QuantCrowd

QuantCrowd was a financial / social experiment that I ran with my former co-founder at Whaley. Through exchange data generated from a closed alpha test with Whaley, we found measurable correlations between profitable investment decisions and crowd-sourced sentiment towards individual equities.

Our thesis was, if we provided a fund where investors could vote up and down what should be in the fund at any given time, the crowd intelligence will be more powerful than the individual. We coupled this with a layer of AI which rebalanced the portfolio and allocated more / less power to users based on their past performance. We even added a third layer of intelligence that allocated more / less power over the fund based on the individual users ability to understand markets at a particular time of year, proficiency in a niche sector and understanding of individual companies.

We started with creating a basic landing page that was stripped directly from the original Whaley bootstrap template and used the existing PHP instance on a local LAMP stack.

We needed to convince 100 people that the fund was genuinely real but only gave ourselves 6-8 weeks to prove or disprove the thesis. We achieved this through the landing page which I designed and built in the space of a week and stating that we had seeded the fund with $100k of our personal savings. We made it apparent that the users who contributed the most to the fund would receive first-access to invest in 2019.

We reached out through our networks and anonymously through several Reddit threads that the fund was opening its alpha in a matter of weeks. This gave us 5 weeks to built a native iOS app with an integration with an existing Rails Active Admin instance (forked from Whaley’s fund management tool).

The fund mathematics was calculated locally using Javascript functions in Google Sheets. It wasn’t automated but it worked well with the IEX market data API’s and allowed for ease of import / export functions with the existing Rails Admin controls.

Within 5 weeks, we had designed, built and integrated IEX market data and introduced 105 votable assets in the fund. We had 500+ people who opted in for the alpha test.

The alpha was run for 5 weeks. Fortunately, we had several weeks of volatile market conditions to test the theory (thanks to the US/China trade war).

From the 500 users, we had about 10% actively returning every day to vote on the fund. The rebalancing algorithm worked well aside from its down-grading mathematics in which users who made slightly poor decisions were heavily penalised.

After 5 weeks, the fund had returned over 17% and outperformed the relevant benchmark by 5% for the period. We didn’t pursue to launch the fund at the time but are considering taking it further in the future.

Want to chat more or see other things I’ve built?

Check out my selected works or get in touch.