1: \begin{abstract}

2: The first moment and second central moments of the portfolio return,

3: a.k.a. mean and variance, have been widely employed to assess the

4: expected profit and risk of the portfolio. Investors pursue higher

5: mean and lower variance when designing the portfolios. The two moments

6: can well describe the distribution of the portfolio return when it

7: follows the Gaussian distribution. However, the real world distribution

8: of assets return is usually asymmetric and heavy-tailed, which is

9: far from being a Gaussian distribution. The asymmetry and the heavy-tailedness

10: are characterized by the third and fourth central moments, i.e., skewness

11: and kurtosis, respectively. Higher skewness and lower kurtosis are

12: preferred to reduce the probability of extreme losses. However, incorporating

13: high-order moments in the portfolio design is very difficult due to

14: their non-convexity and rapidly increasing computational cost with

15: the dimension. In this paper, we propose a very efficient and convergence-provable

16: algorithm framework based on the successive convex approximation (SCA)

17: algorithm to solve high-order portfolios. The efficiency of the proposed

18: algorithm framework is demonstrated by the numerical experiments.

19: \end{abstract}

20: