Goldman Sachs ML Sentiment Analysis

Sentiment Analysis

Applied NewsAPI for web scraping to compute relevant stock sentiment primarily through Bloomberg, CNN, Business Insider, Financial Times, and Reuters. Used Twitter API for real-time posts relating to the stock embedded into the app. This is a trending topic in Event-Driven Investing, used by many quant firms and hedge funds.

Machine Learning

Used a Long-Short-Term-Memory Recurrent Neural Network (LSTM-RNN) Model that could track temporal changes of stock price, growth score, financial return score, multiple score, and integrated score with a 50-day moving average, which is a commonly used technical indicator for short-term price movement. Trained the neural network on Google Cloud Services and NVIDIA Tesla P100 GPU.

Web Application

Developed a web application using Bootstrap that uses a custom search bar to query historical stock trade data from a MySQL database. The plan was to utilize the data to provide trends and graphs.

Thank you @lilianweng for the TensorFlow base model

Built With

• beautiful-soup
• bootstrap
• css3
• django
• html5
• iexfinance
• marquee
• newsapi
• pandas
• python
• tensorflow
• twitter
• vadersentiment