Pairs trading com criptomoedasavaliação de estratégias econométricas e empíricas

Resumo

This study investigates the application of pairs trading strategies in the cryptocurrency market, focusing on the interactions between Bitcoin (BTC) and Ethereum (ETH), the two leading digital assets by market capitalization. The main objective is to assess the efficiency and robustness of different quantitative long-short trading approaches in an environment characterized by high volatility, rapid evolution, and informational asymmetries. The period from 2009 to 2025 is covered, using daily data and a rigorous econometric framework that includes stationarity tests (ADF), Johansen cointegration, Granger causality, half-life calculation, Kalman filter, and both univariate and multivariate GARCH models. Six distinct pairs trading strategies are implemented, ranging in sophistication from traditional models to adaptive approaches with dynamic estimation and conditional volatility. The results reveal long-term equilibrium relationships between BTC and ETH, with mean reversion observed in various configurations. Strategies with dynamic parameterization, such as univariate and multivariate GARCH, stand out for their higher predictive power, improved risk-return profiles, and superior hit rates compared to conventional approaches. The performance metrics analyzed—such as win rate, profit factor, and Sharpe ratio—indicate that adaptive models provide competitive advantages in the crypto environment. This dissertation contributes to both academic literature and digital asset management practice by demonstrating that systematic and quantitative approaches can be effective in volatile markets. Additionally, it deepens the discussion on causality between cryptocurrency pairs, particularly regarding the identification of the direction of information flows between BTC and ETH. Causality tests, such as Granger, show that correctly determining the causal direction is crucial for effective strategies, avoiding spurious relationships or causality inversion, which may compromise operational accuracy and risk assessment. Thus, the study highlights the relevance and direct impact of causality on the robustness and efficiency of models applied to the cryptocurrency market.

Ficha do documento

Tipo

Dissertação

Ano

2025

Instituição

Fundação Getulio Vargas

Fonte

Repositório da FGV

Idioma

Português

Acesso

Acesso aberto

Identificador

oai:repositorio.fgv.br:10438/37389

Temas

EconomiaTecnologia

Palavras-chave

Pairs tradingCryptocurrenciesCointegrationKalman FilterQuantitative strategiesMarket efficiencyCriptomoedasCointegraçãoFiltro de KalmanGARCHEstratégias quantitativasEficiência de mercadoEconomiaOperações com pares (Finanças)Kalman, Filtragem deModelos econométricos