Investment involves allocating funds to gain future profits, one way being the purchase of stocks representing company ownership. Investors seek high returns with low risk, but stock price fluctuations introduce risk. Diversification through a stock portfolio helps minimize this risk. The Mean Variance method by Markowitz in 1952 optimizes portfolios based on risk and return, but it assumes data must be normally distributed, often misaligned with financial data. This study adopts the Mean-Semivariance optimization method, which does not require normality assumptions and is more suitable for non-normal data. The study uses 6 stocks from the IDX30 index, to form 2 portfolios with 3 stocks each. The results show an optimal portfolio composed of BMRI stocks with a weight of 48,69%, PGEO stocks with a weight of 17,01%, and INKP stocks with a weight of 34,31%. This portfolio has a Sharpe index of 0,03985, indicating better risk optimization using the Mean-Semivariance method.