Someone could download a dump of the whole thing (the SEC has a link and updates it every night) and assemble all of the various documents in a way to get a decent estimate.
An example Form 4 (which will show cost basis of acquisitions):
https://www.sec.gov/Archives/edgar/data/1045810/000104581025...
An example Form 144 (which would give you a pretty good guess as to which shares are being sold; you can then correlate with a known Form 4):
https://www.sec.gov/Archives/edgar/data/1045810/000192109425...
The random sale model doesn't account for pensions, mutual funds, and individuals that hold. Where would firms like Berkshire Hathaway fit into this model?
I imagine the distribution of stock hold duration is nonlinear/logarithmic
I don't consider they serious, because they could buy penny stocks, which are unregulated and are very frequent target for fraud.
> pensions, mutual funds
Funds are better than individuals, because usually they have some strategy and hiring professionals to control things, but they also prone for mistakes.
It's just a measure of liquidity. A stock can trade 1$ or 0$ per day and still have a market cap of 300B ( see private eq.)
(Maybe you are onto something, if stock iliquid, they got stocks at 0. If stock very liquid they got stocks at actual price.)