The effect of investors sentiment on the robo-advisor's financial model

The goal of this master thesis is to measure the effect of investors sentiment on the robo-advisor’s financial model. Consequently, two research hypotheses have been formulated. The first one aims at confirming or disproving that the Black-Litterman model without investors views offers a better risk-return trade-off to an investor than the Modern Portfolio Theory. To answer the first hypothesis, the Implied Equilibrium Excess return vector has been built in order to build the efficient frontier of the Black-Litterman model without investors views. In the second hypothesis, we claim that the Black-Litterman model with investors views provides a better risk-return trade-off than the Modern Portfolio Theory. To this end, a sentiment analysis has been used to provide the Black-Litterman model with investors views. Since the sentiment time series showed some nonlinear dependencies with ETFs stock returns. We have used the sentiments as inputs of a Long Short Term Memory neural network to predict future returns. The thesis highlights that the first research hypothesis is partially validated as for some level of risk an investor is better off with the Modern Portfolio Theory. However, we confirm that the Black-Litterman model with investors views offers a better risk-return trade-off to any investor.