4 ms·
There’s also this list where they correct for “synthetic longs” and show that some of these have dramatically lower short int %: https://mobile.twitter.com/S3Pa
by faizshah 6y ago
There’s also this list where they correct for “synthetic longs” and show that some of these have dramatically lower short int %: https://mobile.twitter.com/S3Partners/status/1354851186533396481 https://mobile.twitter.com/S3Partners/status/135485118653339...
I won’t pretend to fully understand this but the analyst answered a lot of FAQs on this data on his Twitter: https://mobile.twitter.com/ihors3 https://mobile.twitter.com/ihors3
What differs in this data: “S3 data is shares shorted, our competition's data is broker stock borrows. Problem is, stock borrows do not equal shares shorted. Stock borrow includes financing trades, broker to broker transactions but does not include all internal rehypothication used to cover short sales”
- bo1024 6y agoI'm no analyst, but this still looks really bad to me. If you buy a random share, there's a 55% chance that share is currently loaned out and a short-seller has to buy it and give it back. There's a 55% chance that share the short-seller buys is currently loaned out... A single share purchase cascades into 2.2 purchases on average. If any sizable fraction of shares held are committed not to sell, that math gets way worse.
- splistud 6y agoand then there's intraday lol