Bitcoin Forecasting with Classical Time Series Models on Prices and Volatility

Battle Count: 116

The paper provides a baseline evaluation of classical time series models for Bitcoin price and volatility forecasting, which is directly relevant to quantitative trading. ARIMA for short-run price dynamics and EGARCH for asymmetric volatility modeling are foundational tools in crypto trading strategies. However, the paper does not develop trading strategies, backtest P&L, or compare against ML baselines, limiting its direct applicability to production trading systems. It serves more as a methodological foundation than a trading system.

All models used (ARIMA, SARIMA, GARCH, EGARCH) are standard statistical methods available in common software packages (R, Python statsmodels, arch). Preprocessing involves simple log transformation and filtering. The main complexity lies in model selection and parameter tuning, but the specifications used (1,1,1) are straightforward. No custom code or advanced infrastructure is required.

Data is publicly available on GitHub (btc data.csv). Models are standard classical time series methods with well-documented implementations. The 90/10 train-test split is clearly specified. However, no specific code repository or software version details are provided, and the exact filtering and preprocessing steps could benefit from more detail.