Pengelompokan Emiten Berdasarkan Fluktuasi Harga Saham Syariah Menggunakan Algoritma Agglomerative Hierarchical Clustering

Abstract

Kebijakan pemerintah untuk memutus rantai penularan Covid-19 memberikan dampak pada sektor ekonomi di Indonesia. Banyak pekerja yang harus mencari sumber pendapatan lain karena kehilangan pekerjaannya di masa pandemi. Salah satu usaha yang dilakukannya yaitu berinvestasi saham. Indonesia sebagai negara dengan mayoritas penduduk muslim membuat pasar modal syariah khususnya saham syariah semakin diminati. Namun demikian, saham syariah yang memiliki tingkat pengembalian tinggi juga memiliki risiko yang tinggi. Salah satu faktor yang membuat risiko saham tinggi adalah harga saham yang bersifat fluktuatif dan stokastik. Oleh karena itu, perlu adanya sebuah analisis tentang penggerombolan emiten berdasarkan fluktuasi harga saham syariah. Berdasarkan karakteristik data harga saham syariah yang merupakan data deret waktu, algoritma penggerombolan yang akan diterapkan adalah agglomerative hierarchical clustering dengan metode jarak dynamic time warping. Parameter jarak terpilih adalah average linkage karena memiliki koefisien korelasi cophenetic tertinggi yaitu 0,643. Jumlah gerombol yang optimal berdasarkan kriteria koefisien silhouette dan rasio keragaman adalah lima gerombol. Setelah dilakukan uji chow breakpoint pada setiap emiten di setiap gerombol, terdapat tiga emiten di gerombol satu, tiga emiten di gerombol dua, satu emiten di gerombol tiga, dan satu emiten di gerombol lima yang tidak signifikan terjadi perubahan struktural pada periode sebelum dan sesudah munculnya pandemi Covid-19.

The government policies to break the chain of Covid-19 transmission have an impact on the economic sector in Indonesia. Many workers lost their job and have to find other ways to survive. One of them is by investing in stocks. Indonesia, as a country with a predominantly Islamic population, the sharia capital market, especially sharia stocks is high in demand. However, sharia stocks that have a high return rate also have high risks. One of the factors that make stock risk is high is due to the stock price that is volatile and stochastic. Therefore, it is necessary to cluster the stock issuers based on price pattern. Based on the characteristics of sharia stock price which is time series data, the clustering algorithm that were applied is agglomerative hierarchical clustering with dynamic time warping distance. The distance parameter that were chosen is the average linkage because it has the highest cophenetic correlation coefficient, 0.643. The optimal number of clusters based on the criteria of the silhouette coefficient and diversity ratio is five clusters. According to the chow breakpoint test there were three issuers in cluster one, three issuers in cluster two, one issuer in cluster three, and one issuer in cluster five with no significant structural changes the period before and after the emergence of Covid-19. The test was performed independently on each issuer in all cluster.