Frédéric VrinsZone, CorentinCorentinZone2025-05-142025-05-142025-05-142020https://hdl.handle.net/2078.2/17366Option pricing is still a sensible question today. A lot of alternative models have been developed since the first apparition of the Black and Scholes option pricing model. However, no consensus emerges on the most adequate model to employ which can lead to arbitrages inside the market. This thesis presents an option classification technique between two categories, over-priced options and under-priced options, through machine learning models. We discriminate the options on the basis of their optimal volatility computed with a Delta-neutral payoff strategy following the Black and Scholes model. Based on this classification, we construct an investment strategy, by selling the options, that allows us to obtain a higher profit than selling these options randomly. We apply the methodology on European option quotes on 15 stocks over a time span of 90 working days. We analyse the predicted accuracies obtained with machine learning models on these options and try to employ them during a real investment situation.financeoptionmachine learningBlack Scholestext::thesis::master thesisthesis:25776