Pembentukan Portofolio Saham IDX80 Berbasis Minimum Spanning Tree dan Pewarnaan Graf Bipartit dengan Variasi Pembobotan

Abstract

Pembentukan portofolio saham merupakan keputusan investasi yang kompleks karena investor berupaya memperoleh return yang tinggi dengan risiko yang terkendali. Penelitian ini bertujuan membentuk portofolio saham dari Indeks IDX80 menggunakan metode Minimum Spanning Tree (MST) dan pewarnaan graf bipartit yang dilakukan melalui algoritma Breadth First Search (BFS). Melalui proses iteratif menggunakan MST dan pewarnaan graf bipartit, diperoleh tujuh portofolio dasar. Setiap portofolio selanjutnya diberi bobot investasi menggunakan tiga metode pembobotan, yaitu equal weight, maximum Sharpe ratio, dan maximum utility, sehingga dihasilkan 21 portofolio. Portofolio dievaluasi pada data training menggunakan expected return, risiko, Sharpe ratio, dan utilitas. Selanjutnya, tiga portofolio dengan kinerja terbaik dievaluasi pada data testing menggunakan geometric return mingguan. Hasil penelitian menunjukkan bahwa Portofolio 6 dengan metode pembobotan maximum utility menghasilkan geometric return mingguan tertinggi. Temuan ini menunjukkan bahwa kombinasi metode MST, pewarnaan graf bipartit, dan metode pembobotan portofolio merupakan alternatif yang efektif dalam membentuk portofolio saham yang terdiversifikasi di pasar modal Indonesia.

Stock portfolio construction is a complex investment decision that aims to achieve high returns while maintaining controlled risk. This study aims to construct stock portfolios from the IDX80 index using the Minimum Spanning Tree (MST) method and bipartite graph coloring performed through the Breadth First Search (BFS) algorithm. Through an iterative process involving MST and bipartite graph coloring, seven initial portfolios were generated. Each initial portfolio was subsequently assigned investment weights using three weighting methods, namely equal weight, maximum Sharpe ratio, and maximum utility, resulting in a total of 21 portfolios. The portfolios were evaluated on the training data using expected return, risk, the Sharpe ratio, and utility. The three best performing portfolios were then evaluated on the testing data using weekly geometric returns. The results show that Portfolio 6 constructed using the maximum utility weighting method achieved the highest weekly geometric return. These findings demonstrate that the combination of MST, bipartite graph coloring, and portfolio weighting methods provides an effective alternative approach for constructing diversified stock portfolios in the Indonesian capital market.