Pembentukan Portofolio Optimal Hierarchical Risk Parity Dan Nested Clustered Opimization Terhadap Forecasting Lstm Harga Cryptocurrency Trending Twitter

Nugraha, Nathanael Satria (2022) Pembentukan Portofolio Optimal Hierarchical Risk Parity Dan Nested Clustered Opimization Terhadap Forecasting Lstm Harga Cryptocurrency Trending Twitter. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06311840000032_Undergraduate_Thesis.pdf Download (5MB)

Abstract

Investasi meningkat di era pandemi, dimana tercatat kenaikan sebesar 2 juta investor menurut data Single Investor Identification dari 4 juta menjadi 6 juta. Salah satu investasi yang sedang ramai adalah investasi cryptocurrencies. Cryptocurrencies adalah mata uang digital yang bisa digunakan untuk bertransaksi. Banyak orang tertarik dengan investasi ini karena memberikan return yang besar. Hal itu semakin ramai dengan adanya media sosial yang sering membagikan hal-hal manis dari investasi cryptocurrencies. Namun cryptocurrencies memiliki volatilitas yang tinggi. Karena itu diperlukan analisis yang tepat untuk bisa berinvestasi di cryptocurrencies. Forecasting adalah salah satu metode yang bisa meramalkan harga di masa yang akan datang. Penelitian ini memanfaatkan data yang didapat dari Twitter yang membahas cryptocurrencies. Didapatkan 10 crypto yang sedang trending di Twitter pada periode 2021. Cryptocurrencies tersebut adalah Arweave (AR-USD), REN (REN-USD), ThunderCore (TT-USD), Bitcoin (BTC-USD), TerraUSD (UST-USD), Amp (AMP-USD), Ethereum (ETH-USD), Aragon (ANT-USD), Kin (KIN-USD), Huobi Token (HT-USD). Kemudian dilakukan forecasting menggunakan metode Long Short-Term Memory (LSTM) terhadap harga closing cryptocurrencies terpilih untuk 30 hari pertama bulan 2022. Setelah didapatkan hasil forecasting tersebut disusunlah portofolio optimal menggunakan machine learning algorithm yakni metode Hierarchical Risk Parity (HRP) dan Nested Clustered Optimization (NCO). Pada optimasi menggunakakan metode HRP didapatkan bahwa portofolio tidak terdiversifikasi dengan satu crypto yang mendominasi portofolio yaitu UST-USD, sehingga dibentuk alternatif portofolio HRP dengan menghilangkan UST-USD. Pada portofolio menggunakan metode NCO portofolio cukup terdiversifikasi dikarenakan portofolio tersusun dari beberapa crypto. Berdasarkan nilai sharpe ratio dan sortino ratio portofolio menggunakan NCO lebih baik dibanding portofolio dengan menggunakan metode HRP maupun alternatif HRP dengan nilai sharpe ratio 0,105 dan nilai sortino ratio 3,744 dibandingkan dengan portofolio HRP dan alternatif HRP dengan sharpe ratio sebesar 0,008 dan 0,075 serta sortino ratio sebesar 2,592 dan 2,973. Dapat disimpulkan bahwa pembentukan portofolio optimal menggunakan metode NCO lebih optimal dibandingkan dengan HRP maupaun alternatif HRP
==============================================================================================================================
There was an increase in investment in the pandemic era, which recorded an increase of 2 million investors according to Single Investor Identification data from the previous 4 million to 6 million. One of theinvestment that is currently trending is invest in the cryptocurrencies. Cryptocurrencies are digital currencies that can be used for transactions. Many people are interested in this investment because it promises a large return. This is exacerbated by the existence of social media which often shares good things from investing in cryptocurrencies. But on the other hand cryptocurrencies have very high volatility. Therefore, proper analysis is needed to be able to invest in cryptocurrencies. Forecasting is one method that can predict future prices. This study utilizes data obtained from Twitter which discusses cryptocurrencies. There are 10 cryptos that are trending on Twitter in the 2021 period. The cryptos are Arweave(AR-USD), REN(REN-USD), ThunderCore(TT-USD), Bitcoin(BTC�USD), TerraUSD(UST-USD), Amp(AMP-USD), Ethereum(ETH-USD), Aragon(ANT-USD), Kin(KIN-USD), Huobi Token(HT-USD). Then forecasting is carried out using the Long Short-Term Memory (LSTM) method on the closing prices of selected cryptocurrencies for the first 30 days of 2022. After obtaining the value from the forecasting results, an optimal portfolio is formed using a machine learning algorithm using the Hierarchical Risk Parity (HRP) method and Nested Clustered Optimization (NCO). In optimization using the HRP method, it was found that the portfolio was not diversified with one crypto that dominates the portfolio, namely UST-USD, so an alternative HRP portfolio was formed by eliminating UST-USD. In the portfolio using the NCO method the portfolio is quite diversified because the portfolio is composed of several cryptocurrencies. Based on the sharpe ratio and sortino ratio values, the portfolio using NCO is better than the portfolio using the HRP method or alternative HRP with a sharpe ratio value of 0,105 and a sortino ratio value of 3,744 compared to the HRP portfolio and alternative HRP with a sharpe ratio of 0,008 and 0,0747 and sortino ratio of 2,592 and 2,973. It can be concluded that the formation of an optimal portfolio using the NCO method is more optimal than the HRP or alternative HRP

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Cryptocurrencies, Long Short-Term Memory, Portofolio Optimal, Hierarchical Risk Parity, Nested Clustered Optimization, Sharpe Ratio, Sortino Ratio
Subjects:	H Social Sciences > HG Finance > HG4529 Investment analysis
Divisions:	Faculty of Mathematics, Computation, and Data Science > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	26 Nov 2024 03:47
Last Modified:	26 Nov 2024 03:47
URI:	http://repository.its.ac.id/id/eprint/115834

Actions (login required)

View Item