FIR filters - page 15
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
BandPass is the differential between the two LPFs (LowPass) similar to how the bastard MACD is derived from the difference of the mashes
if you want an order of 2500+ filter or a cutoff period of 300+, use Behemoth's filter.
but its filter is not suitable for all cases...
tried to do 1000 dB of damping or 0.000000000000001 dB of roll-off, but failed to achieve the same sharp resonance in the AFC as the oscillator filters
Didn't think anyone would be interested in such high suppression. The reason is most likely rounding error. One or two orders of magnitude higher, double besselAccuracy=0.00000001, and increase the resolution of the PI. It is possible to think how to calculate a little differently, but I'll pass here.Regarding the MAHA. ANY MA is the same FIR filter. The ONLY difference is the sliding window coefficients.
>> >>. For example, exp. The "MASKA" is BIX - its imp. characteristic is infinite.
Don't be silly. The period of the MA is the length of its TX
Don't talk nonsense. The period of MA is the length of its XI??? Read something about filters and then talk. For example on wikipedia.
EMA is a classical filter with an infinite impulse response. And it has no period per se. There is a feedback coefficient (which is why it is BIH, not FIR), which is recalculated from the period in the standard MA implementation in MT4. k=2/(period+1)
EMA with a "period" of 3 has a feedback coefficient of 0.5, and the result of filtering on series 1,2,3,4,5 will be different from filtering on series 9,8,3,4,5 or 3,4.5. In case of FIR (simple slip) the result will always be the same.
FIR filter has no feedback coefficient, but a set of weight coefficients per sample (period). Coefficients of simple MA are all the same and equal to 1/period. Other FIRs may have different coefficients, as long as their sum equals 1. Background outside this sample has no effect on the result. Therefore it is FIR.
Just in case, here are the links on Wikipedia:
CIH, BIH
Just in case, here are the links on Wikipedia:
KIH, BIH.
I was wrong, I got it wrong. Sorry, didn't see the word exp. in your post.
Didn't think anyone could be interested in such high suppression. The reason is most likely a rounding error. You could increase the accuracy of the Bessel function calculation by one or two orders of magnitude double besselAccuracy=0.00000001, increase the PI bit. It is possible to think how to calculate a little differently, but I'll pass here.Yes I did, I set it straight away:
besselAccuracy=0.000000000001
pi=3.14159265358979323846
but it's OK, I need a 600-period cutoff with a smooth frequency response.
Yes I did, I set it straight away:
besselAccuracy=0.000000000001
pi=3.14159265358979323846
but it's OK, I need a 600-period cutoff with a smooth frequency response.
Well, I was actually wondering where the error might accumulate at high order. Because with a transition zone of 0.2% of Fd and 60dB suppression you already get an order of 1800. But what if you need more suppression? One more possible source of error is coefficient normalization. In general, it is interesting how MQL calculates mathematical functions.Yes I did, I set it straight away:
besselAccuracy=0.000000000001
pi=3.14159265358979323846
but that's OK, I needed a 600-period cutoff with a smooth frequency response
maybe it's because of the inaccuracy of the PI number setting, I encountered this problem once and the error was building up because of it - here's the 'Pi' solution
see if that's where the problem lies.
I was wrong, I got a little excited. Sorry, didn't see the word exp. in your post.
Bullshit question - let it go. >>)))