I don't trust such estimates too much. The most trustworthy results come from matches which share solid segments with a person. Like if you have a couple of Sephardic matches on a specific area of a chromosome, that would be better evidence than ethnicity estimates of some sort, which might be skewed by bad fitting references or very old admixture events.
If we are talking about 1500 and later, there is a chance, if the admixture is real, that you still have some matches in that direction. To check that seems to be the better way to go, IMHO.
If we are talking about 1500 and later, there is a chance, if the admixture is real, that you still have some matches in that direction. To check that seems to be the better way to go, IMHO.