This study focuses on improving the accuracy of forecasting cryptocurrency market data patterns using daily Bitcoin (BTC) close price, return, and volatility risk data. The dataset consists of 1535 observations spanning from January 2020 to January 2024. The modeling approach incorporates a nonlinear spectral model called the Maximum Overlapping Discrete Wavelet Transform (MODWT) with various mathematical functions such as Haar, d4, la8, bl14, and c6. The study found that the ARIMA-direct model with an ARIMA(0,1,0) configuration performed best for close prices in the 80% subset. In the 20% subset, the d4 model with an ARIMA(3,1,0) configuration showed the lowest error measures. For return data in the 80% subset, the la8 model with an ARIMA(1,1,0) configuration had relatively low errors, while the ARIMA-direct model with an ARIMA(0,1,0) configuration had low RMSE and MAE values. Regarding volatility risk, in the 80% subset, the Haar model with an ARIMA(3,1,0) configuration performs well, achieving low error measures. In the 20% subset, the d4 model with an ARIMA(1,1,0) configuration with drift stands out as the best performer based on ME, RMSE, MAE, MAPE, and MPE metrics. These results highlight the effectiveness of the MODWT model in forecasting cryptocurrency market data, specifically in capturing close price, return, and volatility risk patterns. The findings suggest that the proposed methodology can enhance forecasting accuracy for other cryptocurrencies as well, contributing to improved decision-making within the economic domains.