Cross Correlation lag issue

Dear Team;

i am conducting cross correlation between two signals using the following code but i am alwyas get the lag at which correlation is max at end of the vector

for example for if both P2 , P1 has length of 100 , lags 2 is 101

i did some preprocessing for the signal such as normalization , remove trend but still results do not change

when i used finddelay function it alyaws give zero

what do you think the issue is?

 [xc2,lags2] = xcorr(P2,P3,'coeff');
[rho2,lags2] = max(abs(xc2));

Answer 1

[Chunru]

P2 = randn(10, 1);
P3 = [zeros(5, 1); P2];   % P3 is P2 delayed by 5
[xc2,lags2] = xcorr(P2,P3);
plot(lags2, xc2)
[rho2,imax] = max(abs(xc2))

rho2 = 7.7737

imax = 10

hold on
plot(lags2(imax), xc2(imax), 'r*')

[Mustafa Al-Nasser]

Ok but if the signal is shifted by 5 samples while lag is showing 10

[Mustafa Al-Nasser]

I mean why?

[Chunru]

To know why, you need to show your problem. Your incomplete code can not tell why and we can only show what is right.

[Mustafa Al-Nasser]

Ok now I found the problem and I am find the lag at which cross correlation is max

but actually my problem is that i am dealing with real signal which is very noisy , so, i am looking to use low pass filter or smooth signal to remove the noise but I have a concern that I may lose some useful information