Correct calculation of currency indices. - page 7
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
I should have put chaos in quotes.
of course we are limited by the sampling rate, spread and speed of situation estimation (if it's manual trading)
There is also such a thing as signal to noise ratio. The most important and detrimental indicator of the received signal in radar. So in my opinion, on larger timeframes, it is large it is good, so there and it shows better useful signal.
There is also the concept of signal-to-noise ratio. The most important and detrimental indicator of the received signal in radar. So, in my opinion, on larger timeframes, it's a good thing, that's why the useful signal appears better there.
I believe that there is no noise in the market. There is a lack of sampling. If we subject the tick history to spectral analysis, the result will be the same as on higher TFs.
There is also the concept of signal-to-noise ratio. The most important and detrimental indicator of the received signal in radar. So in my opinion, on larger timeframes, it is big is good, so it shows better useful signal.
but on longer timeframes we need to use big stop losses ((
Strong spikes in the minute bars deprive the higher bars of relative power (signal/noise) - it can be seen on my picture
I believe there is no noise in the market. There is a lack of discretisation. If we subject the tick history to spectral analysis, the result will be the same as on higher TF.
Unfortunately, it is not that simple.
ticks show more speculative factor, while weeks show more fundamental ones
If you want to make a profit on forex, you have to decide 3 questions. 1) When to open? 2 in which direction? 3 When to close?
The market has periodic harmonic components, but the process is not stationary. (Proved by Associate Professor A.O. Sivertsev.)
Non-stationarity changes smoothly. (My personal observations)
If changes in stationarity in the measurement segment are not great, they can be neglected. (Since even an approximate calculation can show points of extremums).
In brief, it looks like this. In more detail you are invited to other theme where this question is discussed.
In my opinion, in green I have highlighted the most important things that exist in any market and that should be used in trading.
In red I highlighted the wrong approach in spectral analysis. I believe that we should look for patterns of change in stationarity or non-stationarity.
In essence, one needs a spectral analysis of changes. But solving this problem is fraught with consequences... If we solve this problem at this level, we will include changes of a higher order. It all comes down to processing power.
If you solve it, please do not shout it everywhere!
I believe there is no noise in the market. There is a lack of discretisation. If you subject the tick history to spectral analysis, the result will be the same as on higher TFs.
No it isn't. It's completely different. You analyze the function with a different sampling rate, hence the spectrum is different. Take a sine wave and experiment. And you have to look at the spectrum, when a pure sine wave goes to the input, and then feed the same sine wave with a 4-point accuracy, i.e. simulate the way a DC quotes a sine wave with an amplitude of 10 points and a quantization error of + - 1 point. And you will see noise !!!
In essence, a spectral analysis of change is needed. But there are consequences to solving this problem... If we solve this problem at this level, we will have higher-order changes included. It all comes down to computational power.
The computational cost of doing this is not that high. Look for FFT method (via periodic compensation) here http://www.diclib.com/cgi-bin/d1.cgi?l=ru&base=bse&page=showid&id=80887.
The idea is simple: you get a new signal, subtract an old one from it. If there is no change it will be 0. If there is a change it will show up in the spectrum.
The computational cost of doing this is not that high. Look up the FVC method (via periodic compensation) here http://www.diclib.com/cgi-bin/d1.cgi?l=ru&base=bse&page=showid&id=80887.
The idea is simple, get a new signal, subtract the old one from it. If there is no change, it will be 0. If there is a change, it will show up in the spectrum.
Yeah, I've got it. It was much simpler in my case.
Only after all the calculations you can get non-stationary changes again and then... You need computing power for that...
MT4 can't handle it.