I had the chance to participate to a hackathon with my school, ENSAE ParisTech, in collaboration with Dauphine (PSL University). The competition was about predicting if the volatility was over-estimated by the market or not, so that we can adjust our strategy.
As all the data was very nicely prepared for the competition we didn't have any cleaning to do, and we had time to perform a deep analysis of the data : We added some features like the Super Sector, the Sharpe, and normalized some of the features by sector or region.
For the model, we chose a Balanced Random Forest, after having tried a XGBoost, which overfitted a lot.
With my teammates from my master (MSDS, Master Spécialisé en Data Science), we ended-up winning the competition, and were awarded a 2000€ prize.