Rolling regression is the analysis of changing the relationship between different variables over time. Visualizing regression outputs like correlation, r-squared, beta and the standard error over time can be helpful in the analysis of risk for stocks, portfolios and factors.