Stonksquote

Inspiration

Tackle real world problems using the beauty of theory to design bleeding-edge optimization algorithms! So bad it turned out to be messy machine learning in the end...

What it does

Stonksquote receives streams of currency prices and trades from its users. Since blindly forwarding trade requests to the actual market is extremely expensive, we designed an algorithm that would avoid going to the market too often and find better alternatives instead. This would be a perfect use case for an online bank product where the bank need to take mininal risk while accessing the queries of its customers.

How we built it

We started from a template provided by Swissquote at https://github.com/Astat/sq-evolution. We simply plugged out logic in it and trained a few ML models separately using Python. Those scripts are called by our Java application to take optimal(?) decisions.

Challenges we ran into

This all ended in the way we were afraid of. We couldn't find an elegant way of solving the problem and we had to design machine learning approaches quickly in order to complete the project.

Accomplishments that we're proud of

Finally had it something working(?)

What we learned

With limited time, start to go from easy solutions to more advanced ones rather than the other way around. This makes the whole project way more enjoyable and less stressful!