Doesn't it make more sense to measure and minimize the variance of the underlying cash flows of the companies one is investing in, rather than the prices?

Price variance is a noisy statistic not based on any underlying data about a company, especially if we believe that stock prices are truly random.