Just as a piece of related information, FTDNA will likely show different results depending upon which test data you upload to them, and between a test taken with them and a test from a different company. In other words, though I have no statistical detail, their genotype imputation isn't accommodating with precision the different SNPs targeted by different companies' tests and chipsets.
In one example, an individual with two accounts at FTDNA--one with uploaded AncestryDNA results and one with FTDNA's own GSA chip test--shows West Slavic segments on the new chromosome painter present on the FTDNA test on chromosomes 4, 14, and 17 that do not show on the uploaded Ancestry data, and on the Ancestry results segments on chromosomes 7, 10, and 19 that are not present on the FTDNA results. Overall, there's about a 4% difference in the "super population" representation. Your mileage, of course, may vary; but I have a feeling that the different SNPs tested will align differently with the cohorts FTDNA is using as reference panels.