mean and standard deviation for kernel estimates

Dear all I try to analyse the distribution of returns before and after a signal. For this, I estimate the pdf using the default kernel estimate (which is the gaussian kernel).

My problem: I have a vector with log returns. I can easily calculate the mean and std of these returns. However, if i estimate the distribution using ksdensity(logreturnvector), I get a curve (looking bit triangular, not very smooth) and if I go in figures under tools and selecting data statistics for this particular curve, i can select MEAN there (to be drawn in the diagram) but this mean is completely different from the one calculated in the command window using mean(logreturnvector) (and by using the default setting, with no change in width, this mean lies far to the left from the peak)

My question is therefore: How can i interpret those two different means?? Shouldn they equal?? And is the std of the log returns calcualted as std the same as the std in the ksdensity??

Any help would be highly appreciated