1st and 2nd derivatives of the MACD - page 41
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
A good filter has some meaningful characteristics that can be used for something. A bad filter has no such characteristics. Somewhere above I posted the characteristics of a MACD and a normal filter.
Is there a speedometer in the car for the forecast?
I just now looked closely and noticed a huge phase delay. You have to shorten the filter. The AFC will of course deteriorate. So the problem is: having a certain filter length (which will determine its group delay = filter length / 2), imposing condition of symmetry of its coefficients (which will give us a linear phase), find these coefficients to get maximum attenuation outside the bandwidth. And why not use the known weight functions?
I remember composing this filter turkey a long time ago
https://www.mql5.com/ru/code/11183
I particularly liked the Hannah window. I'm attaching a corrected version of the turkey. Here is the result (red - Hahn, blue - Blackman, green - Natal):
You can see the group delay equal to (Per-1)/2 where Per is the filter length.
You can shorten the filter, or you can do something else. For example, move it back in time.
From this it should be clear how much it can be shifted and why the overshoot will be minimal and probably not noticeable to the eye at all.
Weighting functions should be used, otherwise you will just get SMA. I can even say that I use the Blackman-Hann window.
And then there are BIH filters, but for my purposes they are not suitable.
A filter is needed for prediction. what does it have to do with meaningfulness?
There are many ways to make predictions. You can differentiate and do something like polynomial extrapolation, you can guess in your mind where it will go next, you can even use (18) I guess :)) Someone needs minimum phase delay, someone needs linear phase, someone needs deep suppression of high frequencies. These are the meaningful characteristics.
There are different ways to make predictions. You can differentiate and do something like polynomial extrapolation, you can guess in your mind where it will go next, you can even use (18) I guess :)) Someone needs minimum phase delay, someone needs linear phase, someone needs deep suppression of high frequencies. These are the meaningful characteristics.
I have not seen anything like that in the market. We predict the level (absolute price value), the increment, the direction, the reversal, the volatility. And there is a very high probability of correctness of such forecast, its error. The last two characteristics may be predicted if the quotient is stationary, but it's non-stationary. What does all this have to do with filters?
Of all the above, I'm only interested in the direction. I have a different opinion about the stationarity of the quotient. I work with the chart as a signal, while I think you wrote somewhere that there is no signal in the market. So it is unlikely that we will ever understand each other.
Of all the above, I am only interested in the direction. I have a different opinion about the stationarity of the quote. I work with the chart as a signal, while you seem to have written somewhere that there is no signal in the market. So it is unlikely that we will ever understand each other.
In radio electronics any time curve is a signal.
Yes. But there's a nuance.
There is noise and there is a useful signal.
Separating them is a major problem, especially if you don't know the criteria for usefulness, the variability of those criteria and the variability of the noise.
A cotier is not in itself a useful signal.