AI Trading Bot: How This Marketer Built ₹0 Stock Alerts Using Claude Code

I’m a marketer. I don’t write code. I write subject lines, campaign briefs, and Google Sheets formulas that occasionally work. But last month, I built an AI trading bot that sends me live stock alerts on Telegram every trading day — and I didn’t write a single line of Python myself.

Claude Code + VS Code + plain English descriptions = fully functional trading alert system.

No CS degree. No ₹50,000 trading course. No mysterious GitHub repos I pretended to understand. Just a marketer who got tired of staring at candlestick charts during meetings and decided to automate the staring.

Here’s exactly how I built it, what it cost (spoiler: ₹0 in code), and why this matters even if you’ve never opened a terminal in your life.

In This Post

• Why a Marketer Built a Trading Bot
• My Journey — From Manual Chart-Staring to AI Alerts
• How to Build Your Own AI Trading Bot (Step-by-Step)
• Results, Costs & What I Actually Learned
• Frequently Asked Questions (8 Q&As)
• The 7-Day AI Builder Challenge
• Conclusion — You Don’t Need to Be a Developer

Why a Marketer Built a Trading Bot

Let me be honest. I didn’t start this because I had a grand vision for “democratizing algorithmic trading.” I started because my screen time report showed I was spending 45 minutes every morning watching Nifty 50 charts instead of doing my actual job.

My boss at TechnoRUCS never said anything. But my campaign deadlines definitely did.

Here’s the thing about intraday trading as a side interest:

• It’s time-consuming — you need to watch charts between 9:15 AM and 3:30 PM, which is also when your actual job happens.
• It’s emotionally draining — FOMO on a breakout you missed because you were in a meeting = instant mood destroyer.
• Most “trading tools” cost ₹5,000-₹20,000/month — for a side interest, that’s insane.
• The information overload is real — 50 YouTube “gurus” saying 50 different things. Signal-to-noise ratio = basically zero.

So I thought: what if I could build something that watches the charts FOR me and just pings my phone when there’s a real setup?

My colleague Ramesh laughed when I told him. “Bro, you can’t even fix the office printer. You’re going to build a trading bot?”

Fair point. But I had Claude Code. And Claude Code doesn’t judge your printing skills.

My Journey — From Manual Chart-Staring to AI Alerts

Week 1: The “I Have No Idea What I’m Doing” Phase

I installed VS Code, added Claude Code via npm, and typed my first prompt: “I want to build a system that watches Indian stock prices and sends me Telegram alerts when there’s a good intraday trade setup.”

Claude Code didn’t laugh. It asked clarifying questions, suggested Angel One’s SmartAPI for live price data, and wrote the first version of the connection script. I understood maybe 30% of what it generated. But it worked.

Week 2: The “It Actually Sends Me Messages” Phase

Set up the Telegram bot via BotFather in 5 minutes. Connected it to the script. First test alert hit my phone at 9:47 AM during a marketing standup. My phone buzzed: “RELIANCE — Long Entry ₹2,847 | SL ₹2,831 | Target ₹2,879.”

My teammate Priya saw my face and said, “You look like someone who just got a marriage proposal from their code.”

She wasn’t entirely wrong.

Week 3-4: The “Let Me Break Everything and Rebuild” Phase

Built and discarded two earlier versions before landing on the final strategy:

• Version 1: An end-of-day swing screener that mixed fundamentals (PE, PB, ROE) with technicals (EMA, RSI, MACD). Result: too slow for intraday.
• Version 2: A scoring engine that tried to do everything. Result: overkill. Analysis paralysis in bot form.
• Version 3 (final): Opening Range Breakout (ORB) — simple, rule-based, fast. Result: actually usable.

Lesson learned: complexity ≠ quality. The simplest strategy won.

Month 2: The “It Runs Without Me” Phase

Moved the bot from my laptop to a free Oracle Cloud instance (India South, Hyderabad). Now it runs on weekday schedule — market hours only — without me touching it. I check my phone, see alerts, decide if I want to trade. No chart-watching. No FOMO. No angry deadlines.

My friend Karthik, who’s been trading for 5 years with paid tools, saw my setup and said: “Wait, this is free? And YOU built it?”

Marketing degree: 1. Computer Science degree: 0.

How to Build Your Own AI Trading Bot (Step-by-Step)

Total cost = ₹0. Total coding knowledge required = ability to describe what you want in English.

Step 1: Set Up Angel One SmartAPI

Angel One is the brokerage that powers the system — live price ticks + intraday candle data through its SmartAPI.

What you need from your Angel One account:

• API key — generated from the SmartAPI developer portal
• Client code — your Angel One trading account ID
• Trading PIN — same PIN you use to log into the app
• TOTP secret — for 2FA, handled in code via pyotp library

Pro Tip: Store all four in a .env file (via python-dotenv). Never hardcode secrets. Claude Code set this up for me and explained why — security isn’t optional, even for side projects.

Step 2: Create Your Telegram Bot

1. Message @BotFather on Telegram → run /newbot
2. Name your bot → get a unique bot token
3. Start a chat with the new bot → get your chat ID
4. Drop both into the .env file

Time taken: 5 minutes. The Telegram bot now receives formatted messages — entry price, stop loss, target, position size — the moment a setup triggers.

Step 3: Set Up VS Code + Claude Code

• Language: Python (Claude Code writes it for you)
• Core libraries: smartapi-python, pyotp, pandas + pandas-ta, python-dotenv
• Supplemental: yfinance for backtesting
• Runtime: Node.js (required for Claude Code)

My process: Describe what I want in plain English inside the VS Code chat panel. Claude Code writes, tests, and explains. I never had to open a terminal and guess.

Step 4: Implement the Strategy (ORB)

The core strategy = Opening Range Breakout (ORB) — one of the simplest, most back-tested setups in day trading.

How it works:

• Script marks the high and low of first 15 minutes (9:15-9:30 AM)
• Break above the range + volume confirmation = long signal
• Break below = short signal
• Entry, stop loss, and target calculated automatically
• Position size set using ATR (Average True Range) so risk per trade stays consistent

Step 5: Deploy to Cloud (Free)

Oracle Cloud free-tier → India South (Hyderabad) region. Bot runs on a weekday schedule during market hours. Cost: ₹0/month.

The loop: Live price data comes in → script checks against opening range → confirmed breakout gets sized and sent to Telegram. No breakout = no message. It just keeps watching. Like a very patient, very emotionless version of me.

Common Mistakes (I Made All of Them)

• Trying to automate execution: Don’t let the bot click “buy” automatically. Alert-only is the right scope. Trust me.
• Ignoring brokerage costs: STT, brokerage, and slippage eat into profits. A target that looks good gross can be barely profitable net.
• Following “trading gurus”: Most YouTube trading content is hype-driven garbage. Filter aggressively.
• Paper trading too long: Paper trades don’t carry psychological weight. Start live with small capital.

Results, Costs & What I Actually Learned

Let’s talk numbers.

Total build cost = ₹0.

Total saved vs. paid trading tools = ₹5,000-20,000/month.

Annual savings = ₹60,000-2,40,000. That’s a vacation in Sri Lanka — check our Sri Lanka guide if you need ideas.

What Actually Changed

• Morning screen time: 45 min/day → 5 min/day (just checking Telegram alerts)
• FOMO during meetings: Gone. Bot watches. I work.
• Trading decisions: Emotion-driven → rule-based. The bot doesn’t panic. The bot doesn’t revenge-trade.
• Confidence in building things: 0 → “I can probably automate anything.” Dangerous energy for a marketer to have.

The biggest learning? You don’t need to be a developer to build developer-level tools. You need clear thinking + the right AI assistant. If you can write a marketing brief, you can describe what you want to Claude Code.

Marketing brain + AI tools = unfair advantage.

Frequently Asked Questions

Q1: Do I need to know Python or coding to build this?

No. That’s the whole point. I’m a marketing professional with zero coding background. Claude Code inside VS Code handles all the Python — I just describe what I want in plain English. The AI writes the code, explains what it did, and fixes bugs when things break. If you can write a clear email, you can build this. My colleague Priya started building her own version after watching me — she’s a content writer, not an engineer.

Q2: Does the bot actually place trades automatically?

No, and that’s intentional. This is an alert-only system. It sends signals to Telegram — entry price, stop loss, target, position size — and I decide whether to execute manually in the Angel One app. Automated execution raises accuracy, licensing, and “what if it goes rogue” questions I’m not ready to own. Alert-only = full control.

Q3: How much does the entire setup cost?

₹0. Not “free trial” zero — actually zero. Angel One SmartAPI is free. Telegram bots are free. Oracle Cloud free-tier runs the server. Python libraries are open source. Claude Code has a free tier. The only cost is your time — and honestly, it took me less time to build this than to watch one complete “trading masterclass” on YouTube.

Q4: Which broker works for this? Only Angel One?

I used Angel One because their SmartAPI is well-documented, free, and Claude Code could work with it immediately. But the same approach works with Zerodha’s Kite Connect (₹2,000/month API fee), Upstox, or Fyers. Angel One was the ₹0 choice with the least friction. If you already have a different broker, ask Claude Code to adapt — it handles API differences without you knowing the details.

Q5: What strategy does the bot use? Can I change it?

Opening Range Breakout (ORB) — marks the high/low of the first 15 minutes of trading, then alerts on confirmed breakouts. It’s simple, rule-based, and heavily back-tested. And yes, you can absolutely change it. I built two other strategies before ORB (a swing screener and a fundamental scorer), and swapping strategies is just a conversation with Claude Code: “Replace ORB with VWAP crossover” and it rewrites the logic. You describe. It builds.

Q6: Is this legal? Can I get in trouble?

Building a personal alert bot = completely legal. You’re using publicly available APIs from your own brokerage account to send yourself notifications. You’re not offering trading advice, managing other people’s money, or front-running. It’s the same as setting a price alert on any trading app — just smarter and customizable. If you start selling signals or managing money, that’s a different regulatory conversation. Don’t do that without SEBI registration.

Q7: How reliable is it? Does it crash?

It runs on Oracle Cloud’s free tier, which has been surprisingly stable — uptime has been solid for the past few weeks. The script has error handling for API disconnections, data gaps, and market holidays (Claude Code added these automatically when I described edge cases). If something fails, it sends me a “bot error” Telegram message instead of silently dying. Is it enterprise-grade? No. Is it better than me staring at charts? Absolutely.

Q8: What’s next? Are you building more bots?

Already in progress. Next up is a swing trading bot — different from the intraday ORB system. This one combines fundamentals + technicals + news sentiment before sending alerts, runs end-of-day instead of real-time, and doesn’t even need the Angel One API (uses yfinance for data). Same Oracle server, same Telegram delivery. One marketer, two bots, zero lines of code written manually. My friend Karthik calls it “the laziest way to be productive.” I call it efficient.

The 7-Day AI Builder Challenge

You don’t need to build a trading bot. But you should build something with AI tools — just to prove to yourself that you can.

Day 1-2: Set up your environment. Install VS Code. Add Claude Code. Open the chat panel and say hello. Describe a simple problem you want solved — doesn’t have to be trading. Maybe it’s automating a spreadsheet, building a Telegram reminder bot, or scraping product prices.

Day 3-4: Build your first script. Let Claude Code write it. Read what it generates. You won’t understand everything — that’s fine. Focus on what the code does, not how it does it.

Day 5: Make it useful. Connect it to something real — Telegram, email, a spreadsheet. The moment your script does something in the real world, you’ll feel the shift.

Day 6: Break it and fix it. Ask Claude Code to add a feature. Watch it break. Ask it to fix it. This is where you learn the most — debugging is just asking better questions.

Day 7: Show someone. Demo it to a friend, a colleague, anyone. The look on their face when you say “I built this and I don’t code” is worth the entire week.

Comment “AI Builder Day 1!” below if you’re starting!

Conclusion — You Don’t Need to Be a Developer

Here’s what I want you to take away:

• AI tools like Claude Code have made the “can’t code” excuse obsolete. If you can describe what you want, you can build it.
• The entire setup cost ₹0 — Angel One API, Telegram bot, Oracle Cloud, Python libraries. All free.
• Alert-only is the smart scope — don’t let a bot trade your money automatically until you deeply understand every rule it follows.
• Start simple — ORB beat two fancier strategies because it was clear, rule-based, and fast.
• The skills transfer — once you build one thing with AI, you realize you can automate anything. That’s dangerous knowledge for a marketer to have.

Ramesh doesn’t laugh anymore, by the way. He asked me to build him one too.

If you’re a marketer, a writer, a designer — anyone who thinks “I’m not technical enough” — this post is my proof that you are. The tools are ready. The APIs are free. The only missing piece is your idea.

Comment “AI Builder Day 1!” below and tell me what you want to build!

Also read: If this got you excited about AI tools, check out how I built a ₹0 AI SEO Agent for Bloggers using the same approach. And if you’re wondering whether you have the right marketing mindset for projects like this, read our Top 5 Qualities of a Better Marketer. For the motivational side of starting ambitious projects, don’t miss Why You Start Everything But Finish Nothing.

About the Author

Jayaprakash — Marketing pro (8+ yrs), no-code AI builder who accidentally became dangerous with Python, VibezDude founder. Currently leading marketing at TechnoRUCS by day and building trading bots by night. Has automated everything from SEO workflows to stock alerts using AI, and will enthusiastically convince you that marketers are the best builders because we think in outcomes, not syntax.