This post is part of series of articles written by Betterment’s 2013 summer interns.
When I first applied to work at Betterment as part of their software engineer internship program, I knew close to nothing about what my future career goals were. And while I’m still figuring them out (hey, I’m only 18!) the internship was a great way to apply what I have been studying at Harvard, where I am majoring in statistics and economics.
One of the coolest projects I worked on last summer involved writing a Python script that generated financial models of Betterment’s portfolio performance for varying risk allocations, time horizons, and client investment levels for the company’s website. (If you don’t know what Python is, it’s not a big snake. Rather, it’s a object-oriented programming language comparable to Java.)
My task was to develop a script to automate the data retrieval behind Betterment’s past performance widget on a monthly basis. The widget uses dividends-adjusted historical market data to calculate the portfolio’s performance over time.
Most importantly, this widget is featured prominently on the company’s website and the information is useful to customers when they make an investment decision.
Learning a new language
Coding in Python wasn’t something I had much experience with, but when Product Manager Alex Benke asked me to take on a project involving scripting for the investing team, I jumped at the opportunity to learn it.
On the back end (where the heavy lifting is done), this data is stored as two static JSON objects—(JSON is a kind of database protocol)—generated by weaving together each ticker’s returns as a single “mega-stock” or “mega-bond” time series weighted by its share in its basket. It’s sort of like making two braids out of different yarns with varying thickness.
But making these “braids” was not as easy as it seemed at first. Querying, storing, and manipulating the ticker data presented many challenges. In particular: Where and how would this data be stored? How flexible is this tool for portfolio changes? How easy would it be ensure the accuracy of our data?
After querying the Yahoo Finance API for prices, I also needed to calculate monthly returns, track portfolio growth over time, and rebalance the portfolio at every end of quarter.
This task turned out to be the perfect opportunity to apply what I had learned in another project I had worked on using Google Spreadsheets as a cloud-based database system. By manipulating individual worksheets, rows and cells, I could isolate each step of the process and visualize the results while the script ran.
The result? A simple, sleek script that could be set to run automatically on our servers and injected the updated data seamlessly into the backend of the past performance widget.
Of course, coding was not the only thing I learned this summer. I learned about working in a tech office (you’d better like ping pong and dogs!) and living in New York City. But after a great summer, I am closer to knowing that I enjoy coding and that it will likely be part of my career.