Things I was also contemplating to implement you can freely steal:

1. Giving a list of potential interesting investments - these could also be behind paywall.

2. Summarize based on sectors to see where there's downturns coming - again maybe paywall this

3. Connect with news-sites or twitter etc.. to see what is happening and what is the sentiment. This is a lot harder as you'll need to either scrape or pay to get that info.

Best of luck

Just two cents based on my usual workflow --

1. I like to compare companies within the same industry, so it would be good if i can ask follow up questions after reading a report like "How does company Y compare with company X on XX metrics?"

2. I understand that the reports are generated by a model; if that is true, maybe you could cross-reference the fillings with their corresponding earnings call transcripts (should be available for free on many investment websites) and highlight what is being discussed/asked in the calls as well

Again, thank you for the great work. Already pinned your site to my browser's investment workspace XD

Charts like that show more detail sure, but everyone freaks out in reaction to them. Always zero out your graphs.

Form 10-K is usually what you'd need to parse for those, maybe with recent stats taken from the 10-Q.

It'd be really great if one could get a report on a company in an industry they're unfamiliar with, with statistics like profit margin available at a glance, and a good description of how they actually operate, what is important to them, what areas of the business actually bring in the money (this is often surprising!) etc.

https://boards.industrial-linguistics.com/

It only looks at information about directors on company boards, and unlike your much better project, I don't have any clear idea how to commercialise it.

I have to build a project for data analysis with ML (and optionally AI) for a course and certification I am passing, this is a great inspiration. I will probably do something simpler on European stocks.

You say it was a year long effort, was it full time ? Do you have an estimation of how many hours you spent on it ? Did you do it alone ?

PS if the next question is whether I’d pay for it, the answer is likely no ;)

A suggestion: I forget the HTML tag off the top of my head, but there is something you can do that turns off the keyboard autocorrect. Setting the search box input type to something specific or something like that… stock tickers aren’t dictionary words :)

Minor suggestion: I would make the home page feed show popular stocks (so that it is more interesting to new visitors and also engaging to regular visitors). You could just pick 100 stock symbols or combine that with something like what you are seeing people search for (I would manually check the popular search ones - some may not make sense to add to the home page feed).

In theory, this is what value investors are doing to some extent. However, embedding vectors often to do match intuition very well.

If you are unfamiliar this was passed mostly secretly in 2019 during the whole Kavanaugh confirmation, almost without issue crossing party lines. Its made every SEC required statement effectively worthless for the entities involved in any project or work that touches on national security.

There are at least 7 of the largest companies that make up the majority of the DOW/other indexes that may be covered.

I'm trying to do something unrelated with public companies, where did you get a list of all companies filing with the SEC? Or was it built based on whoever filed 8-Ks? A google search leads me to this list [0] but it's missing a few companies.

[0] https://www.sec.gov/files/rules/other/4-460list.htm

If this is a side project, that's amazing and I love it. If you did want to take it further though, I would though encourage you to spend some time interviewing potential users and thinking through how you are creating value.

As a fundamentals investor, to win in stockpicking (especially against quant traders), you need to get really deep into the details. As mentioned there are lots of tools that summarize earnings calls, but more importantly, I don't think summarizing earnings calls or releases are super useful for investors.

There are basically two ways to build useful software for investors. One way is to create summaries / off-the-shelf analyses that tell investors if a company is good or bad. Right now as-is, that is sort of where this product is. The problem is, an earnings call summary is insufficient. Yes, AAPL revenue may be up 4.0%, but the fact that you figured that out in 15 seconds with an LLM is not a differentiator since everyone else can figure it out in 45 seconds by just doing mental math. And, what did _you_ expect it to be? Very different trade answer if you thought it should be down 10% or up 40% and it in reality hit 4.0%.

I think generally this is also tough because the people really good about building recommenders (black box or otherwise) just work at hedge funds and make a ton of money for themselves / their clients. It is almost always going to be more profitable to do this, in no small part because once you reveal your secret sauce to the market, it gets priced in.

There's also the issue of user personas. There is no one generic investment strategy. A "good" investment is very different for a long-only vs. a long-short vs. a macro fund vs. an activist fund (and yes these are not mutually exclusive at all).

So, I think in order for this to become truly useful you need to make it more of an interactive tool to allow people to run their own analysis (vs having a "one size fits all" analysis). A few ways you could do this:

1) investments are done behind an investment thesis. Can you TRACK theses over time? For example, a merger arb thesis might be that a big M&A deal is on track (or not on track) to close. Can you track my THESIS, which almost becomes a detailed LLM prompt. E.g., "I think that Company X's acquisition of Company Y will close at the current offered price of $Z by date A, and key milestones are (1) regulatory approval on or around date B, (2) stock price converging toward the offered price of $Z as date A approaches, (3) positive sentiment on earnings calls about the acquisition by both Company X and Company Y"; let me know if earnings releases or calls or 8-Ks should make me update my thesis ("8-K came out that said that the target date of A has slipped by 90 days"). Or, "Company X said they bought Company Y and at the time expected revenues of $Z in three years; what compound quarterly growth rate does this imply" and then track that every quarter to see if they are on track or not

2) Can you help me track a specific question through each quarter—almost like automated BI / reported? Bloomberg/CapIQ/etc. will do this for GAAP metrics like revenue or COGS but often companies will report bespoke metrics. Can you help me track WeWork's community-adjusted EBITDA (lol) every quarter? Can you help me track business unit non-GAAP EBITDA with an addback for taxes every quarter? These are things that won't show up in your GAAP 3-statement model but for _individual_ companies may be super important

3) similarly can you help me run comparisons across companies? E.g., can I compare Vornado's total revenue (almost all office) to Bridge's office segment's revenue to SL Green's total revenue (also almost all office) and every quarter see how their relative revenue (market share) is changing? What about their occupancy and rent levels?

4) generic RAG/natural language search (when did SmartRent talk about their acquisition of Planon), although as noted, many competitors are doing this

All of the above are things that analysts do in real life and basically just track manually in Excel.

Again don't mean to be discouraging—I posted this in part because I think there is a ton of value here to (a) having a clean pipeline and (b) having a basic working knowledge of applying LLMs in a useful way. If helpful I'm happy to discuss more.

Also, rename "News" to "Latest Filings" or something more direct. Generally websites use "news" to refer to news about themselves, which nobody cares about.