Inspiration

How does a first-time freshman hacker compete with teams possessing several or even dozens of hackathons under their belt? Our team's response is buzz-words, all of them.

With a diverse team of majors ranging from Mechanical Engineering to International and Public Affairs, we each possessed distinct skillsets. Laughing about buzzwords transitioned our team and after some introspection, we realized behind all the fanfare and satire there were real hidden opportunities.

All of us previously tinkered around with Bitcoin and other Cryptocurrencies and as millennials we're well aware of the power of social media and big data. Our project comes from this intersection of our personal experiences and hidden opportunities within buzzwords.

What it does

Leverages social media analytics for forecasting cryptocurrency trends; provides an efficient, automated trading algorithm. Synthesizes moving averages analysis, recent twitter data, and a modified relative strength index to provide a robust strategy for algorithmic trading. Uses sentiment analysis to parse relevant tweets and extract consumer sentiment and predict its influence on highly volatile cryptocurrency markets.

How I built it

Frontend built with Adobe Dreamweaver in HTML5/CSS3, JavaScript, jQuery, and Bootstrap. Backend written in JavaScript and Python. Leveraged Twitter REST API and Python's tweepy API for scraping social media data, Quandl API and Alpha Vantage API for collecting historical financial data, TraderView API and Rickshaw.js for data visualization, nltk for natural language processing, and pandas and numpy for data analysis.

Challenges I ran into

Twitter API was very tricky to use -- rate limits made large-scale data analysis difficult and time-consuming. Error handling and optimizing our trading algorithm consumed the bulk of our time.

Accomplishments that I'm proud of

Simulated on historical cryptocurrency data and yields 562.60% return on investments for Bitcoin over past 20 days (compared to 486.8% ROI achieved by holding). Naive Bayes classifier trained on dataset of 30,000 tweets and achieved 24847.01% ROI between August 2016 and September 2016 (compared to 19.4% ROI achieved by holding).

What I learned

Learned large-scale data analysis and financial analytics. Familiarized with Twitter REST and Streaming APIs and realtime data collection.

What's next for Everest Futures

After finalizing risk management systems for Cryptocurrency trading, we would like to apply our method to other financial instruments which are similarly influenced by general sentiment. We also plan to further develop our trading algorithm and seek investment and mentorship.

Share this project:
×

Updates