Inspiration

Knowing that equal-cap weighting in portfolios has historically outperformed cap-weighted portfolios in the realm of pure equity, we wanted to design a tool that allows individual investors and advisors better realize their options for equally weighted portfolios. Upon realizing the complexities in the required variables to perform the necessary calculations, we had no other option than to turn to machine learning. Due to the heavy influences of financial institutions in HackGT, we felt it appropriate to create a tool that both they and the judges would find value.

What it does

Trifecta displays BlackRock's 7 Core Alpha 100% Equity (Stock) Funds and allows for a user to input how much they want to invest. The program then determines how each fund is weighted in terms of small-cap, mid-cap, and large-cap, by assessing its holdings. Next, the Trifecta considers how much you are willing to put into the mutual funds you selected. TenserFlow then process this information to determine how many shares of each fund you should purchase with a goal of 33% small, mid, and large cap holdings.

How we built it

At first, we thought finding the fund distributions would be an easy task. However, we soon realized that finding these distributions was related to a concept in machine learning. As a result, we used TensorFlow.js to solve the multivariate regression problem. We chose to display the results using HTML, CSS, JS, and jQuery because it was most appropriate for integrating the results from TensorFlow.

Challenges we ran into

The computational algorithm was the largest challenge we faced. Another challenge that proved difficult was the lack of desired information available in the hackathon version of the BlackRock API, Alladin. The market cap weighting is unavailable by native call to the platform, causing us to draw values through other means.

Accomplishments that we're proud of

Devising a method of determining how many shares of each fund an investor should purchase is absolutely the bread and butter of our project. If you're reading this, we challenge you to devise another way of doing so! Looking at the variables on paper was daunting - but we prevailed.

What we learned

For starters, the team now has a better understanding of mutual funds and general investing strategy. As well, when we went into the project, we had no idea TensorFlow had a JavaScript library. After implementing our own calculations, we now have a better understanding of solving regression problems. Additionally, this project served to reinforce our understanding of version control as we used Git and GitLab to share and contribute to our project.

What's next for Trifecta?

This is only the beginning! The future holds a dynamic web crawler that will scrape the values of what we hope to be all securities. Next, we will be integrating an analytics section to the site. To accomplish this, the use of BlackRock's Aladdin API is going to be a necessity. The plan is manipulate and display relevant charts for the funds our users have selected and compare their performance side by side as well as with a cumulative recap. Of course there is more, but we don't want to give away our trade secrets.

Share this project:
×

Updates