RootsChat.Com

I've got -

Matches with confirmed ancestry:

430 from Ancestry
12 from MyHeritage
7 from 23AndMe
3 from FTDNA

Matches with common ancestors, but unconfirmed link to self:

450 from Ancestry
21 from MyHeritage

So also < 4% from MyHeritage.

MyHeritage is better for continental Europe. In my limited expedience Ancestry has close to no coverage for Germany and Scandinavia. Almost all matches for such lineages will likely be from immigrants, mainly to the US.

The main benefit of Ancestry is the database size - https://isogg.org/wiki/Autosomal_DNA_testing_comparison_chart

Ancestry at 27m and MyHeritage at 9m.

With Ancestry's skew to tests in the US, Canada, UK, Ireland, Australia and New Zealand, it means people with heritage US and British ancestry will get far many more matches on Ancestry. You will find even higher number of matches if you have Jewish ancestry and Mormon ancestry or relatives. As those people take so many tests.

Then there is the situation with Ancestry's timber system, which strips out what they believe are false positives. The rate of false positives on MyHeritage is around 70-80%. So the majority of your matches are not reasonably close relatives, but people who share a common region of ancestry. This seriously hampers finding real DNA relatives, as if you find common ancestry in a cluster of matches on MyHeritage, you have a high chance they are not actually closer DNA relatives. I'd say the false positives on Ancestry >= 30cM are 0%. Below that it's difficult to tell, but it's much less than MyHeritage.

So if you have US, British ancestry you will be able to find multiples more matches, in the range of 40X-50X, with a high degree of confidence. MyHeritage is useful for closer matches, mostly above 100cm. Below after around that level you get false positives.

The MyHeritage cM values are also less helpful, because they don't strip out identical by state (false positive) segments. Below 100cM matches, you need to multiply the cM by 0.6 to get a closer idea of the identical by descent shared cM. So for a typical US or British tester, by the time you've got through your 1st page of DNA matches, you are probably down to people who are 4th-5th cousins. By the time you are through to page 7-10, you are dealing with people with whom you share about 15cM, which might be identical by descent, but is more likely to not be. And 15cM is a very ambiguous amount, as it's at the threshold where it's not not uncommon for 15cM to be inherited (nearly) unbroken over multiple generations. Around 15cM is the average size of segments inherited from either parent.

In terms of Ancestry:

>= 30cm - you will likely be able to find a relationship if there is no NPE, a tree and no brick walls, relationships usually quite predictable
20-29cM - these sizes can be much more ambiguous in terms of relationship, owing to the possibility of smaller segments being inherited (mostly) unbroken
13-19cM - progressively more ambiguous
8-12cM - the range where segments could potentially push outside of the reasonably feasible to research window of ancestry

While you probably only have 10-20 >= 30cM identical by descent on MyHeritage, on Ancestry you'll have more like 200.

With Ancestry, you will typically get several to about 50 matches in a cluster, even for 4th-5th cousins. With multiple people related in the main window that is not so difficult to research (1775-). You look through their trees to find common surnames, places and ultimate ancestors. Heritage just doesn't have enough matches to be able to do that very consistently. Since many matches will have no tree, NPE, or dead ends that aren't easy/possible to solve.

If I go through my Ancestry matches down to about 20cM, if you exclude the ones that have no tree or not enough of on to reasonably find out their ancestry, I know how about 75% are related to me, or I have them in files where common ancestors have been found among the matches. (I have three NPE lines and a suspected adoption, which most of those relate to). Yet if I go through the MyHeritage matches down to Ancestry-comparable 20cM, I don't know how I am related to 90% of them. For the vast majority of them I don't even known what side (paternal, maternal) I am related to them on, never mind the actual line of ancestry. While with Ancestry, for 20cM+ matches, I know which line of ancestry they relate to in most cases. Owing to Ancestry having many more matches.

With Ancestry DNA I've been able to break down 7-8 brick walls, prove most of my lines back to between about 1650-1775, identified a likely adoption, have solid leads and some known ancestors for three NPEs, and expect to be able to solve more lines into the future. Also solved the paternity of a friend's paternal grandfather (who it turns out was my grandfather's half 3rd cousin) and united a few unknown half-siblings. Comparing to MyHeritage, I've just got matches who have provided no clues to break down brick walls, and an unknown ~2nd cousin match where I don't even known if they are a paternal or maternal match.

That's why Ancestry wins hands down, and the competition are not close. Despite their bleeding of their loyal customers and having developed their system very little.

The best process for getting the most out of genealogy DNA is:

1) test with Ancestry
2) upload that test to MyHeritage + FTDNA for free
3) test with 23AndMe (not sure what will happen with that)
4) systematically go through the matches to find common ancestors among them

R-M269 is my haplogroup.

Knowing it is of Zero use to me in researching my family history.

Knowing that for the last few thousands of years my ancestors were in Western Europe is not telling me anything useful as it is far to general.

Even forking out excessive amounts of cash for a Big y test is only going to refine the result to a sub-branch which then is still going to be of little Genealogical use as again it is generic by nature.

Hopefully MyHeritage or Ancestry will buy FTDNA one day, as it seems clear that FTDNA don't have the necessary attributes to take Y + mt testing to a consumer audience. It's mostly used by hardened genealogists and people searching for their father. I am fairly sure the price of Y-DNA tests could be significantly reduced. Their autosomal business has obviously been nuked and all they have left really is the seemingly overpriced Y tests. You can get full genome tests for about the same now.

I also see that Peter Thiel's fund is looking to acquire 23AndMe to their already existing DNA company:

Nucleus Genomics, a New York-based whole-genome testing company, is connected to Peter Thiel. It is backed by Thiel’s Founders Fund, among other investors, and its CEO, Kian Sadeghi, has expressed interest in acquiring 23andMe, particularly for its telehealth subsidiary, Lemonaid.

Nucleus Genomics is a New York-based company specializing in whole-genome sequencing (WGS) and genetic analysis to provide personalized health insights. Unlike many genetic tests that analyze less than 0.1% of DNA, Nucleus sequences the entire genome—approximately 6 billion letters—to identify millions of variants that may impact health.

Their full genome test is actually less than the FTDNA Big-Y test, which give an impression of how overpriced it is.

It would be nice if a boat load of VC cash could be used to bootstrap another outfit offering genealogy DNA services, to boost the number of people you can match to and hopefully get better coverage for Y and mt matching.

From what I'm reading whole genome would offer better sifting of identical by state and descent segments, in short better weeding out of false positives.

Autosomal Tests: These typically analyze ~600,000–700,000 single nucleotide polymorphisms (SNPs) across the 22 autosomal chromosomes, covering less than 0.1% of the genome. They focus on common SNPs selected for ancestry and health insights but miss many rare or private variants.

WGS: Sequences all ~6 billion base pairs of the genome, including all autosomal SNPs, rare variants, and structural variations (e.g., insertions, deletions). This provides millions of data points, capturing virtually all genetic variation relevant to relatedness.

The denser data from WGS allows detection of smaller and more numerous shared DNA segments, increasing the chance of identifying matches with distant relatives who share fewer or less common segments.

Autosomal Tests: These rely on identifying identical by descent (IBD) segments, typically requiring a minimum length (e.g., 7–10 centiMorgans [cM]) to confirm a match. For 4th cousins, who share 0.2% of DNA (13–20 cM on average), the segments are often small or fragmented, leading to a ~50% detection rate due to limitations in SNP coverage or recombination events breaking up segments.

WGS: Captures all DNA variants, enabling detection of shorter IBD segments (e.g., <7 cM) with higher confidence. It also identifies rare variants shared by descent, which autosomal tests might miss. This increases the sensitivity for detecting distant relationships.

***

I've got five y-DNA test so far. Only one big-Y, but you can often determine some of your recent haplogroups if your closest matches have taken big-Y. From that, you can extrapolate out migration path. Though it is somewhat circumstantial, the results are more of a best guess. One example:



For this specific line it appears it was probably part of the latest Bell Beaker immigration to Britain, c. 1800 BC, rather than the later Celtic migrations from c. 1250 BC to the Roman conquest.

So with the data you can get a reasonable idea of what historic peoples your line was part of. There is quite a wealth of archeological research on prehistoric cultures. You can get a reasonable idea of which pre-Roman tribes your ancestor belonged to in Celtic Britain, Germanic peoples, and early Celtic cultures of Europe. You can also see fairly clearly if your line ended up switching from Celtic to Germanic via conquest, or vice versa. Since the area that is now Germany was the intermediary zone between the Bell Beaker culture that likely spoke a predecessor of Celtic and the adjacent Corded Ware culture, that likely spoke Balto-Slavic.

It is probably worth saving the details of your most promising matches from 23AndMe.

I logged in as they told me I had some new relatives. When I looked at my match list I found that my two closes matches had disappeared. These were a 1st cousin of my aunt and her son. Though this was via a birth out of wedlock. So I am lucky to have saved another info to be able to know who they are.

To this date I've only noticed on other match disappear from MyHeritage, in that case I had it saved in my bookmarks, which contained enough info, as well as my memory, to know who it was.

T-FT8342

It may not have many matches, as the matches may be from populations that are not much tested, i.e. outside of the US.

The best that can be said is that around 6,000 years ago the lineage appeared to be in the Middle East, maybe Arabia specifically. So there is not much information to suggest a possible lineage. Given there are two bearers of the parent haplogroup in Italy, it could have found it's way in during the Roman Empire. Rome had many trade links with what is now Arabia.