101–109 of 109 posts
Hi, great idea! I tried to do something similar for earnings call transcripts and got very stuck very fast, so I appreciate that this is quite a tough problem. My suggestion would be to indeed see if your algos can extract useful structured data from earnings calls too, as often c levels give information here that’s not necessarily included in the regulatory filings. Good luck!
Are you guys able to access the Ledge site? Seems down for me but the main site is fine
Just curious, are you doing anything to prevent companies from running drafts of their disclosures through your algorithm? That would allow them to tweak the disclosure until your AI gives it a positive score.
We don't sell to filers. Ever. Got to protect against conflicts of interest. That's one of the reasons we don't have a true free version of the product. That said, we're using language models so replacing words with synonyms won't evade the model as long as its expressing the same thing.
Out of interest - why didn't you start a hedge fund instead?
Earlier quoted context omitted.
I think big part of a strategy like this is long term. If you want to short a stock you need to make sure the fraud is discovered by other investors. So you probably need to have a strong marketing department that basically thrashes other companies, this may not be a goal a lot of people aspire to. With a short you're also exposed to lose an infinite amount. And finally "markets can stay irrational longer than you ca…
My guess is they also have plans to move in on other sorts of fraud detection, for example bond issues or privately held firms. That’s my good faith interpretation. My bad faith interpretation is they don’t trust their algos enough to trade on it themselves.
It seems to me all you have to do to fool a ML model is rephrase sentences and just use verbiage it has never seen before. So why isn't your tool just the latest item on a pre-IPO company's checklist that is looking to commit fraud to scrub against? Also if Muddy Waters can do something similar to this (if not better?) doesn't that mean you should just hire them and throw away your ML model?
It isn't that easy to reverse engineer our models. We test our models for a certain degree of robustness (same concepts expressed in different ways will be generally picked up). Plus, if companies go out of their way to express risks in cryptic terms, our models picks up needlessly convoluted text as well. We also constantly retrain our models so I don't foresee this becoming a problem in the foreseeable future.
Out of interest - why didn't you start a hedge fund instead?
A few reasons but the primary one is impact. Our mission is corporate accountability through information transparency. Once you become a hedge fund your incentives for information dissemination become warped.
I could easily be missing something related to your specific goals or situation, but on the surface I don't see it? Have you seriously explored it? I mean, if this thing you have works, you might be leaving a billion dollars on the table...
Earlier quoted context omitted.
My guess is they also have plans to move in on other sorts of fraud detection, for example bond issues or privately held firms. That’s my good faith interpretation. My bad faith interpretation is they don’t trust their algos enough to trade on it themselves.
See response above. And not just fraud detection. We plan to expand to other areas including credit/bankruptcy, ESG and tracking macro changes and trends
I don't know if there is an opportunity related to ESG.
Update: I see you mentioned 10b-5 as your gold labels