Hack Cambridge

(Oh yes, I forgot to mention, I’m a Cambridge Student now, studying Maths at Trinity College UwU)

I was in the last two days very privileged to participate in Cambridge University’s flagship hackathon event, Hack Cambridge, having made a last minute entry. It was my great pleasure to participate in a team alongside Trinity College computer science students Bence, Jakob and from Kings college, Kornel. I am also thankful for the opportunity to meet with many sponsors of the event, such as Optiver, Jane Street and Blackrock.

Hacking started at 12pm, and continued until 12pm the following day. The first few hours was spent attending workshops and brainstorming ideas. Out of several ideas, ranging from Balanced News Recommendation Systems to Fintech, we combined some ideas across the team and settled on an toolkit for pricing Carbon Allowances. We began setting up our Machine Learning tools around 7pm, breaked for lunch and worked till 12am when I went back to my dorm to get some sleep.

Work resumed around 6am, and with the help of some coffee we were able to pull through and finish all features with around 10 minutes to spare. Many Kudos to Jakob’s superb frontend skills, Bence’s good animation and Kornel’s side auto-trade project.

Our team as a whole was very much pleased to have produced a very good piece of software in 24 hours of hacking. We, to our great surprise, ended up placing in the top 6 projects overall, and received the Wolfram Prize for our efforts.

The event was very tiring, but I enjoyed it very much and the rapport and teamwork between our team was amazing. This is not to mention the heaps of free swag from sponsors; the prize won was only the cherry on the cake.

End Product

Our end product was called Green Analytics, a toolkit to help ordinary investors trade an currently illiquid asset “European Carbon Allowances” containing 3 machine learning models (sentiment analysis, time series price forecasting and electricity consumption prediction) to help price the asset. Our hopes was that it would drive up demand in secondary markets for the asset and thus increase its price, and thereby make it more costly for companies to pollute.

In all we captured the essence of the idea well, and although our models fall well short of production quality, nevertheless achieve nontrivial performance

1) The github repo is here

2) A ppt describing our project here

3) An image of the interface