Random Flow Theory and FOREX - page 31
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
Candid grasn
Here's the thing: if I said that, before you all would have thought that Prival was just crazy. I want you all to come to the same conclusions as I did, it's simple logic explains a lot.
I will be pushing with questions.
After all, sampling frequency is related to sampling period by the formula Fdisk=1/ delta_t. Delta_t is nothing more than the data period (in maths terms "tick lag"). Ask him tick lags is the sl. value (as long as the type of distribution law is not important). If the mathematician says YES, then answer the sampling rate will also be a random variable ?
Second. As an expert in DSP.
Try to imagine that the instrument (which measures the MT4 price) has Kotelnikov's theorem fulfilled. And tell me what a gap would look like. For example a gap from Friday to Monday (and then extend that to a gap during a news release).
Note that you will come to these conclusions, it's just that I am pushing YOU all to them.
It gets even more interesting :-)
P.S. On topic, on topic...
P.P.S. Prival, YES, this is a highly unsteady s.p. What will you fill in the missing samples with if you want to arrive at a single sampling rate? Or will you model the frequency change?
P.P.S. Prival, YES, this is a highly unsteady s.p. What will you fill in the missing samples with if you want to arrive at a single sampling rate? Or will you model the frequency change?
to Prival
In the "hardware" part, as well as in DSP itself, I am self-taught and no expert...
It always seemed to me that the sampling rate, in simple terms, is the number of measurements per unit time, by some tricky device. In general, it's operator/designer-controlled and is chosen based on the required quality of the OUTPUT signal digitization.
The sampling rate should be twice as high as the highest frequency component of the OUTPUT, i.e.
Fd>2*fmax >>>> Everything else is, by and large, bullshit.
In my stupid mind - this does NOT explain the gap and everything else, including the world does not govern in any way. And the fact that the fmax will be random is no big deal, it's already understood and also doesn't help in any way.
PS: The formula Fdisk=1/delta_t, is slightly incorrectly interpreted. "Delta_t" is not the arrival period but the sampling period , and these are quite different things (!!!). In other words, the original signal is replaced by a lattice of the form x(n*T). i.e. the moments of data sampling are defined, not the arrival of data
to Prival
PS: The formula Fdisk=1/delta_t, is a bit incorrect. "Delta_t" is not a data arrival period, but a sampling period, and these are quite different things (!!!). In other words, the original signal is replaced by a lattice signal of the form x(n*T). i.e. the moments of data sampling are defined, not data arrival
I consider myself an expert in DSP, for an experiment, so that you understand what I'm talking about. Draw a sine wave and make 2 counts per period. By Kotelnikov's theorem this sine wave can be reconstructed. Now imagine that you don't know the sampling rate (=period=interval between samples). Try to reconstruct a simple sine wave.
P.S. In a dead market we will be able to recover, roughly speaking, only the constant, and during strong news releases even some of the high-frequency harmonics. The market determines what we can do.
P.P.S. The problem is different. Kotelnikov's theorem says that a continuous signal under the right conditions is recoverable by the formula:
Everything is tip-top for constant delta_t. Under extreme non-stationarity of this quantity - how to modify the formula so that it optimally reconstructs the signal?
And another thing: even if we have perfect samples (ticks) with constant lag of 1 s at any time of day, we cannot reconstruct sine waves with period less than 2 s in principle. What to do on strong news when there is a high fraction of high frequency data? Our reconstructed function will be too low frequency in such conditions and will inevitably lag.
P.P.S. Prival, YES, this is a highly unsteady s.p. What will you fill in the missing samples with if you want to arrive at a single sampling rate? Or will you model the frequency change?
I consider myself an expert in DSP, for an experiment, so that you understand what I'm talking about. Draw a sine wave and make 2 counts per period. By Kotelnikov's theorem this sine wave can be reconstructed. Now imagine that you don't know the sampling rate (=period=interval between samples). Try to reconstruct a simple sine wave.
Prival, I am self-taught in DSP and have never concealed it. It may sound ridiculous, but I understand Nyquist frequency, sampling rate, and many other useful things. :о)) And I also understand that it doesn't rule the world, doesn't explain the market and the gap in particular. I think you are overdoing it a bit, probably from great knowledge.
I consider myself a DSP specialist, for an experiment, so that you understand what I'm talking about. Draw a sine wave and make 2 counts per period. By Kotelnikov's theorem this sine wave can be reconstructed. Now imagine you don't know the sampling rate (=period=interval between samples). Try to reconstruct a simple sine wave.
Prival, I am self-taught in DSP and have never concealed it. It may sound ridiculous, but I understand Nyquist frequency, sampling rate, and many other useful things. :о)) And I also understand that it doesn't rule the world, doesn't explain the market and the gap in particular. I think you are overdoing it a bit, probably from great knowledge.
I agree. There is no perfect number. And there probably won't be.