Automating Modeling - CVNA Test
Thanks Google!
Hi all, another update on my AI escapades in relation to finance work flows.
A step in the research process for many is of course finding historical KPI’s and financials and trying to get a rough sense for something like the following:
How a company makes money
How much money it makes
At what margins
How that has trended over time
There are a lot of really nice tools for this like AlphaSense’s doc search+Canalyst or free/cheap solutions like Tikr/Quart/Yahoo finance. Typically though to actually get everything in an excel file including not just financials, but KPI disclosures, you’d need a Canalyst subscription, an intern, or to do it yourself.
Thanks to the complete wizards over at Google, this is no longer the case!
Over the past week I’ve been iterating on some code that basically reads a model, then auto-completes it using the appropriate filings. This initial test was for CVNA, but it’s scalable to basically any publicly listed company and even works pretty well for international filings given similar disclosures.
A link to that sheet (the Google sheet version, thanks Substack) can be found here: Link
It currently takes about 15 minutes to fully parse a filing and spit out an updated model, which I’m rather confident is faster than just about any human, and also drastically cheaper!
I’m writing this article because I think that’s a really cool thing, in addition to the other stuff I’m working on. We’re starting to get to the stage of trialing our AI workflows where we probably work closely with 5-10 funds/investors to rapidly iterate and scale out coverage on their names before we get to covering everything publicly listed.
We want to have a product where you can
Have access to SEC filings/earnings transcripts/relevant news articles as well as AI tools to assist with analyzing those.
Have access to model generations that can be the template for valuation work or used to quickly spin up on a name
Utilize our AI generations to find data sources/information sources on industry KPI’s/catalysts/blogs/podcasts/etc. Stuff like Manheim vehicle index and industry reports, podcasts with C-Suite, Substack writers who cover the stock, and more
Receive daily/weekly updates on developments at the company, alt data, macro factors, competitor earnings, etc
Understand sentiment over time for key business arms
How narratives have changed around the stock over the past few years and what events led to those shifts. Why was CVNA down 99% and then up 100x? Great question!
So much more.
A lot of that functionality is already live and we’re continuously improving on it. If you’re interested in working with us would love to chat. The models are slowly becoming absurd productivity unlocks, allowing much more depth and breadth of coverage. If that sounds useful to you, shoot an email to Kyle@Clarity-Markets.com or shoot me a DM on Twitter.
Cheers!