Idea Validation, Web, using ML to analyze company 10k fillings, ceo letters and calls, looking for CEO candor and dishonest accounting
Buffet and others have said that Candor from a CEO, trust, accountability are very important.
Avoid those that hide behind deception.
https://www.profgalloway.com/yogababble
There is already those you can hire to analize the letters of CEO's I propose some ml that does this for you at a SaaS price.
(10 stocks a month for x, 100 for x)
Also its been shown that AI is very effective at identifying accounting fraud, so at the same time as the above, and for the same price, get analysis on companies that would identify the frauds of the like of Enron.
It would be targeted at the smaller investor, keeping it to something around $24.99 a month, running an API on Azure for this is very inexpensive, and once you have the model build (which I have already) the rest is cheap. Could have an enterprise offering, unlimited scans, or even have it inbed results in enterprise CRM's / Portfolio Tools.
I'll come at it from a different answer than others will. I have no idea if your idea is good, but here are some questions you could answer to find out:
(Market Risk)
(Product Risk)
Best of luck!
Thanks, appreciate you taking the time to respond. Great questions..
People are looking for not so much this tool, but the answer the tool provides, entire companies exist that answer just the candor question alone, ( http://www.rittenhouserankings.com/ ) 100's of large companies like KPMG, Deloitte are hired to look for accounting fraud for Hedge Funds and Wealth and Asset Management firms to give them an edge.
Now this tool will not do as good a job as those firms, just like the quantitative scanners you can subscribe to monthly wont do as good as a team of quants at a bank or as good as professionally made algo's. Still the every day investor and small financial managers need the same analysis done (If there taking their job seriously) and like buffet they sit there day after day reading 10k after 10k. Well, now they can scan for qualitative data just like they can scan for quantative.
So no tools, but the concept, which is what I think maybe make its a good SaaS tool, providing a need to what people currently do on pen and paper or hire large expensive companies to do.
Hard to say.. This would be a weakness.. Market research or a landing page might help, but i think it would be a lot of marketing. Although bundling might be better, there are some good SaaS portfolio tools out there, they may be willing to pay a small fee to bundle this in. Or some of the quant scanner guys might want to bundle this in as a qualitative tool.
Time is what it solves. It solves you spending large amounts of time reading and forming the same conclusions yourself, in what takes a team days, or an individual months this tool could do in seconds.
This is not so much the sentiment of the text, but the obfuscation, and the focus. There a key comments CEO's make. Take Tricia Griffith, a very customer focused person, very focused on the team and their training, how the company can grow and safely manage its float, the companies return on its invested capital. If you compare that to a company that talks non-stop about its SharePrice (That was simply manipulated via buybacks with money borrowed on low interest loans anyway, something that shows up with the Accounting Scan). Or constanly draws attention to made up non-gaap metrics, then you have what I think this quote sums up nicely. “The CEO who misleads others in public may eventually mislead himself in private.” So the context of the words is more important, and we have a ton of AI research in that area, its used millions of times a second and is freely available on several AI aa S platforms, Spam filtering. Some transfer learning, aspect mining with the right data and you have an amazing well performing model. The gold here is not the AI, not at all, its the data used to train the AI and that I have in the 1000's. My work on the AI for my personal investing has been very effective.
An examples of SOTA transfer learning if interested
https://arxiv.org/abs/1801.06146
I think these are great first stabs :) Let me push you a bit, if you're down.
Okay, but what are they using currently to solve this problem?
If they have a team of ten doing this analysis full-time, you better bet they care about quality and correctness of the answer. If so, you need to rethink your positioning here, because they will never buy a $25/mo product that works only 'okay' (from their perspective).
If they are a day-trader trying to read newspaper articles for two hours a day, every day, then you've found a market at the price point you care about, but still have to prove that they will buy it instead of continuing with their current habits.
If you start interviewing users and they don't do anything to solve this problem, you're putting yourself in a risky position.
I wrote about why it's risky in this post: https://www.indiehackers.com/post/you-need-to-understand-the-3-reasons-people-buy-0d952dadde
My gut feel is that either this is super useful as a product and you're way under-charging, or the 'good-enough' solution isn't good enough to bet stocks on.
If the software is so good, why put the effort into selling it at $25 a month?
If you really can live up to your claims, it feels like you need to sell this thing for $1,000/mo, minimum.
Thanks again, really appreciate you taking the time to help here.
This tool would be one piece of the pie, you couldn’t make an investment solely on this tool, it’s actually the opposite. The tool helps you weed out companies that are committing fraud or trying to mislead or just are more worried about their next bonus than the growth of the company.
I’ve seen screeners, good quant screeners not your yahoo screener charging $500 - $3,000 a year. That’s going back 5 years and just one market, say NYSE. If you want 20 years of data over multiple markets the price goes up and up. $24 per user per month is for the last 5 10k, the last 15 quarterly calls in one market. The up sell is more history, more markets, ability to run your own text through such as news articles or interviews that you have found. But that’s not saving much time, you could just read them yourself. Searching for all candid, honest, internal focused companies, with documents released in the last 2 months, then filter out those with a +80% likely hood of fraud and show the fundamental numbers. Thats what you use to make your investment decision.. But this has made me think, the quant side is needed, it really needs to be a tool that offers the full pie.
This isn’t day trading or swing trading it’s picking solid companies to hold for a few years, would be called GAAP. Even with solid returns of 10%-15% IRR above the benchmark (which is exceptional) its only useful if you can deploy it on large sums of capital. If your starting with a few 100,000 then your only growing it at 30,000 a year, and if your taking that out you limiting the whole compound interest goal.. It’s not enough to live on..
But 500 users paying $24 a month.. Which may be more users than there are but that’s what pays the bills.. Or just adds to your capital for growth.
Some good thoughts though.. Gives me a perspective on actually there needs to be more to the tool.. It also needs a fair few users.. Maybe by adding the value-quant side it can charge more and attract more users..
Glad it's helpful :) I know next to nothing about your industry, so I'm moreso just raising questions that hopefully help you explore for yourself.
Hope you are also talking to some potential users as well!
Sadly there more enterprise users.. Making it enterprise makes a lot more money but then you have to manage a large team to support that. It’s really painful managing a large team.. My dream is to be a solo developer..
Thanks for the questions, it’s not just helped me validate this idea but given me a framework to use on future ideas too.
Awesome!