Sentiment Analysis for Bitcoin Market Trends with NLP

This project aims to develop a modular trading strategy platform that enables non-programmer traders to easily automate their strategies. By utilizing a visual UI/UX interface, investors can create and backtest trading strategies without the need for coding. Once satisfied with the results, users can seamlessly connect to exchange APIs to implement their strategies in real-time, reducing the emotional impact of market fluctuations and minimizing monitoring time.
Github Repo Link: github.com/CaslowChien/Crypto-autotrading-platform

My Contribution

Developed a sentiment analysis model using transformer model from HuggingFace, implemented a streamlined pipeline for data scraping, cleaning, and tokenization with Python package to process real-time Twitter data.
Created a custom sentiment index using NLP model outputs to quantify Bitcoin market sentiment trends.

Final Presentation Video

Result Demonstration

Final Presentation Slides

Git

Working with a team of front-end and back-end engineer, our team had a strict version control system and pre-checks. As a result, I gained more experience with Git.

Machine Learning

Through this project, I implemented a sentiment analysis model for Twitter and fine-tuned it.

Data Scrapping

Scraped data from Twitter using specified tags and keywords.