MyHeritage uses two different parameters to calculate DNA matches:
The total amount of shared DNA and the number of shared segments.
The algorithms and thresholds that are used to calculate matches might vary between the different providers, and therefore you might not see the exact same matches in the different sites.
Keep in mind that DNA Matches listed at the very bottom, who share very little to almost no DNA with you, could be the result of identity by state rather than identity by descent. This is a technical way of saying that these are false positives, and the DNA they share with you is a result of coincidence. It means that indeed, the DNA segment is similar between you and your match, yet it is unlikely related to a recent known ancestor, but rather to some ancient DNA or random sequence.
Starting with 4th cousins and going further, you should take DNA Matches with a grain of salt, and look for additional pieces of information as an indication on whether a family connection exists (such as shared ancestral surnames, or similar geographical locations between their family tree and yours).
We will continue to enhance and fine-tune the algorithm regularly. We are working to implement an exciting feature road-map, and you can look forward to improvements on a regular basis. DNA Matching on MyHeritage will keep getting better and better, and more matches will be found as more people upload their DNA data every day.
To learn more about this topic, read this article on the MyHeritage Knowledge Base: What’s Considered a Strong DNA Match?. On the MyHeritage Knowledge Base, you'll find additional articles, webinars, and how-to videos that can help you master your genealogy skills.