This paper discusses an extension of Markowitz's portfolio construction method that addresses practical issues and handles uncertainty in return statistics. Markowitz's original method, which trades off expected return and risk (defined as portfolio return variance), has become the dominant quantitative method for portfolio construction despite criticisms, such as sensitivity to forecast errors. The extension described here is a convex optimization problem that can be solved reliably and efficiently. The paper outlines the original Markowitz idea, which involves optimizing portfolio weights based on expected returns and covariance matrix. It also addresses alleged deficiencies of Markowitz's method, such as sensitivity to data errors, symmetric risk definition, and neglect of higher moments like skewness and kurtosis. The paper argues that modern optimization techniques effectively handle these issues without altering Markowitz's vision. The extension of Markowitz's method incorporates practical constraints and objectives, such as transaction costs and leverage limits, and uses robust optimization and regularization to mitigate sensitivity to input data. It also addresses the issue of sensitivity to estimation errors in return forecasts. The paper describes a convex optimization formulation that includes additional constraints and objectives, such as holding and trading constraints, and provides numerical experiments to demonstrate its effectiveness. It also discusses the use of convex optimization in portfolio construction, including the use of domain-specific languages and solvers. The paper concludes that Markowitz's method remains relevant and useful, with the proposed extension providing a practical and effective approach to portfolio construction. It emphasizes the importance of convex optimization in handling the complexity of modern portfolio construction and the need for robust and efficient solvers.This paper discusses an extension of Markowitz's portfolio construction method that addresses practical issues and handles uncertainty in return statistics. Markowitz's original method, which trades off expected return and risk (defined as portfolio return variance), has become the dominant quantitative method for portfolio construction despite criticisms, such as sensitivity to forecast errors. The extension described here is a convex optimization problem that can be solved reliably and efficiently. The paper outlines the original Markowitz idea, which involves optimizing portfolio weights based on expected returns and covariance matrix. It also addresses alleged deficiencies of Markowitz's method, such as sensitivity to data errors, symmetric risk definition, and neglect of higher moments like skewness and kurtosis. The paper argues that modern optimization techniques effectively handle these issues without altering Markowitz's vision. The extension of Markowitz's method incorporates practical constraints and objectives, such as transaction costs and leverage limits, and uses robust optimization and regularization to mitigate sensitivity to input data. It also addresses the issue of sensitivity to estimation errors in return forecasts. The paper describes a convex optimization formulation that includes additional constraints and objectives, such as holding and trading constraints, and provides numerical experiments to demonstrate its effectiveness. It also discusses the use of convex optimization in portfolio construction, including the use of domain-specific languages and solvers. The paper concludes that Markowitz's method remains relevant and useful, with the proposed extension providing a practical and effective approach to portfolio construction. It emphasizes the importance of convex optimization in handling the complexity of modern portfolio construction and the need for robust and efficient solvers.