Robo-Advisors, does complexity mean performance ?

Abstract

This master thesis aims to study the existence of a link between the sophistication of roboadvisors and the returns they generate. Our analysis covers the period from may 2015 to April 2021. The data set we exploit comes from a major German comparison website for online brokers and digital asset managers, brokervergleich.de. Since returns are not published in the same way and some robo-advisors were created at dates later than our first data, we divided our sample into 5 sub-samples. In addition to monthly returns, through several risk adjusted performance measures such as the Sharpe ratio, Omega ratio, Calmar ratio, or Jensen’s alpha we measured the performance of robo-advisors. The variables we have chosen to reflect complexity of robo-advisors are: the number of offers, the human involvement in the portfolio management process, the management style, the frequency of rebalancing and the investment strategy. We have tried to account for the degree of influence that these variables have on the performance measures calculated on the basis of the monthly returns, first in a discrete way, then in a compiled way via a score constructed on the basis of these complexity proxies variables. Our findings show that there is indeed a link between complexity and robo-advisors’ fees. But the low statistical significance and small sample size do not allow us to assert any link between complexity and performance measures