Overview
In this post, I’ll walk you through a full-stack data science side project where I built a crypto trading signal bot. It uses classic candlestick logic on historical OHLCV data from KuCoin and sends trading alerts directly to my phone via Telegram Messenger.
Why does this matter? For any aspiring or professional data scientist, being able to deploy small, end-to-end tools like this isn't just satisfying—it’s a career booster. It shows you're more than a model builder; you're a system thinker.
Motivation
I wanted to create a lightweight, real-time signal detection bot that didn’t rely on advanced indicators or black-box models. Instead, I used simple candlestick rules—like reversals after two down candles followed by an up candle—to detect potential “Long” or “Short” entries.
Tech Stack
- Python — scripting language of choice
- ccxt — for pulling real-time OHLCV data from KuCoin
- pandas — for processing time-series data
- asyncio — to run the reporting asynchronously
- Telegram Bot API — to send alerts and CSV reports to my phone
- GitHub — to document and share the project
How It Works
1. Data Collection: Fetches the last 5 days of OHLCV candles (daily + 4-hour).
2. Signal Logic: Checks for patterns like:
- 2 red candles + 1 green candle → Long
- 2 green candles + 1 red candle → Short
3. Reporting: If a signal exists, a CSV file is created and automatically sent to my Telegram account.
4. Scheduling: It’s designed to run periodically (e.g., daily at midnight) via cron or scheduled Python task.
Result Snapshot
| Symbol | Signal_Daily | Signal_4H |
|--------------|--------------|-----------|
| BTC/USDT | Long | No Signal |
| ETH/USDT | Short | Short |
| ARB/USDT | No Signal | Long |
What's Next?
- Add indicators like RSI or MACD
- Include backtesting logic
- Support Telegram or email alerts
- Deploy it to a cloud scheduler
Conclusion
This project is proof that you don’t need deep learning to do something useful. Sometimes, a clean rule-based system combined with smart deployment is all it takes to stand out.