Yesterday I watched a Youtube video from a group called Family Tree Fanatics which was quite good at explaining how to work out if a match is a good/true match or likely to be a false positive.
The man talked about working out statistics for good matches over all the DNA databases and how their algorithms and number of test results can show you a more manageable way to investigate your DNA matches. He had his DNA on all databases, but I don't, and am not likely too. But the premise of working out which is better for your matches probably okay.
The recommendation was that the smaller the cM match and more segments makes it more a possible false match, just a sort-of regional match. He sorted his matches only down to 35cM.
Looking at it from his American point of view was interesting, so I tried to adjust it for my Australian perspective - that is there are a lot less matches on GedMatch and probably 23 and Me and others, than Ancestry.
I put all those really small cM matches (less than 20) into the hidden files, unless they have a family worth investigating, or in one case, a common ancestor nominated. That one bears more investigation as it seems to make the link across several family trees. Maybe I will re-visit the hidden files in the future.
If you are interested in statistics and algorithms for DNA databases, here is the link:
https://www.youtube.com/watch?v=O9tV...&feature=share
Di