In one of the last videos Jeremy mentioned companies that are on a stock market and have to share their data.
With that data you ought to be able to make better predictions for those companies.
If I remember correctly he also mentioned a paper about this.
I would love to rewatch this and read the paper, but I can’t seem to locate it anymore.
It’s not in the timelines sadly, so I was hoping you guys could help me out here.
1 Like
I’m not in ‘Part 2’ yet, but can answer some of this.
First - here is a rather recent paper on the topic - its quite an interesting approach and results.
https://arxiv.org/abs/1711.04837
Publicly listed companies (ones that have stocks that trade) have to publicly release financial data quarterly. In the US they file on Edgar (https://www.sec.gov/edgar/searchedgar/companysearch.html). There are api’s for accessing this data. In Canada look at SEDAR (http://www.sedar.com/ - less friendly). For others, you’ll have to look it up. This said, most people don’t go to the raw reports directly.
If you are looking for stock / financial data there are a number of ways to get started. You can get basic daily time series via yahoo finance (lots of code out there for this). There are sites that give both free and paid access to fundamental and price data. Quandl is a great place to start for this - they have price, economic, fundamental (financials etc) and other data (some sentiment). You can also look at sentiment on news and social media (twitter, stocktwits) - ravenpack has data sources for a bunch of this (https://www.ravenpack.com/ - not free). If you are an academic, there are services at most universities and you may be able to get access to some datasets from University of Chicago.
Most professional now pay a lot for their data - sources like Bloomberg, Reuters, Factset, Comstock, S&P Capital IQ - and then still spend a lot of time cleaning and preprocessing.
That is a great reply! Thank you kindly, kind sir. The paper’s approach sounds exactly like Jeremy’s, though it’s a newer paper and thus a different one.
I’m still wondering if the one Jeremy mentioned has something extra, therefore the request still stands. So if anyone else comes across it, please let me know.