A time-series bootstrapping simulation method to distinguish sell-side analysts’ skill from luck

Data mining is quite common in econometric modeling when a given dataset is applied multiple times for the purpose of inference; it in turn could bias inference. Given the existence of data mining, it is likely that any reported investment performance is simply due to random chance (luck). This study develops a time-series bootstrapping simulation method to distinguish skill from luck in the investment process. Empirically, we find little evidence showing that investment strategies based on UK analyst recommendation revisions can generate statistically significant abnormal returns. Our rolling window based bootstrapping simulations confirm that the reported insignificant portfolio performance is due to sell-side analysts’ lack of skill in making valuable stock recommendations, rather than their bad luck, irrespective of whether they work for more prestigious brokerage houses or not.