In order to test this, you have to account for some alternative. In other words, had I randomly bought that same security over the 3 year period, what were the returns and are your results statistically significantly different then the null comparison. Otherwise the results are meaningless.