One of our team members is a Commerce student who is looking to work in finance in the future. After reading a PhD paper describing how a team of engineers built a similar algorithm and backtested their algorithm against data to show how the algorithm outperformed the market. She was curious to see if this is something that would work in real life and if this was something that could be hacked in 36 hours.
What it does
Social media is an excellent PR tool that has the power to either propel a business towards its best year or become the downfall of an otherwise successful company. While it is important to be informed about how companies are doing in a fluctuating stock market, users might spend valuable time scrolling through countless tweets, assessing how a company's digital presence could influence the stock market instead of using that time to make critical and educated decisions for their stock portfolio.
Enter MarketChirp: a web app designed to track tweets containing a specific stock symbol using a Watson API to categorize the tweet as either a positive or negative sentiment. Users can visit our site to gather data about how a stock's sentiment can change over a period of days and use that information to identify sentiment change and predict how these could potentially affect the stock market.
How we built it
MarketChirp was built using a Twitter API to feed tweets to a Watson Tone Analyzer, which would then attribute a sentiment to a tweet based on keywords mentioned in the tweet. The results of the sentiment associated with a particular stock symbol in the tweet were catalogued to the site using a d3.js and presented in the form of a visually intuitive line graph.
Challenges we ran into
A few key challenges that we ran into were:
- Learning Bootstrap in response to so many elements in web design
- Compromising a few web design features in order to streamline the creative process
- Optimizing a program that took five minutes to complete a task, which meant reworking the code
Accomplishments that we're proud of
The team considers the following to be their greatest accomplishments:
- Learning Bootstrap from scratch and being able to successfully integrate it into the project
- Getting to create a visually stunning design that evolved in response to the group's ideas
- Reworking the code for their program so that it ran in 15 seconds instead of 5 minutes
What we learned
- Working as a group where everyone is willing to learn and bounce ideas off each other has been productive and gratifying
What's next for MarketChirp
In the future, the group is looking to refine the algorithm MarketChirp uses to track a stock symbol's sentiment to catalogue changes from over a period of days to over the course of a few hours.