Following the implementation of Basel III and its forthcoming finalization, risk management has been already paid high attention by firms, financial institutions and even the Basel Committee on Banking Supervision. To examine the long-term and short-term market risk, a time-varying scale function is introduced based on parametric models. The market risk is decomposed into a short-term risk and a long-term risk. Due to the free parametric model, the estimation of the scale function is without parametric assumption. Iterative plug-in algorithms are developed to estimate the bandwidth and the power parameter. The moving block bootstra is carried out to estimate the confidence interval of the power parameter. The empirical findings indicate a good performance of the general power transformed semiparametric models.VaR and ES are also predicted by the general semiparametric models. Well known models, such as the GARCH class models, are as parametric extensions. Backtesting with the semiparametric approach for both VaR and ES are also discussed. Although the Kupiecs POF test and independence test are carried out, the robustness of the results is challenged. Following Basel III, the traffic light test, considering the cumulative probability, is applied. For ES, a breach indicator is introduced. Loss functions from regulators and firms are also discussed. In the empirical cases, the semiparametric models are proved to be necessary tools in risk management.