I’m creating an AI stock analyzer that rates stocks. Right now, it rates based on four categories: Industry, Financials, Technicals, and Sentiment, each rated from 0 to 20, giving a total rating out of 80. Here’s the flow: https://imgur.com/a/dbh3cYk
I’m curious, what’s your approach to stock analysis? I want to use your ideas in my AI agent and feed it real-time market data and tools.
Would love to hear from someone more experienced in finance about how you approach analyzing stocks.
There isn’t really a single ‘right’ way to do it. If you want an accurate model, you need accurate assumptions, and that’s really tough. Most fund managers focus more on the performance of the whole portfolio than individual assets.
@Kai
But surely there must be a way to rate individual stocks or companies, right? How do you decide which stocks to buy, sell, hold, or add/remove from your portfolio?
Zan said:
@Kai
But surely there must be a way to rate individual stocks or companies, right? How do you decide which stocks to buy, sell, hold, or add/remove from your portfolio?
Sure, you could do a discounted cash flow (DCF) analysis to figure out a company’s real value. You could also compare companies in the same industry using metrics like P/E ratios. Another option is technical analysis, though I don’t trust it. There’s no proof that it actually works.
The downside of DCF is that it relies on assumptions that might not be correct. DCF gives you an ‘intrinsic’ value based on future cash flows discounted back to today.
Comparative analysis, though, just compares companies to each other. It doesn’t give you the company’s true value. Often if one company in an industry is overvalued, the others are too. Take the 2001 tech bubble, for example.
@Kai
Thanks for the detailed response. Do you think there’s any value in offering a tool that can automatically do the DCF, comparative analysis, and other analyses for people, even if it’s not super accurate in the long term?
@Zan
Not really. Automating DCF doesn’t make sense because each company is unique. A good DCF is personalized to the company, taking its specific factors into account.
Zan said:
@Kai
But surely there must be a way to rate individual stocks or companies, right? How do you decide which stocks to buy, sell, hold, or add/remove from your portfolio?
But why does there need to be a way to predict the future?
@Pippin
If there’s no way to predict stocks, then it’s all random. Wouldn’t all companies perform the same under the same conditions and talent? You can’t predict the future, but you can figure out which companies have a better chance of doing well.
@Zan
You’re assuming that ‘stock value’ ties to ‘future profit’ or ‘assets,’ but that’s not really the case anymore. Stock value nowadays is about whether the price will go up or down, not whether the company is actually making money or has good products.
Take Tesla. It’s hugely overvalued, and the products are mediocre. Yet, the stock price doesn’t show that. Why? Or look at JC Penny—had tons of assets, but stock dropped below the value of its property. Crazy, right?
@Zan
It’s just too complex. It’s unpredictable, too volatile in the short-term. Few people are good at predicting, and those who are don’t build AI models for short-term trading, at least not the ones who would share their methods.
Zan said:
@Kai
But surely there must be a way to rate individual stocks or companies, right? How do you decide which stocks to buy, sell, hold, or add/remove from your portfolio?
I don’t bother. I just buy the index.
People who can consistently beat the market are paid huge sums to find market edges.
Here’s a link showing the pay for people in quant roles: https://www.reddit.com/r/quant/comments/18lst38/2023_quant_total_compensation_thread/
Any so-called ‘free lunch’ from finding certain stock patterns is already being taken advantage of by huge firms with tons of resources. Invest passively, and keep your hobby just that. You’ll end up with more money in the long run.
@Phoenix
I didn’t mean to focus on beating the market. I just wanted to create an automated system that rates stocks daily, weekly, or monthly. This would give individual investors the ability to access ratings and updates, kind of like having a decent financial analyst look at the major stocks each month at an affordable price.
@Zan
If you can’t beat the market, then what’s the value of what you’re offering?
Orion said:
@Zan
If you can’t beat the market, then what’s the value of what you’re offering?
Orion said:
@Zan
If you can’t beat the market, then what’s the value of what you’re offering?
Your comment has been removed due to violating the forum’s rules. See here for more info. If you think your comment was removed by mistake, please message the moderators.
I’m a bot, and this action was performed automatically. Please contact the moderators if you have any questions or concerns.
Orion said:
@Zan
If you can’t beat the market, then what’s the value of what you’re offering?
Well, no one I know can afford to hire an elite analyst.
AI models are advancing fast. Soon they’ll be as smart as the top experts.
I’m building AI agents that can use data from services like Alpha Vantage to do what those analysts do, for a fraction of the cost. It’ll be accessible to people who can’t afford to hire expensive analysts, but still want professional analysis.
If those models keep improving, AI might replace human analysts entirely.
@Zan
That doesn’t answer the question.
If you can’t beat the market, then what’s the point of paying for your analysis? The real value is in providing info that can help you beat the market. If your analysis doesn’t do that, why would anyone pay for it?
@Orion
I’m just starting with tests of the AI agent. I can’t say it will beat the market yet because it’s early in development. Right now, it just analyzes data from Alpha Vantage and gives ratings on stocks.
If top analysts can beat the market using data analysis, I believe this AI agent could do the same eventually.
Even if it doesn’t beat the market, it’s still better than most analysts at analyzing data. It can handle thousands of assets every day and give daily, weekly, or monthly updates for specific stocks in your portfolio.
The point is, the AI agent is efficient—it does the work of 100 analysts at a fraction of the cost.
@Zan
People are misunderstanding your approach. I think what you’re doing is about data aggregation and presentation. You’re not claiming to beat the market, just making it easier for individuals to get solid, automated data for their own research. Sure, you’ll face competition, and some people would rather do it themselves, but for others, your tool could be useful. I’m building my own stock dashboard too, using Python and yfinance, for example.