Model peramalan indeks kripto setelah masa pandemi Covid-19 dengan Markov Switching Autoregressive

Abstract

This study applies the Markov Switching Model Autoregresive Models (MSAR) with two regimes to predict movements in the Bitwise crypto index. An autoregressive model with one lag (AR1) is used to capture the complex dynamics of the crypto market, focusing on two main phases: bullish and bearish. The model accounts for the probability of transitioning between regimes, allowing it to identify when the market shifts between positive and negative conditions. The estimation results show that the model fits the data well, with key variables such as historical prices and other indicators providing accurate predictions in both regimes. In both phases, the model exhibits near-perfect fit, indicating it explains almost all variability in the data. However, despite the strong fit, some parameters do not show high statistical significance, which may point to challenges in the estimation process. The transition probabilities reveal that the bearish condition is more dominant and tends to persist longer, while transitions to the bullish phase occur with lower probability. Thus, the use of the Markov Switching Model offers deeper insights into crypto market movement patterns, especially in identifying sudden shifts between different market phases. These findings are relevant for investors and analysts, providing a better understanding of volatility and aiding in investment decision-making strategies in the crypto market.