Lopen posted... Anyway I'm not claiming to be an expert here. I haven't even dabbled in that kinda stuff since 2012. So yeah it's very possible the entire idea is different now-- but the idea of machine learning and training/testing sets, conceptually that hasn't really changed to my knowledge. It has no doubt advanced but the basic concept is the same I'm sure.
But yeah I just know a LOT of the market is automated first hand (I mean it should be obvious it is just looking at it, but I have an idea firsthand of how it's done) and you're probably giving short sellers/market makers more credit than they deserve if you say "well they wouldn't do that because they'll clearly lose money"-- they will make moves that will clearly lose money because on average with a stock market full of stocks to exploit they won't lose money making those moves. The amount of coordination and awareness needed to beat the algorithms is such that they won't lose all that often.
Also there is just a lot of information on BBIG that can't even be boiled down into "data" such that if you've got it flagged as an easy short, the data that it shouldn't be shorted isn't something an AI can really digest. A dividend is coming that has no cash value and must be covered when shorting-- there's no way to feed that into a machine.
So if you want my answer to "why would they do this, it's so hard to make money continuing to short here" the basic tl;dr answer is
"they're not consciously doing it, and right now BBIG is a blip on the radar in terms of the overall balance sheet"
Maybe if it flies to hundreds Gamestop did they'll notice. Until then load up on the cheap.
I guess that makes sense, and the algos may not have ever actually "seen" what the humans all saw could have happened - the price climbing into the thousands on the strength of actual margin calls in the billions of dollars. So they may not be appropriately evaluating the risk.