Are you sure you're not comparing liquid vs illiquid assets? I.e. public shares that you can sell at a stock exchange vs. a highly speculative valuation that may pop >10x like WeWork did.

There's also this wonderful disclaimer:

>Summary statistics from the AngelList dataset of 684 nonnegative investments that we consider in this paper.

Given that 90% of startups fail, they are are looking at the top 10% investments. I am pretty certain if you hand-pick the 10% top-performing public stocks, you will also get great numbers. The problem is, this only works in retrospective.