Stock Market Prediction

Inspiration

The inspiration for this project came as I was watching GME go up and down over the last month or so. I wanted to be able to predict these kind of spikes beforehand, so I decided to build a model that could help me do so.

What it does

This project generates in-depth analysis of technical indicators of stocks (when significant volume/price shifts happen, analyst ratings, earnings surprises etc.) and also analyzes the buzz around a stock using sentiment analysis of Twitter tweets. It also attempts to predict the next day price of a stock using the closing prices data.

How we built it

I built it using Python and its numerous libraries for scraping, sentiment analysis and machine learning. In addition, I used API's such as Tweepy and yfinance to get extensive Twitter and stock data. An integration of data generation and various libraries for functionality ended up making this project, as it was mostly developed in Python.

Challenges we ran into

Once I got the idea, it was very hard to exactly develop every component of the project. This project is multi faceted in many ways, and since it was just me working on the project, I had to quickly work on each one. Thus, I was under a time crunch and couldn't build out every component exactly as how I wanted.

Accomplishments that we're proud of

Even though I couldn't build everything exactly as I wanted, I was still very proud that I could build out this whole project by myself. It's my first time doing a hackathon of this magnitude, so even submitting on time with a semi working project is something I'm extremely proud of.

What we learned

I learned so many things, ranging from a small to a larger magnitude. As a whole, I learned about many machine learning algorithms that I didn't know otherwise, and I also learned a lot of useful back end skills (scraping, api integration, api keys storage, etc.). I also learned the smaller details about how to work on a project when under a time crunch, and how to tackle my issues one at a time without putting excessive stress on myself.

What's next for Stock Market Prediction

I hope to add a GUI to make it more presentable to users first of all. I also hope to increase the accuracy of the machine learning model and find a way to predict what time range will return the most accurate prediction (maybe using another machine learning model for this).

Built With

• beautiful-soup
• keras
• numpy
• pandas
• python
• tensorflow
• twitter