The article explores ways to improve predictions of financial risk by studying how different types of moments in financial data change over time. By comparing various models, researchers found that a specific model with certain characteristics performed just as well as more complex models in predicting value-at-risk accurately. This simpler model is also easier to use computationally.