AI Stock Bot on Telegram: Daily Picks Every morning

What Doosol Points Out

• Every weekday at 7 AM, a Telegram bot sends me a curated list of S&P 500 stocks that are down 15-30% from their 52-week highs — with AI analysis on which ones are worth watching.
• It also recommends sector ETFs based on what’s happening in the world that week (oil crisis → energy ETFs, AI buildout → semiconductor ETFs).
• I’m not a developer. I built this in one afternoon with AI guiding me through every step. Total cost: roughly $0.50-1.00/day for the OpenAI API. Everything else is free.
• The bot also monitors financial news every 2 hours and alerts me when something market-moving happens — earnings surprises, Fed decisions, geopolitical shocks.
• This isn’t a trading bot. It doesn’t buy or sell anything. It’s a research assistant that does the boring screening work so I can make better decisions.

Building an AI stock bot on Telegram sounded like something only developers could do. Turns out, it took me one afternoon. I have a confession. I used to spend my mornings the same way most investors do — opening five different apps, scrolling through news feeds, checking stock prices one by one, and trying to figure out if I missed anything important while I was sleeping.

It took about 30-40 minutes every day. And most days, the answer was: nothing changed much.

So I built a bot that does all of that for me and sends the results to my phone before I even get out of bed.

What It Actually Does

Here’s what my AI stock bot sends me every weekday morning:

Throughout the day, if something big happens — a Fed rate decision, a major earnings miss, a geopolitical shock — I get a news alert with the bot’s assessment of whether it’s bullish or bearish and which stocks or sectors it affects.

All of this runs automatically. I don’t open any apps. I don’t check any websites. The information comes to me.

Why a AI stock bot – Telegram Bot (Not an App)

People ask me why I chose to build my AI stock bot as a Telegram bot instead of a proper app.

It’s on my phone. I don’t need to open a separate app or remember to check a dashboard. Notifications just arrive, like messages from a friend who happens to be really into stocks.

It’s conversational. I can also ask the bot things on demand. Type /us and it runs a fresh analysis right then. Type /add NVDA and NVIDIA gets added to my personal watchlist. Type /mystock and it analyzes all my watched stocks with AI commentary. It’s not a static report — it’s interactive.

Zero friction. There’s no login, no loading screen, no ads, no premium upsell. Just information, delivered cleanly, when I need it.

How It Works (The Simple Version)

The AI stock bot does three things:

1. Scans for beaten-down quality stocks. Every morning, it pulls price data for the top 50 US stocks (S&P 500 components by market cap) and checks which ones are 15-30% below their 52-week high. Not 40-50% down — those might be falling for good reason. The 15-30% range is the sweet spot: enough of a drop to be interesting, but not so much that recovery could take years.

2. Checks the news. For each stock that passes the screen, it pulls the latest headlines from the past week using NewsAPI. This adds context — is the stock down because of a temporary setback, or is there a structural problem?

3. Asks AI to analyze everything. All the price data and news get sent to GPT-4o, which evaluates each stock on fundamentals, valuation, recent catalysts, and long-term outlook. It picks the top 7 and explains why each one is worth watching — plus flags the risks. It also identifies 2-3 sectors that look promising based on current macro trends, and recommends specific ETFs for each.

How I Actually Built It (Honest Account)

I want to be upfront: I don’t code for a living. I work in product management. Everything I know about servers and Python, I learned from AI walking me through it step by step.

Here’s what the process actually looked like:

The plan was simple. A Python script on a cloud server, connected to free APIs for stock data and news, with OpenAI’s API for analysis, sending results via Telegram’s Bot API.

The reality was messy. I accidentally created an SSH key with the filename “yes” because I typed it at the wrong prompt. I tried to run server code on my local machine instead of the remote server — different terminal windows, same-looking black screen. I discovered that Telegram bot commands only support English characters (my first command names didn’t work). I got “Permission denied” errors three separate times for three different reasons.

Each time, I described the error to AI, and it told me exactly what went wrong and how to fix it. The whole thing took about 3 hours, including all the mistakes.

The Commands

Once it’s running, the bot responds to simple commands:

Command	What it does
/us	Fresh analysis of top US stocks + ETF picks
/add AAPL	Add Apple to your personal watchlist
/add TSLA	Add Tesla to your watchlist
/list	See all your watched stocks
/mystock	Full AI analysis of your watchlist
/remove PYPL	Remove PayPal from watchlist
/help	Show all commands

The /mystock command is the most useful one. It checks live prices for every stock on your watchlist, calculates how each one has moved, and asks GPT to give a brief opinion on each — hold, buy more, or sit tight. It’s like having a quick portfolio check-in without opening a single app.

What I’ve Learned After Using It

I’ve been running this bot for about a week now, and a few things have surprised me.

The AI recommendations are decent starting points, not final answers. When the bot recommended Disney, I did my own deeper research and agreed with the thesis — the streaming turnaround story is real. When it flagged Intel, I was more skeptical given the execution challenges. The bot gives you a shortlist — you still need judgment.

The news alerts are more valuable than the stock picks. Getting a push notification that says “🔴 Oil hit $100, energy stocks rallying, here’s what it means for your portfolio” is genuinely useful. It’s faster than checking news apps and more relevant because it’s filtered through an investment lens.

The 15-30% drop filter is surprisingly effective. Stocks that are down 40%+ from their highs are often down for good reason — deteriorating fundamentals, industry decline, management problems. The 15-30% range catches stocks that are temporarily out of favor but still fundamentally sound. It’s not perfect, but it’s a much better starting universe than “everything in the S&P 500.”

I actually check my phone less. This sounds counterintuitive, but because the bot sends me everything I need in one message a day, I’ve stopped compulsively checking stock apps throughout the day. The information comes to me on a schedule. If something urgent happens, the news alert catches it. Otherwise, I can focus on my actual work.

What’s Next

This is version 1.0. Here’s what I’m planning to add:

Connect it to my MCP server. I already built a Claude MCP server for deeper stock research on my desktop (wrote about that here). The plan is to bridge the two: Telegram for quick daily alerts on my phone, Claude Desktop for deep analysis when I want to dig in.

Better screening criteria. Right now it only looks at price drops from 52-week highs. I want to add P/E ratio filters, earnings growth rates, and free cash flow data to make the initial screen smarter before AI even touches it.

Portfolio tracking over time. Instead of manually adding each stock, I want the bot to track price changes since I added them — a simple performance tracker for my watchlist.

Accuracy tracking. Did last week’s recommendations actually go up? I want the bot to track its own hit rate over time so I can calibrate how much weight to give its picks.

Should You Build One?

If you invest regularly, this kind of AI stock bot pays for itself in time saved. The setup is a one-afternoon project, and once it’s running, it’s completely hands-off.

If terminal commands terrify you — wait for my detailed setup tutorial (coming soon), or use the concept as inspiration for what’s possible. The tools are getting easier every month.

The bigger insight for me wasn’t the bot itself. It was realizing that AI + automation + your existing tools can create something genuinely useful in a few hours. Not a toy. Not a demo. Something you actually use every single day.

My phone buzzes at 7 AM. I read the picks over coffee. Some days I act on them, most days I don’t. But I’m always informed, and I never feel like I’m behind.

That’s worth an afternoon of setup.

---

Disclaimer: This bot is a personal research tool, not financial advice. AI-generated stock analysis should be one input among many in your investment decisions. Always do your own research and consult with a qualified financial advisor. The author uses OpenAI’s API (paid) and free-tier data services. Past stock recommendations do not guarantee future performance.