Understanding finance as a beginner is a lot like googling "how to code" for the first time. There is a plethora of information available to you, but you just don't know where to start. We wanted to allow our users to pick up basic finance in an interactive way.

What problem it solves:

Many concepts in economics and finance are based on the expected. A future prediction that no one can make for certain. We thus created a machine learning algorithm which attempts to predict a stocks near future based on trends. This information allows us to walk the user through the different strategies to employ when the information is in front of them. This way, when they make their own predictions they are better prepared.

For ease of use, we created a front end, comprehensive chat bot which explains every step of the way. It also can easily provide a definition or clarify a concept.

How we built it:

We began by creating the underlying time series ML in R and the chatbot using dialogflow. React is used as the web frontend framework and sends requests to the Flask backend server and the Stdlib service which connects to Dialogflow SDK to handle the chatbot interactions. Requests are sent either straight to the Flask server when a ticker is searched for, or to the Stdlib service and when the response comes back the request is sent to the backend.

Backend is hosted with a Ngrok with a connection between Flask and R studios. React app is hosted on heroku.

Challenges I ran into

We found it very difficult to link all of our different projects together to work in one streamlined and clean way. With so many moving parts and a lot of ML and AI working together, it was very difficult to make sure that everything was interpenetrated right and worked together to produce our desired output.

Accomplishments that were proud of

We are extremely proud of our ability to create a machine learning model from scratch that works in a practical situation. Furthermore, the ability to implement something as complex as an artificial intelligence simultaneously while working on the ML is a great feat. We also think that it should be noted that we were able to take an extremely complicated subject of portfolio theory and make it simple and easy to understand.

What I learned

Every member of the team worked in an area that we have never worked on before. By design we all chose tasks that we were unfamiliar with but thought we could learn. In a lot of cases, we were working in areas that other members of our team had worked before allowing for peer support.

What's next for

Sell the algy to a hedge fund, buy a yatch, etc.

Share this project: