K-means clustering is one of the multivariable methods that aims to group objects based on the characteristics of the objects. The concept of K-Mean Clustering in this research is utilized to select stocks in a stock-portfolio formation. Then, the proportion or weight of each stock constructing the portfolio is specified by employing Mean Absolute Deviation (MAD). The analyzed data in this research is daily closed price stocks indexed by SRI-KEHATI from 13 August 2023 until 13 February 2024. There are two portfolios obtained from K-Means Clustering. The first portfolio is generated by BBTN, JSMR, SILO, and TLKM, while the second portfolio is formed by BBNI, BBRI, and BMRI. Using MAD, the proportion of each asset constructing a portfolio can be identified. For the first portfolio, the proportion of BBTN, JSNR, SILO, and TLKM subsequently are 30%, 30%, 10%, and 30%. Meanwhile, the proportion of BBNI, BBRI, and BMRI analyzed by MAD successively are 40%, 20%, and 40%. Moreover, the performance of the first and the second portfolios measured by Sharp Ratio are 0.06612 and 0.13668. It means that the second portfolio is better than the first portfolio.