Developing recommender systems for cryptocurrencies

Abstract

Post the introduction of Bitcoin, the first complete cryptographic currency by Satoshi Nakamoto in 2008, there have been new varieties of digital currencies appearing each year. In today’s crypto market, there are over 10,500 publicly traded cryptocurrencies with a total market capitalization of close to 1.74 trillion dollars! The most prominent cryptocurrency, Bitcoin, is valued at 731.43 billion dollars, which is ~42% of the entire cryptocurrency market cap. Cryptocurrencies built on the underlying blockchain technology have not only enabled modernization in traditional financial systems but have also opened investment opportunities for investors. Despite risks associated with investing, cryptocurrencies as an investment instrument have gained significant popularity driven by steady innovations in the digital landscape. Investments in cryptocurrencies have seen an upsurge since the year 2017, though most of these investments have been driven by speculation. The increasing interest of fund managers in a cryptocurrency fund has led to research focusing on portfolio management, portfolio optimization, crypto price prediction, and price trend forecasting. This paper aims to study the existing cryptocurrency landscape and suggest creating a portfolio based on cryptocurrencies on a recent data set. For creating a portfolio, cryptocurrencies have been carefully chosen considering multiple aspects such as market cap, current supply, use cases, community, and pricing history, among others. For cryptocurrency portfolio optimization, we used several metrics such as Naive portfolio. mean-variance portfolio, and optimization using the Sharpe ratio. The overall objective is to develop a recommendation system to recommend cryptocurrency portfolios, allowing investors to profitably invest in cryptocurrencies.