Time series forecasting is an important area of financial forecasting. With advances in machine learning and AI, the speed of information is driving market efficiency. A robust financial model assumes that an efficient market exists, in which all currently available information is factored into prices, and future prices are determined by uncertainty. Today's portfolio theory is based on the Markowitz framework, which focuses on market uncertainty analysis rather than price prediction. The Markowitz framework makes strong assumptions about the probability distribution of future returns. To overcome this drawback, we propose using a generative adversarial network method, a quantum computer-based QuGAN, for portfolio optimization. Generative models in QuGAN learn the probability distribution of asset prices to match the probability distribution of the real market. After training the model, we construct an optimal portfolio that minimizes risk and maximizes profit observed under various simulations. This study compares the portfolio constructed using the QuGAN methodology with the classic Markowitz portfolio.