QuantCrowd was a two-month experiment to prove AI controlled, crowd-sourced intelligence could outperform the top managed funds.
Core Thesis: Thousands of investors with a broad array of backgrounds and skillsets have a collective knowledge to predict macro-economic trends in financial markets.
Prototype & Solution: Over the course of 8 weeks we built a simulated quant fund which used a voting system to determine investment decisions. As users voted, the fund weighted power based on the success rates of individual users. Over the course of the eight week closed alpha, the theoretical fund (perceived as real by the users) of $100K returned 18% returns with 5 year projections averaging 15% returns. The fund also outperformed the equivalent benchmark by ~5%.
- Ruby on Rails MVC application for funds rebalancing (controlled using Active Admin)
- Live calculations with JS scripting within Google Sheets
- Native Swift iOS app for customer facing product
- Integration with IEX Market Data API’s