Machine learning in trading: theory, models, practice and algo-trading - page 2276
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
https://www.google.com/search?q=Piecewise+Aggregate+Approximation&oq=Piecewise+Aggregate+Approximation&aqs=chrome..69i57j0j0i22i30l5.177j0j7&sourceid=chrome&ie=UTF-8
https://www.google.com/search?q=Piecewise+Aggregate+Approximation&oq=Piecewise+Aggregate+Approximation&aqs=chrome..69i57j0j0i22i30l5.177j0j7&sourceid=chrome&ie=UTF-8
I need a way to be invariant to time, amplitude, frequency
how does logarithmic help to recognize the patterns I drew above?
That's what I thought...
you can proxy anything, I need a way to be invariant to time, amplitude, frequency
first approximation, then normalize
first upprox, then normalize
how do you propose to normalize time?
How do you propose to normalize time?
are you stupid? for time upsampling or downsampling via interpolation/approximation
Are you stupid? For time upsampling or downsampling via interpolation/approximation.
No...
interpolation/extrapolation is the same as doing it with different sized windows, it's the same fucked up.
Do you know how much time it takes? At each iteration.
how does logarithmic help you recognize the patterns I drew above?
Just logarithm each of these pieces. Meaning, go from a series X(t) to a series log(X(t)). Then you have to remember the shift. And then you can already DTW.
No...
interpolation/extrapolation is the same as screwing with different sized windows, the same losses.
Do you realize how much time it takes? For every iteration.
What's it all for?
Night fluctuations in 2 hours and day fluctuations in 20 minutes may be similar in form, but they have different causes, different driving force, different patterns, different participants.
And the daily 20-minute and 3-hour fluctuations are also different.
I think only commensurable ones should be combined as one pattern, for example if they differ by no more than 30-50% in time and/or amplitude.
And why all this?
Because there are no equally repetitive patterns on the market... in a fixed scale exactly, I want to check in another scale