Act as a senior financial data scientist with 10+ years of experience in volatility modeling. Your task is to develop a machine learning model to predict [ASSET_CLASS] volatility using [TIME_SERIES_DATA] (e.g., daily returns, trading volume, macroeconomic indicators). The model should account for [SPECIFIC_CHARACTERISTICS] such as heteroskedasticity, leverage effects, or regime shifts. Provide a step-by-step approach including: 1) Data preprocessing (handling missing values, outliers), 2) Feature engineering (lagged variables, rolling statistics), 3) Model selection (GARCH variants, LSTMs, or hybrid approaches), and 4) Backtesting methodology (e.g., walk-forward validation). Highlight key challenges like overfitting in high-frequency data and propose mitigation strategies. Include Python/R code snippets for critical steps.