Aguacate

The Great Bart

We were eager to find out some creative strategies to trade and having the opportunity to do sentiment analysis together with it just felt like the right thing! We started with basically no clue of how we could approach this, as none of us has any experiences with finance or trading whatsoever. We started with the sentiment analysis, trying out and fitting approximately 1983397 different models. None of those were giving us the speed we were looking for. However later we realised we didn't need the magnitude of speed we were looking for. To train the model to make better predictions, we accumulated the news and their effect on each stock, then translated from English to many languages and back to change the words a bit and train the model with that. That turned out really well. On Saturday, we were the whole day trying to do arbitrage, which failed miserably. Therefore we looked for another totally different solution to make money. Using large "walls" as fail safe (we'll explain that later). At the end it turned out pretty well. It was really fun!!