Two comments regarding your data points: A) As far as I see all these companies were created in a different market, the most recent one from your list created

Two comments regarding your data points: A) As far as I see all these companies were created in a different market, the most recent one from your list created 17 years ago: 1. Apple founded in 1976 2. Google founded in 1998 3. Microsoft founded in 1975 4. Amazon founded in 1994 5. Facebook founded in 2004 6. Tesla founded in 2003 I think it matters to take into consideration when we focus our analysis. The times were then and now. So analysis companies from 70' or early 2000's is different than analysing founded companies in 2020's. I did not verified yet the dataset attached to the study, but I see that they say > "Across the 2.7 million founders in the United States between 2007–2014" so it appears they are studying 2007-2014. So I have the same skepticism against the article. 2020 is a lot different than 2014. There are some similarities but also a lot of differences in the market, capital, know-how, competition, users expectations and more. What worked back then will not work now. B) On one side Outliers can be fascinating to study. But for 99% of the population also unuseful in a way. Cannot replicate their pattern individually. They are in a way a probabilistic "side-effect": they appear when there are enough interactions or data points, but nobody can pin point exactly from which initial data point they will appear. If this would be possible then with time everybody will invest early only companies that will become trillion dollars companies.