Millionaire-Maker: Stock Price Predictor
Team Members Heading link
- Yiheng An
- Weijie Guan
- Jieyu Zhao
- Peng Zou
Advisor: Vladimir Goncharoff, PhD
Project Description Heading link
The goal of this senior project was to design software algorithms to help investors make buy/sell/hold decisions for a specific stock according to various factors: the historical performance of this and other related stocks, the stock’s price-earnings ratio, and current events that may affect the stock price. In order to guide investors’ decisions in real-time, we have designed a system having two algorithms: the first is a stock price predictor applying the Long Short-Term Memory (LSTM) neural network architecture, which is often used in the field of deep learning; it is trained on the past months’ prices of a group of stocks of interest to predict future trends. We retrain this predictor every day with new data after the market closes. The second algorithm takes prediction results from the first algorithm and also input from the user regarding current events and whether he/she considers an event to have positive or negative predicted effect on stock price, to produce buy/sell/hold suggestions. Our project is unique in that two predictors are tied together to give user-specific advice based on both user input and general market trends.