I don't know what a "peaky neature curve" is. If you want to know how well two signals correlate on an element-for-element basis, you can use corrcoeff(). If you want to find out where (what shift) one signal overlaps another best, you can use xcorr. There are some things you need to look out for and be aware of, like xcorr() might indicate a shfit where the ends of the signals are shifted and there's only one - they don't overlap. That might be ok if one signal is much smaller, like it's a template, and you want to find out where in a longer signal it appears. But if the two signals are just two arbitrary signals of about the same length, then I think you better think about what you're asking and what you really need.