Stockvisor

Just like shopping on Amazon, we've built an AI using ML that is able to predict and recommend securities that you may be interested in based on Marquee data and qualities of your purchases.

Comment

Inspiration

Online stores has always been able to predict what the shopper wants, and today, it's very good at it. However, buying securities and investing is slow, requires much research, and sometimes, is very time consuming and stressful to keep up with. Usually, people don't blindly purchase securities, they consider different qualities of it, essentially do their research. We believed that it is possible to use Marquee's database as well as Machine Learning to recommend other stocks that the user may be interested in.

What it does

We made a barebones implementation of an AI software that dynamically trains its machine learning model with every query by the investor. It uses this model to recommend other stocks that the investor may be interested in using an algorithm that scores based on how similar the other stock is to the one the investor bought as well as the independent financial performance metrics provided by Marquee.

How I built it

We used Jupyter notebook to request the data, explore the data, then clean the data for training. We then moved the test code to a python3 script to be used with flask.

Challenges I ran into

We had trouble first understanding the documentation of Marquee, mainly because it was our first time working on a finance related project. However, we were able to learn a lot through this opportunity and were able to finish the project.

Accomplishments that I'm proud of

This was the first time we tackled a project where we had to find the data, prepare the data, and also train models using the data we prepared. We are very proud to have been able to do this in one day too!

What I learned

How to search for open source data sets.

What's next for Stockvisor

We want to expand from just the small list of stocks we trained our model on. We would also like to integrate more functionality beyond just telling the AI what stock you want, but also maybe allowing the user to ask questions, request data visualizations, and much more!

Built With

Share this project:

Updates