Stock Savant

• 

  stock savant

Inspiration

As soon as the hackathon started, we were thinking of a way to incorporate our skills in python and data science to create an awesome financial application that would be fun in addition to serving a practical use for anyone with an internet connection. As a result, we decided to create Stock Savant, which is a culmination of our skills (new and old!) we have learned this far!

What it does

Our application allows the user to input a stock ticker, and in return they receive the predicted price for the next month. This can be used to tell if a stock would be a good buy based on different indicators that we believe have the potential to alter the stock's price in the future. We developed an AI application that combines open source AI model analysis of news sentiment and social media trends, and openly available market and economic data to predict company financial stock performance.

How we built it

For the backend, we used the Python libraries yFinance, Pytorch, Flask, and Pandas. We also used NewsAPI, Reddit API, and an open source BERT AI model that sifts through bodies of text and detects whether they have positive or negative sentiments on a scale from 1 to 5 (negative to positive). For the frontend, we used vanilla HTML, CSS, and JavaScript.

Challenges we ran into

It took us a while to land on a solid stock market library to be able to access past stock performance, in addition to wasting countless hours of trying to run an open source large language model that didn't end up panning out because of our time constraint. In addition, we didn't have any experience with a lot of the backend processes, so it was definitely a learning curve to say the least!

Accomplishments that we're proud of

Creating our first backend system for a website and creating our first project that utilizes machine learning models. Through our hours of development, there were moments where we felt like we hit a wall and couldn't progress, yet there were also moments where we finally figured things out and it felt absolutely amazing. In addition, making algorithms that predict the future is a heavy task, but we believe it payed off, as the final product is something we are proud of!

What we learned

We learned tons about machine learning models, flask, web development, and quantitative finance. We learned that a big part of hackathons is being able to work well with your teammates, because we all know once it gets late enough, things can start to get out of hand!

What's next for Stock Savant

We are hoping to deploy the app online so anyone can use it for any device. This will hopefully be done by the end of the hackathon. We also want to add more datapoints for a more accurate prediction of the stock.

We wanted to thank Goldman Sachs and Fidelity for the opportunity to create such a cool project off of their prompts, and the HackUTD team for spending long hours putting this amazing event together!

Built With

• ai
• bert
• css
• flask
• html
• javascript
• machine-learning
• newsapi
• pandas
• python
• pytorch
• redditapi
• yfinance