FinBay

A credit card advisor for beginners to develop healthy financial habits.

Comment

Inspiration

A good financial posture is critical in today’s daily life, and that begins with credit cards. How does one choose a credit card for themselves, if all the information they receive is vendor-biased? If it is recommended that you maintain many credit cards for credit building, using which one, and where, where will benefit the user the most? How does a university student with no credit history, make a decision about what card to use? As the only application that offers our functionalities, we present a seamless application that is integrated securely using Plaid, delivering educational value and optimal financial recommendations, completely free of any bank bias.

What it does

With FinBay, here are some of our key features:

  • Forecast end-of-week, end-of-month, end-of-year expenditures based on frequent categories (Groceries, Restaurants, Travel) and overall trends
  • Recommend the most optimal credit card for the current purchase based on expected expenses, maximizing all rewards and incentives
  • Allow the user to securely authenticate themselves to a variety of leading banks, to manage their financial portfolio
  • Aggregate and display the interest rates/other benefits tied to each credit card in an interactive dashboard
  • Our recommendation and forecasting features are agnostic to financial institutions

How we built it

Mobile App

We developed the mobile App in Flutter, written in Dart, with an emphasis on an interactive and easy-to-use design. The mobile app is connected to our backend applications through firebase and cloud functions.

Backend

Our backend is developed with TypeScript and centered around the Plaid API to securely retrieve and store card, transaction, merchant, and other dashboard information.

Algorithms

We used the ARIMA model with Python, to forecast the expected value of future transactions based on a history of previous transactions aggregated at a category level. We then use this information to determine the benefits from each card, standardizing all benefits to a monetary level. Based on the expected benefits, we recommend a card to the user.

Google Cloud

We used firestore as our no-sql database implementation to store all retrieved transactions, and other aggregated information for the dashboard with firebase for authentication. With an asynchronous server-less pipeline, all our backend functions and ]algorithms are hosted on cloud functions, triggered incrementally with event triggers. For example: the cloud function which holds our forecasting algorithm is directly triggered when an update or a new user with transactions has been created.

Languages: Dart, TypeScript, Python

Challenges we ran into

  • Integration with firestore, such as writing to firestore from a cloud function
  • Setting up the asynchronous triggers, such as triggering the ML algorithm when a transaction is added or created

What's next for FinBay

Next, we would like to further work on the app to expand our card and stores selection, and improve the recommendation algorithm before making this available to the public. Also we want to explore further ways to import transactional data from other vendors (some people may not be comfortable solely relying on Plaid) and supporting more financial products to better educate everyone.

Research and Statistics

Share this project:

Updates