The current objective of my adversarial planning project is to create an anti-NASDAQ of sorts. I want to create a real time metric for leverage to catalyze proletarian-self activity.
Is it possible to scrape the web for business/industrial reports and aggregate them? I've noticed that LLMs are pretty good at summarization; can they be used to aggregate economic reports from across the web?
Yeah, it should be possible, but if you're using LLMs it would take a lot of compute. Also you'll have to deal with the lag time for financial reporting to happen. Ideally we would have a system where workers could report this themselves.
@casperadmin Any way to infer from real-time data? How frequent are financial reports? LLM might not be necessary if you can get the data directly?
The idea is to rip enough data from the web to use cockshott's dissagregation algorithm to get the whole thing.
Sorry for the late reply.
Most financial reports are quarterly. There are some datapoints that come out weekly or daily, however, see the latest releases from Fred Federal Reserve Economic Data | FRED | St. Louis Fed (stlouisfed.org)
@casperadmin That's really good to know, that + the max-ent stuff I found could make this work in 'real' time (monthly maybe) thanks!