A modeling technique is presented that encapsulates the behavior of a crowd that may or may not become hostile. The various parameters such a slogans shouted, grouping, density, age, occasion, leadership, etc. among many other factors can easily be estimated and submitted to the model. The model will then begin a prediction process that can be corrected as more data is obtained. A predictor corrector process is described that re-guides the prediction process. In addition, we use the Metropolis simulation algorithm with input from the Boltzman weighting factor to determine how individuals within the crowd may be influenced to follow a particular path, and cellular automata to control the sphere of influence by one individual over another. Lastly, we provide some sample output from the model to illustrate the flow of such a dynamical environment.