Crypto Price Forecasts Using Markov Switching Models

Cryptocurrency adoption in Indonesia has grown rapidly, with more people registering to trade digital coins than those investing in traditional stocks. However, even as investor numbers climb, the total value of daily transactions keeps dropping. This drop is linked to how unstable crypto prices can be, making many investors hesitant to put in more money.

To help reduce risk and losses, researchers looked at ways to predict crypto prices better. They focused on the five biggest cryptocurrencies by market size: Bitcoin (BTC), Binance Coin (BNB), Ethereum (ETH), XRP, and Cardano (ADA). The team used something called the Markov Switching Autoregressive model. This method helps spot hidden patterns in price data by switching between different states over time.

The results showed that each coin worked best with a slightly different version of the model. For BTC, BNB, and ETH, the top choice was MS(3)AR(1). For XRP and ADA, it was MS(3)AR(2). These models were able to capture major shifts in market behavior and still give useful short-term forecasts.

When checking how accurate the predictions were, the study found that Bitcoin had the lowest error rate. In other words, BTC prices were predicted more accurately than the other coins. This could be because Bitcoin has more stable trading patterns compared to smaller or newer cryptocurrencies.

Overall, the study suggests that advanced statistical models like the Markov Switching approach can help investors make smarter choices. By understanding which model fits each coin best, traders may feel more confident navigating the risky world of crypto investing.