When I run these tests, my database had about 3900 stocks. Every day I run a cron job to check delistings from SEC fillings. Then, this number decreases a few units every day.

So, replying your bullet points: 1. Full data is about 3900 common stocks listed in Nasdaq and NYSE.

2. I had outperformances for both 5y, 10y and 20y ago until Today (first chart).

3. I don't understood your point 3. Seems the opposite of what happened to me.

4. Agree with you. Market adjusts accordingly to the winners. Also a pull quote from the article. Nice!

Thank You for your comment!

However, this way you'll ended up more exposed to the index than your own choices and decisions. Right? If you had put 80% in the index in 2006 and only 4 stocks (20% of your portfolio) in 'expected good fundamental stocks', yet the majority if your whole portfolio returns would be the 80% of the invested (103% return in 10 years).

So, a conservative strategy definitely could mix a LOT of index/ETF, but I believe this would also drag the returns to be conservative. The exposition to individual stocks (if we could ever be able to have a investing strategy) will produce returns based in the risk you're buying.

Yeah.. After I run the correlation over the full stocks of a window I noticed that maybe that could be a good path. I just didn't wanted to keep digging things and put in this article. It would be a never endless battle to "justify" my stock rankings. Right?

But Consider It done. I'll definitely check my data and see how horrible fundamentals, high debt and other bad things could help us.

One drawback I'm still trying to thing about is the problem with the Cyclical companies. If I take the data as I did before, cyclical usually would rank bad (and they do in my current ranking). So, I'm still thinking how that will distort the data.

I'm not writing using AI. I'm just focusing in this side project for the last 2 years. It's right now my main side-project to transform a hobby in, maybe, a revenue source... So, I've been polishing and reviewing that article a dozen times since the beginning of this month. It has a LOT of iterations with friends on missing Data Analysis arguments and bad english.

Thank you also for your topic comment: That's kind of my direction now. I want to improve the tool to focus in helping people see that a business is not a ticker and a chart. They're about knowing the forces of revenue, net income, controlling cash and debt.

I just wanted to start with a feature that could easily compare my screener to others, so I build the rankings on Fundamental Data (and also Historical return).

My idea is to bring more features in the quality side of analysis: Michael Porter's 5 forces, SWAT table, release earnings comments, etc.

I'm not a finance expert also, but I could learn over the years the (not quantitative truth) that fundamental analysis at least is more reliable than buy moment or day-trading. Right?

So, my conclusion is in line to what you mentioned: The ranking tool I build (and tried to sell as a predictive tool) also fails in that plan. It's actually just too enought to separete the gems from the rubble to help a little bit the investor who wants to consider fundamental analysis...

You're right: I blindly thought I could find a correlation between good fundamentals and historical returns. Since 2024 I've been working in that as a side project. Like, I pivoted the "features" of the web app a LOT of times.. Adjustable weights, added and removed Net Deb/ Ebtida from the score, etc.

The core feeling and reason for me is that I never found a good screener for common people to understand what is a good company to invest. People usually follow hype news and just bet in the stock market. I wanted to create a tool to fight against that.

And only recently, when I was about to launch my website I got AI to review it and noticed the point-in-time bias. :grimmacing:

What do you mean write your own HN posts? (This is my second post)

Comments URL: https://news.ycombinator.com/item?id=49257407

Points: 10

# Comments: 18

So, It's like AI SDK, but for email providers? I'd like that!

All of us attach to a given client until we see that their pricing is going up or that a better solution exists. Then, you want to just switch the provider, not your code that sends the transactional emails, etc.. I'd use your tool

Build Id!

Comments URL: https://news.ycombinator.com/item?id=49167559

Points: 2

# Comments: 0