This paper employs ARMA-GARCH model to analyze and predict the volatility of the U.S. stock market and evaluates the effectiveness of the model using the three main stock indexes ranging from S&P, Dow Jones Industrial Average to NASDAQ Composite Index. The value at risk is estimated in this paper to give a measure of the risk of loss and an early warning for the American stock market crisis. The result shows a high risk in the overall U.S. stock market and there exists stationary, autocorrelation, and volatility clustering effect in the return series. Moreover, the volatility model has proper goodness of fit and strong robustness.