Looks like I'm going to have to spell it out for you.
When a locus is tested, that is 'one time'. When 100 loci are tested, that is 100 times. When 5 of those 100 loci are more similar between than within groups, they are more similar 5% of the time (due to 5% of the loci tested). When you say "38% of the time" that means "38% of the loci tested" because that's how you determine the percentage of time wherein there is similarity -- this is what your quoted study is talking about.
"of the time" is referring to genetic similarity between one random European and a random Asian. It is not referring to a genetic locus. You made that up and your interpretation is baseless.
This is usually where I directly quote the study's methodology (not the abstract, which for some reason you think will explain the methodology). However, it's jargon heavy, not an easy read at all (took me many minutes) and the average person isn't going to understand it. I've already attempted to explain what it means, but you've decided to label that as "baseless" lol.
So, I'll try another way.
Bamshad (2003), Guo (2015), Alloco (2007) have all produced results showing that self-ascribed Asians fit into the Asian category 97%+ of the time, even when K > 2 (your Asian-white example being K = 2), and even when far less genetic markers were used compared to your cited study
Race Realism: Critical understandings (debateart.com) .
You're essentially claiming that when K = 2, Asians only fit their category 62% of the time -- way off what most studies are showing.
Do you see that now? If not, how do you explain all this research that contradicts what you've argued?
Well map every single gene on the genome and figure out what all of it does, come back with a list of impacts and figure out which loci are important and which are not. Assign them a significance value, say 1-10. Then you can say "this gene matters, this gene doesn't matter" etc. Let me guess: You're going to select the ancestral markers as the only important loci.
I've already explained to you why researchers don't do the entire genome. You've indicated that you don't believe this is a reasonable standard yourself by citing a paper which doesn't measure the entire genome, yet using it to make an argument.
The reason I mentioned "significance value" was that I was trying to explain how your study found the 38% number. Don't take that out of context.