth of history for identifying a useful signal? - page 21
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
You'll see when you let go.
You won't. As a doctor.
It won't let you go. As a doctor I say.
Why 32 bars, may be it only because of the computational load, what timeframe we should use to get something less or less using 32 bars, like H1 or H4? You cannot make 32-bars for 1 minute, 5 minutes or even 15 minutes if you forecast by a couple of bars only.
So why exactly 32 bars and exactly on one timeframe (by the way, which one do you use?)?
It may be any TF. And why should I not think to work with minutes? From the point of view of physics and mathematics all TFs are the same. By the way, good indicator performance only on certain TFs is, in my opinion, a sure sign of falsity.
The work is gradually moving forward. Here is what I have obtained so far:
64 bars of history are used to calculate it. We can use them to calculate 64 bars of the forecast but only the first 32 bars can be trusted, that's why 32 bars are displayed. They can lie about the amplitude, but only in its decreasing direction, i.e. it is possible to miss a profitable deal, but one cannot make a mistake about the direction. On the next 32 bars it is possible to invert the phase, so fuck it.
I would calculate more, but the machine cannot pull more, therefore 32.
To be very happy, we need a graph of external influences that cause deviations from forecasts but it is a very complex calculation that requires OpenCL...
The TF can be anything. And why shouldn't I think of working on minutes? In terms of physics and mathematics all TFs are the same. By the way, good indicator performance only on certain TFs is, in my opinion, a sure sign of falsity and bullshit.
The work is gradually moving forward. Here is what I have obtained so far:
64 bars of history are used to calculate it. We can use them to calculate 64 bars of the forecast but only the first 32 bars can be trusted, that's why 32 bars are displayed. They can lie about the amplitude, but only in its decreasing direction, i.e. it is possible to miss a profitable deal, but one cannot make a mistake about the direction. On the next 32 bars it is possible to invert the phase, so fuck it.
I would calculate more, but the machine cannot pull more, therefore 32.
What we miss here is a chart of external factors that lead to deviations from the forecast, but it's really ugly calculations, I can't do without OpenCL...
It's funny plotting the forecast on a graph;)
I'm interested in how it is implemented. Do you take the data from a file? But it's nice.
About the blue buffer. Doesn't seem to be an offset.
Funny construction of the forecast on the chart;)
Interested in the implementation of the plotting. Do you take the data from a file? But it is beautiful.
About the blue buffer. Doesn't seem to be an offset.
I do not quite understand the question.
Offset of what in relation to what? And why does it only apply to the blue buffer?
All data is stored in buffers, there are no files there.
Alexey, can you draw a histogram with a coincidence of forecasts (more than zero - coincided, less than zero - did not coincide) for some period.
PS about the forecast - you have guessed the direction of one bar - you will be in profit.
Alexey, can you draw a histogram with a coincidence of forecasts (more than zero - coincided, less than zero - did not coincide) for some period.
PS about the forecast - you have guessed the direction of one bar - you will already be in profit.
In general I can, but it's not very interesting. The indicator is not intended for permanent trading. It not only shows the direction, but also good entry points. I am interested only in these points.
But if we do not make just the histogram with/ without coincidence but calculate the difference between the real price and forecasted one, we will get the chart of external influences I wrote above. That's worth doing.
Actually, I can, but it's not very interesting. The indicator is not made for constant trading. It not only shows the direction, it also shows good entry points. I am interested only in these points.
But if we do not make just the histogram with/without coincidence but calculate the difference between the actual price and forecast, we will get the chart of external influences I wrote above. That's worth doing.