Resistências, limites e potencial dos criptoativos em portfólios de investimentos: a regionalidade importa?

Abstract

Contextualization: Blockchain technology and cryptocurrencies have unveiled a universe of possibilities across various fields, including Finance, with applications in payment services, loans, fundraising, and transferring resources. They also expand investment options for individual and institutional investors by being recognized as a new asset class. In the investment context, managing cryptocurrency portfolios faces the challenge of their high intrinsic volatility, requiring innovative risk management approaches. The use of autoregressive models focusing on predicting future volatility, and machine learning models with deep neural networks, shows potential to revolutionize investment approaches in crypto assets, allowing for more dynamic and strategic adaptation to rapid market changes for more informed and robust investment strategies.

Objective: To investigate the potential and effectiveness of advanced portfolio management strategies composed of cryptocurrencies and their implications at a regional level.

Method: The first study conducts a systematic literature review focusing on risk management, portfolio use, and market efficiency. Then, the second study estimates the future volatility of a cryptocurrency portfolio using DCC-GARCH and tests the automated trading strategy Reaction Trend System (RTS) by Fiorucci, Silva, and Barboza (2022a) for a set of cryptocurrencies previously selected according to the estimation models’ criteria. To evaluate the strategy’s effectiveness, it was compared with the passive buy and hold strategy for the same set of cryptocurrencies. Finally, the third study utilizes a deep reinforcement learning approach for dynamic portfolio selection and optimization. The algorithms Proximal Policy Optimization, Deep Deterministic Policy Gradient, and Soft Actor-Critic SAC were tested on cryptocurrencies with available data.

Results: In article 1, the systematic literature review identified significant gaps in the literature on risk management and market efficiency for cryptocurrencies, highlighting the need for innovative approaches for portfolio selection and optimization. From testing automated strategies in cryptocurrencies, article 2 demonstrated that the strategy based on the Reaction Trend System (RTS) combined with DCC-GARCH future volatility forecasting was effective for cryptocurrencies, outperforming passive strategies and traditional approaches in terms of risk-adjusted return. However, the strategy requires careful adjustments for each asset due to its sensitivity to configurations. Finally, in article 3, DRL models, especially Soft Actor-Critic (SAC) and Deep Deterministic Policy Gradient (DDPG), showed potential for optimizing cryptocurrency portfolio management and improving returns compared to classic strategies, particularly in portfolios composed of smaller or less liquid assets. These results strengthen the thesis that cryptocurrencies, when managed with advanced modeling and machine learning techniques, can form efficient investment portfolios, contributing to innovation in financial asset management.

Research Alignment with PPGAdm Concentration Area (Regionality and Management) and Research Line: The research addresses the topic of cryptocurrencies, which are global digital assets. However, the adoption, regulation, and impact of cryptocurrencies can significantly vary from one region to another. By studying cryptocurrencies, the research can contribute to a better understanding of how these digital assets are being adopted and managed in different regions. Additionally, the research involves the use of advanced Machine Learning techniques to optimize cryptocurrency investment portfolios. This directly aligns with the management theme, as it involves strategic decision-making on how to allocate resources in an investment portfolio. The research may have implications for organizations in terms of how they manage their cryptocurrency investments, considering specific regional factors such as local regulations and market conditions. Therefore, although the research is highly technical and focused on a specific type of digital asset, it fits well within the PPGAdm area of concentration and the research line "Organizational and Regional Management."

Impact and Innovative Nature in Intellectual Production: The research contributes to the existing literature on investment portfolio optimization and Machine Learning by introducing the use of Deep Reinforcement Learning (DRL) in the management of cryptocurrency portfolios. This represents an innovative approach that could open up new possibilities for future research. Additionally, the research also highlights the importance of considering transaction costs, which can have significant implications for the effectiveness of DRL models.

Economic, Social, and Regional Impact: The research has the potential to influence how organizations and individuals manage their cryptocurrency investments. This can have significant economic implications given the rapid growth and volatility of the cryptocurrency market. Furthermore, by considering regional differences in the adoption and regulation of cryptocurrencies, the research can also contribute to a better understanding of the social and regional impacts of these digital assets. For example, in regions where cryptocurrencies are widely adopted, the research insights may be particularly relevant.

Regional Implications: The expectation is that the results will promote improved efficiency in both the volatility estimation and the management of cryptocurrency portfolios.

Sustainable Development Goals Addressed in the Research: Considering the UN's 2030 Agenda, the Sustainable Development Goals (SDGs) and targets addressed by the research, this thesis aligns with several SDGs, mainly 8, 9, and 10. SDG 8 (Decent Work and Economic Growth) – By investigating the use of advanced technologies for risk management and optimization of cryptocurrency investment portfolios, the research can promote sustainable economic growth and offer more efficient investment methods in the financial market. SDG 9 (Industry, Innovation, and Infrastructure) – through the development of advanced technologies and automated tools for cryptocurrency portfolio management, the research fosters innovation in the financial sector, encouraging more efficient and sustainable investment practices. SDG 10 (Reduced Inequalities) – by democratizing informed access to innovative portfolio management methodologies in crypto assets, it contributes to reducing inequalities in access to the financial market.