The article explores how to better estimate risk in stock market investments by using time-varying factor models. Traditional risk models can be unreliable during market turbulence due to structural changes not accounted for. By using estimates from a single factor model that changes over time, the researchers show improved risk estimation, portfolio selection, and investment performance. This approach focuses on accurately estimating the variance-covariance matrix, which is crucial for optimizing portfolios. Previous research has mainly focused on predicting returns, but there is a lack of evidence on the performance of different risk models. The study highlights the importance of accurately estimating risk for successful investment strategies.