Portfolio optimization is a practical application problem. The task of this problem is to allocate capital to a set of assets and its goal is to maximize investment returns while minimizing the probability of loss (risk). This makes portfolio optimization a multi-objective optimization problem. It is also a noisy problem, but noise is ignored in most research. In classical portfolio optimization, an efficient optimal portfolio is created using past stock dividends. Inevitably, the expected return from the portfolio is subject to uncertainty and noise. Naturally, we have no knowledge of future stock dividends and invest only based on our estimates and expectations of future stock performance, which itself contains a high level of noise. In this research, the Markovitch mean-variance model is used to investigate the effects of noise on the return from the optimal stock portfolio. In this article, we will show that investment decisions can be significantly wrong if noise is ignored. Although the results in this article are negative, the results have significant benefits for investors. When dividends are subject to noise and turbulence, investors should be very cautious about their portfolio selection and investment strategy.