He Built a $24K Bloomberg Terminal For Free With AI. Here Is How.
Prefer to watch? Watch on YouTube · Subscribe for new builds · Read the transcript
A developer named Florin just built a stock screening system that does what Bloomberg Terminal does. It scans all 500 S&P 500 stocks in 15 seconds. It runs Warren Buffett's 10 investment formulas on any stock in 2 seconds. It pulls official SEC filing data. And it cost him exactly zero dollars. Bloomberg charges $24,000 a year for basically the same capability.
This is the full breakdown of how he did it, what tools he used, and how you can build something like it yourself.
Quick summary of what this system does:
- Scans 500 S&P 500 stocks for oversold signals in 15 seconds
- Runs 10 Buffett-style fundamental analysis formulas on any stock in 2-3 seconds
- Pulls real SEC EDGAR filing data (not scraped estimates)
- Uses SQLite caching so repeat scans take under 30 seconds
- Controlled entirely from Telegram on your phone
- Total cost: a few dollars a month in AI API tokens
Why Bloomberg Terminal Costs $24,000 (And Why Most People Don't Need It)
Bloomberg Terminal is the gold standard on Wall Street. Traders, analysts, and hedge fund managers use it for real-time market data, company financials, news, execution, and the Bloomberg messaging network. It's a beast. And it costs $2,000 a month. That's $24,000 a year for a single seat.
But here's the thing. Most individual investors use maybe 10-15% of what Bloomberg offers. They want to screen stocks, check fundamentals, run some technical indicators, and maybe compare a few companies. They don't need the real-time futures execution. They don't need the Bloomberg chat network. They don't need the fixed income analytics.
| Tool | Annual Cost | Best For | Main Limitation |
|---|---|---|---|
| Bloomberg Terminal | $24,000 | Institutional traders, hedge funds | Wildly expensive for retail investors |
| FactSet | ~$12,000 | Research analysts, portfolio managers | Still enterprise pricing |
| Koyfin | $468 (Plus plan) | Serious retail investors | No AI analysis, manual screening only |
| Finviz Elite | $180 | Technical screeners | Limited fundamental depth |
| Morningstar Premium | $249 | Fund research, ratings | Weak on technical analysis |
| DIY AI Agent (this build) | ~$36 (API tokens) | Individual investors who want full control | Requires setup, not real-time execution |
So the question becomes: if you're a regular investor who wants professional-grade screening without the professional-grade price tag, what do you actually need?
The Architecture: Four Layers, Zero Subscriptions
Florin's system uses four components. Each one is either free or costs pennies.
Layer 1: AI Agent (OpenClaw + Claude) handles the orchestration. It's the brain. It takes your natural language commands ("scan for oversold stocks" or "analyze AAPL"), figures out what scripts to run, interprets the results, and sends you a clean summary. This is the only part that costs anything, and it's just API tokens. A few bucks a month depending on how often you scan.
Layer 2: Python Scripts do the actual computation. Three scripts handle three jobs: technical screening, combined screening, and deep-dive fundamental analysis. The AI agent calls these scripts when you ask for something. Simple.
Layer 3: Free Public APIs provide the data. Yahoo Finance (via the yfinance Python library) gives you price data, historical OHLCV, and basic metrics. No API key needed. No signup required. SEC EDGAR gives you official financial statements. Balance sheets, income statements, cash flow. All public, all free, rate limited to 10 requests per second.
Layer 4: Telegram is the interface. You type commands in a chat. Results come back in the same chat. Run it from your phone, your laptop, anywhere. No desktop app required.
That's it. No subscription fees. No data vendor contracts. No enterprise license agreements.
Script 1: The Oversold Scanner (500 Stocks in 15 Seconds)
Type "oversold" into Telegram. Fifteen seconds later, you get a list of every S&P 500 stock trading near its recent low.
The scanner uses an indicator called Williams %R. It measures where a stock's current price sits within its 21-day high-low range. The scale runs from 0 to -100. If a stock scores below -80, it's considered oversold, meaning the price is near the bottom of its recent range and could potentially bounce.
Here's a concrete example. Say a stock hit $100 at its highest point in the last 21 days and $60 at its lowest. If it's currently trading at $62, the Williams %R would read around -95. That's deep oversold territory. The same indicator costs professional traders thousands of dollars to access through Bloomberg's charting tools.
When Florin ran his first scan, he found 93 oversold stocks out of 500. Almost 19% of the S&P 500 in oversold territory. The top five most oversold:
- Visa: -99.3 (price just 0.7% above its 21-day low)
- Axon Enterprise: -98.7
- The Trade Desk: -98.1
- Constellation Energy: -97.6
- Fair Isaac (FICO): -97.5
Professional traders pay real money to see exactly this data. This system surfaces it in 15 seconds for free.
Script 2: Combined Screening (Technical + Fundamental)
The second script goes deeper. It combines the Williams %R oversold detection with fundamental analysis from SEC filings. So you're not just finding stocks that are technically cheap. You're finding stocks that are technically cheap AND fundamentally sound.
This scan takes 1-3 minutes because it's pulling deeper data from SEC EDGAR for each stock. But the result is a shortlist of companies that pass both the technical and fundamental filters. You get stocks that are trading near their recent lows with strong cash positions, manageable debt, and consistent earnings.
That combination is powerful. Most free screeners give you one or the other. Technical tools like TradingView show you chart patterns but don't dig into 10-K filings. Fundamental tools like Finviz show you ratios but miss the timing signals. This system does both in one shot.
Script 3: The Buffett Analyzer (10 Formulas in 2 Seconds)
This is the one that made me say holy moly.
Type "analyze AAPL" and in about 2-3 seconds, the AI agent runs Warren Buffett's actual 10 investment formulas on Apple using real SEC EDGAR data. Not estimates from a third-party website. Not scraped numbers from some finance blog. Official SEC filings that companies are legally required to submit.
The 10 formulas cover:
- The Cash Test - Does the company have sufficient cash reserves?
- Debt Ratio Analysis - Is the company overleveraged?
- Profit Margin Check - Are margins healthy and stable?
- Return on Equity (ROE) - Is the company generating strong returns on shareholder investment?
- Earnings Consistency - Has the company produced reliable earnings year over year?
- Revenue Growth - Is top-line revenue trending upward?
- Book Value Growth - Is the underlying value of the company increasing?
- Capital Expenditure Ratio - How much is reinvested vs. available as free cash flow?
- Current Ratio - Can the company cover short-term obligations?
- Retained Earnings Trend - Is the company reinvesting earnings effectively?
An analyst at a hedge fund might spend 30-60 minutes manually pulling these numbers from SEC filings and running the calculations. This system does it in 2 seconds. And it uses the exact same source data.
The Secret Sauce: Smart Caching With SQLite
Here's where Florin got really clever. The first time you run a full scan, it takes 3-5 minutes because it's fetching fresh data from Yahoo Finance and SEC EDGAR for 500 stocks. Every scan after that? 15-30 seconds.
The trick is SQLite caching with smart expiration rules:
- Price data: cached for 1 day (prices change daily, so you want fresh data each session)
- Fundamental data: cached for 7 days (SEC filings don't change daily, quarterly at most)
This means you're only hitting the APIs when you actually need new data. No wasted requests. No rate limit issues. And your scans go from minutes to seconds after the initial load. It's the difference between a demo and a tool you actually use every morning.
What Bloomberg Has That This Doesn't (The Honest Comparison)
I'm not going to pretend this is a perfect Bloomberg replacement. It's not. If you're a professional trader at a bank, you still need Bloomberg. Here's what's missing:
- Real-time execution: Bloomberg lets you trade directly from the terminal. This system is analysis-only. You still need a broker.
- Fixed income and derivatives data: Bond pricing, options chains, credit default swaps. Bloomberg's fixed income data is unmatched and this build doesn't touch it.
- The Bloomberg messaging network: IB Chat is how Wall Street communicates. There's no replacement for that social layer.
- Real-time streaming data: Yahoo Finance has a slight delay. Bloomberg is instant. For day traders, that matters. For swing traders and long-term investors, it doesn't.
- Multi-asset coverage: Bloomberg covers every asset class globally. This build focuses on US equities via S&P 500.
But here's the flip side. If you're a regular investor looking for good stocks to buy, checking fundamentals, and screening for technical setups, this covers probably 80% of what you'd actually use Bloomberg for. At 0% of the cost.
Why This Pattern Matters (Even If You Don't Trade Stocks)
The stock screener is cool. But the real lesson here is the pattern. This developer took an industry where the standard tool costs $24,000 a year and built a functional alternative for free using AI.
That pattern works everywhere:
| Industry | Expensive Tool | Typical Cost | AI Agent Opportunity |
|---|---|---|---|
| Finance | Bloomberg Terminal | $24,000/year | Stock screening + SEC analysis |
| Real Estate | CoStar / LoopNet Premium | $6,000-15,000/year | Property analysis + market comps |
| Legal | Westlaw / LexisNexis | $10,000+/year | Case research + document analysis |
| Marketing | SEMrush / Ahrefs (full suite) | $1,200-5,000/year | Keyword research + competitor analysis |
| Recruiting | LinkedIn Recruiter | $8,000-12,000/year | Candidate sourcing + outreach |
The formula is always the same: find the expensive tool, identify the 80% of features people actually use, connect free or cheap data sources to an AI agent, and build a focused alternative.
How to Build Your Own Version (Step by Step)
If you want to build something similar, here's the path:
- Pick your data sources. For stock screening, Yahoo Finance (yfinance library) and SEC EDGAR are your foundation. Both free. Both well-documented.
- Write your screening scripts. Python is the move here. The yfinance library makes pulling price data a one-liner. SEC EDGAR has a REST API you can query with basic HTTP requests.
- Set up caching. SQLite is lightweight and built into Python. Cache price data for 1 day, fundamental data for 7 days. This is what makes it fast enough to use daily.
- Connect to an AI agent. OpenClaw with Claude gives you the natural language interface. The agent calls your Python scripts, interprets results, and responds in plain English.
- Add a chat interface. Telegram is the easiest. You get mobile access for free. Type "oversold" from your phone while you're having coffee and get results before you finish the cup.
The whole setup can be running in a weekend if you focus. It's not a massive engineering project. It's three Python scripts, a SQLite database, and an AI agent to tie it all together.
When This Is NOT For You
Real talk. This build is not the right move for everyone.
- If you're a day trader who needs millisecond execution and real-time Level 2 data, this won't cut it. Yahoo Finance data has a delay. You need a direct market data feed for scalping.
- If you trade options or fixed income, the data coverage isn't there. Options chains, bond pricing, and derivatives analysis require specialized data feeds that free APIs don't provide.
- If you want a polished GUI with charts, this is a chat interface. No candlestick charts. No drawing tools. If you need visual technical analysis, pair this with TradingView's free tier for charting.
- If you're not comfortable running Python scripts, there's a learning curve. It's not plug and play. You need basic comfort with command line tools and editing config files.
For long-term fundamental investors and swing traders who want to find good stocks without paying thousands of dollars? This is exactly what you need.
FAQ
Can you really replace Bloomberg Terminal for free?
For individual investors doing fundamental and technical screening, yes. This system covers about 80% of what retail investors use Bloomberg for. The missing 20% is real-time execution, fixed income data, and the Bloomberg messaging network. Most individual investors never touch those features anyway.
What APIs does a DIY Bloomberg Terminal use?
Two free APIs handle most of the work. Yahoo Finance (via the yfinance Python library) provides real-time price data, historical OHLCV, and basic fundamentals with no API key required. SEC EDGAR provides official financial statements including balance sheets, income statements, and cash flow data. Rate limit is 10 requests per second.
How much does Bloomberg Terminal cost in 2026?
Bloomberg Terminal costs $24,000 per year ($2,000 per month) for a single seat. Enterprise pricing with additional terminals typically runs $20,000 per additional seat. Other professional alternatives like FactSet cost around $12,000 per year, and Morningstar Premium runs $249 per year for limited screening.
What is Williams Percent R and how does the AI stock screener use it?
Williams %R measures where a stock's current price sits within its 21-day high-low range on a scale of 0 to -100. Below -80 means the stock is trading near its recent low (oversold territory), which can signal a potential bounce. The AI screener scans all 500 S&P 500 stocks and flags every stock below -80 in about 15 seconds.
Is this AI stock screener legal to use?
Yes. All data sources are publicly available. Yahoo Finance data is free for personal use. SEC EDGAR is a government database that companies are legally required to file with. The AI agent runs on your own machine analyzing public data. No scraping of paid services, no terms of service violations.
How long does it take to set up?
If you're comfortable with Python and command line tools, you can have a basic version running in a weekend. The three core scripts (oversold scanner, combined screener, Buffett analyzer) plus SQLite caching and Telegram integration. Most of the time goes to writing the screening logic. The AI agent setup takes about 30 minutes.
The Bottom Line
A regular developer, not a Wall Street quant, built a stock screening system that runs 10 Buffett formulas in 2 seconds and scans 500 stocks in 15 seconds. Using free public data. Controlled from his phone. For the cost of a few dollars a month in API tokens.
Bloomberg Terminal is still the best tool money can buy. But most people aren't buying it. This is the 80/20 version. The version a regular investor can actually use.
Whatever expensive tool exists in your industry, the same pattern applies. Free data sources + AI agent + smart caching = a tool that handles the heavy lifting for pennies.
If you want to learn how to build projects like this and actually ship them, come join us in Shipping Skool. We do live build sessions six times a week and share real use cases every single day. Not just ideas. Real projects that real people are building and shipping.
Disclaimer: This article is for educational purposes only. Nothing here is financial advice. Always do your own research before making investment decisions.
Chapters
Transcript
Read the full transcript
0:00A guy just built his own Bloomberg terminal with open claw and it didn't even cost him a single dollar. And if you don't know what a Bloomberg terminal is, let me put it this way. It's the tool that Wall Street traders use to scan stocks, analyze companies, and find investments. And it costs $24,000 a year. That's $2,000 a month just to look at stock data. And this guy basically built the same thing for free with open claw running from his phone and I'm going to break it down exactly how he did it step-by-step because this is one of those use cases that is insane. And I get comments and messages all about how can I build a trading bot? How can I analyze stocks better? And how can I make money with this open claw
0:46that actually makes money for me 24/7? This can be a step into that journey down that road building out something similar, which is why I'm making a video for you because I do think this will add value if you are in the finance space. So, I'm going to back up bit and give you some context to paint a picture. This developer, his name is Florin. He posted the whole thing on Medium. He's an everyday investor, you know, not a hedge fund manager and not a Wall Street trader, not some quant at Goldman Sachs. He's just a regular person who is interested in qualitative investing and wanted to find some good stocks to buy. Me, too. You know, the kind of stocks Warren Buffett would buy. Nothing flashy, simple long-term investments. Companies with strong cash positions, low debt, consistent earnings, that kind
1:31of stuff. Here's the problem, though, and this is what Florin ran into is all the tools that actually do like professional stock screening cost a ton of money. Bloomberg terminal is insane. $24,000 a year, I mean, come on. Comment down below if you have that. That's insane. FactSet, which is $12,000 a year, even something like Morningstar Premium, which is $249 a year, and it barely does the kind of stuff that this deep analysis he was looking for like what he built out with open claw. Trade Ideas is $118 a month just for basic scanning on that app. So, you're looking anywhere from like $1,500 to $24,000 a year just to screen stocks like a professional. That's insane. Most regular people can't afford that. Let's
2:16just say nobody that I know is dropping $24,000 a year to get stock data. So, what did Florin do? He built his own with open claw and free public APIs that anyone can access and you can build the same setup yourself. So, when I saw the results, I was kind of going crazy because this is actually better than some paid tools I've seen people use. Like, this isn't some toy demo. This is a real production system that scans 500 stocks in 15 seconds. So, here's what he's built. He set up an open claw agent on his machine running Claude Opus, connected it to Telegram so he can control everything from his phone. Then he wrote a three Python script base that has his agent running on command and each one does something different. The first one is a technical screener. You
3:02just type in the word oversold into your Telegram chat and in 15 seconds his agent scans all 500 stocks in the S&P 500 and tells you which ones are at the lowest price point right now. 15 seconds, 500 stocks from your phone while you're eating at Taco Bell. That's amazing. The second script is full of a screening tool in more of like technical analysis plus fundamental analysis combined. So, this one takes about 1 to 3 minutes because it's pulling deeper data, but it does give you the complete picture. Not just whether a stock is cheap, but whether the company behind it is actually a solid company. And the third one is a deep dive analyzer. You type in analyze Apple and it runs Warren Buffett's actual 10 investment formula
3:47on Apple using the official SEC data. You know, not some random website scraping numbers from who knows where, actual SEC filings, the same data Buffett himself reads in his 10K annual reports. That's the sauce for the sauce on the sauce right there and I love it. Who doesn't love sauce? So, I'm going to break it down the technical side because I love technical analysis. You know, it's something that I like to look at on the side. I have my charts up here and I just like to monitor certain stocks. You know, and I love doing it. You know, I'm a data-driven person. Real quick, before we keep going, if you're watching this and you want to actually build with some of these tools, not just watch videos about them, then you're going to want to check out our community down below, Shipping School. We have a full Claude
4:33code course, a full open claw course, and four live bootcamps every single week where we actually help you get set up from scratch. Like, actually set this thing up, not just watch a tutorial and figure it out by yourself. And we also provide one-on-one coaching so you can book a call with me. We can share screens and I can help you get Claude code or open claw running on your machine. That's it. No fluff. I built this community because watching YouTube's only gets you so far. We launched it just 3 days ago and we have over 55 members. You need people around you who are actually building, people who hold you accountable, and coaches who can help you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. So, this is a really simple
5:18build, way simpler than you might think. You know, for price data, he uses Yahoo Finance through their Python library called YFinance. It's completely free. No API key needed, no sign up required. And the thing is, it's the same data source that TradingView and Robinhood use. It's crazy. So, you're getting the exact same quality price data that professionals get. You just aren't paying thousands of dollars for a fancy interface to look at. For fundamental data, he pulls directly from the SEC Edgar API and this is key. This is the United States Securities and Exchange Commission's own database, official government data. Companies are legally required to file their financial statements here. Balance sheets, income statements, cash flow statements, all of
6:03it, public and free and accessible through a simple API call. You just need to respect the rate limit of 10 requests per second, which is more than enough for scanning 500 stocks. Here's where Florin got more really smart and just blew my mind on this. It's he built a caching layer using SQLite databases. So, the first time you run the scan, it takes about 3 to 5 minutes because it's fetching all of the price data and fundamental data fresh from the internet. But every scan after that, 15 to about 30 seconds because it caches all the price data for 1 day and the fundamental data for 7 days. So, the only thing you're hitting with the APIs is, you know, when you actually need fresh data. That's just good engineering right there. That's the kind of thing
6:48that separates a demo from an actual production tool. I'm going to explain what the oversold scanner actually does because you don't need to be a trader to understand this. It basically uses an indicator called Williams percent and, you know, don't let the name scare you. It's basically just measures where a stock's current price sits within its recent 20-day high or low range. Like, imagine a stock hit $100 at its highest point and then in the last 21 days, it was at $60 at its lowest. So, if it's trading right now at $62, it's way down near the bottom on the RSI. You know, it's that's what Williams percent measures. If a stock scores below negative 80, it's considered oversold, meaning if people might be panic selling
7:34at that price and it could potentially bounce back. You know, that's how it's measuring it. So, I got to say this is not financial advice at all. I'm just talking about how the tool, but it is the same exact indicator that professional traders use every single day on their expensive terminals. When Florin ran his scanner on January 31st, he found that 93 oversold stocks out of 500. That's almost 19% of the entire S&P 500 in oversold territory. The top five most oversold were Visa at negative 99.3, Axon Enterprise at negative 98.7, The Trade Desk at negative 98.1, Constellation Energy at negative 97.6, and Fair Isaac at negative 97.5.
8:22So, when a stock like Visa hits negative 99.3, it means the current price is only 0.7% above its 21-day low. That's an extreme oversold territory and that's the kind of signal professional traders pay big money to detect. Here's where it gets even better. The analyze command runs Buffett's actual 10 investment formulas on any individual stock you pick. So, things like cash test, does the company have enough cash? Debt ratios, is the company over-leveraged? Profit margins, are they actually making money efficiently? Return on equity, are they using shareholder money well? Earnings consistency, do they make money reliably year after year? All 10 formulas, all pulled from real SEC filings, and all
9:07calculated automatically by the agent. And it's all delivered to you on your phone in about 2 to 3 seconds. 2 to 3 seconds to run 10 Buffett formulas. That used to require an analyst in a Bloomberg terminal and now it happens in seconds. So, I'm going to connect this back to why this matters for you even if you don't care about stocks at all. The pattern here is what matters. The guy took something that cost $24,000 a year and rebuilt it for free using an AI agent with an open claw. You can do the same. Think about that for just a second. Let it sink. Wow. Whatever industry you're in, there's probably an expensive tool you're paying for that open claw agent can either replace entirely or supplement in a major way. And look how clean the architecture is. Open claw handles the orchestration and the brain. Python scripts do the heavy
9:52computation. SQLite caches the results so it stays fresh and fast. And Telegram gives you the interface so you can run it from your phone anywhere in the world. That's four layers and not a single one of them cost anything beyond your normal API tokens for Claude. This is what I talk about in the community and shipping school all the time. We have over 136 members in the past 13 days. You know, right now every single week when someone shows up with a cool use case like this, that just makes my jaw drop and it confirms what we're doing, you know, whether you're building a stock screener, a real estate analysis tool, a customer support system, an inventory tracker, whatever it is. The pattern is always the same. Find the expensive thing, build the cheap version with Open Claude, let the agent run it for you and that is exactly what Florian
10:37did. The total cost to run the system just as API tokens for Claude through Open Claude. Maybe a few bucks a month depending on how often he scans, but there's no Bloomberg subscription, no FactSet license, no expensive data feeds, just Open Claude, free public APIs and some clever Python scripts. And now he has a tool that hedge funds and trading firms pay $24,000 a year for. From his home, while eating dinner, at Taco Bell. Can I get an amen on that? That's insane. So, if you're sitting there thinking, I don't know what to build with Open Claude, here's my advice. Look at the expensive tools in your industry. Look at what people are paying hundreds or thousands of dollars a year for and ask yourself, can an agent do 80% of this? Because more and more the answer is yes. We're living in
11:22an amazing time where regular people with no Wall Street connections can compete with billion-dollar firms from their phone and that is crazy. And what it's only going to do is get more wild from here. And if you want to learn to build stuff like this and actually ship real products around real people, come join us in the community. I'll put the link in the description. We do six live build sessions a week. We share use cases like this every single day and we help each other actually build and get things pushed to production, not just talk about ideas, real projects that make real money. And honestly, this is what gets me so fired up where we are right now. It's insane. The playing field is getting leveled. You don't need a finance degree. You don't need a venture capital. You don't need connections on Wall Street. You don't need any of that. You just need Open Claude and some curiosity and the
12:08willingness to learn. That's it, plain and simple. If you guys haven't subscribed to the channel already, we release four to six videos every single day and if you like AI use cases like this, specifically Claude code and Open Claude, hit subscribe, hit the notification bell cuz we got a video coming out in about an hour. So, we'll See you in the next one. Be blessed.
Ready to start building with AI?
Join the Shipping Skool community free. Paid members get live bootcamp calls every week and weekly-updated courses.
Join Shipping SkoolKeep reading
Claude Code Just Doubled Your Limits. Here Is How I Am Using Every Last One.
Anthropic just doubled Claude Code usage limits during off-peak hours. This is how I restructured my entire AI agent pipeline to take advantage before it expires March 27th.
I Quit ChatGPT For Good. Here Is What I Use Instead.
I was spending 3-4 hours a day copy-pasting context into ChatGPT. Then I found OpenClaw and everything changed. Here is why I am never going back.
Anthropic is Quietly Crushing OpenAI (While Everyone Sleeps)
Anthropic hit $2.5B revenue while launching the Institute and deepening partnerships. Here's why Claude is quietly winning the AI race.
