/biz/ - Business & Finance

Good morning anons, I'm back with some updates from the app I've been working on, which attempts to evaluate the difference in Robintrack.net's popularity changes to find actionable predictors of short term spikes among the retail PnD crowd. Last I left you all, I was running the app on its first trial run (which I botched somewhat on Monday), and between yesterday and today's tweaked run I've gathered up some results.
Which are so far pretty underwhelming - the main issues seem to be that Robintrack only updates hourly at the 50 minute mark, and that Robinhood users do not seem to contain a subset of trendsetters. My app did properly predict the IBIO spike just before 1pm yesterday, and today it's predicted the MTP spike this morning, but both of those were only well-predicted because they occurred after the 50 minute mark, and they were both just as predicted by RT.net itself. The combination of these factors means that even if RH users were giving off good signals, Robintrack as a front for their data is simply too infrequent and slow to be usable.

This isn't what I'd been hoping for, but it's close to what I'd expected, and really I've still got more data to scrape before I can draw any final conclusions. I'm not regretting the time spent learning, and I do want to implement the Robinhood API myself to get a faster set of rolling differences to analyze (tragically I am a lad whose currency is $CAD so I may not actually be able to sign up to access their API), but in the end I'd say my confidence in getting actionables from any Robinhood data has been reduced. Depending on how things go, I'll be back with more info about the current app, the API-updated app, or perhaps some new ideas for next steps in automating research. With that said, there's no reason we can't all be learning together, so: do any of you have any ideas for analyzable patterns? If your idea rocks, I'll give it a go!