# Strategy for trading weekly options

We will see that this type of model is not just different conceptually, it is also quite different geometrically, from models that assume constant deterministic preferences (or utilities) perturbed by random errors. If we can communicate to QT est what constraints this model imposes on binary choice probabilities, then the program can carry out a quantitative test.

We discuss in the Online Supplement how one can find a complete and nonredundant list of such constraints. Consider the possibility that the decision maker, at any point in time, rank orders the gambles from best to worst in a fashion consistent with. Mixture, aka, random preference models quantify this variability with a probability distribution over preference patterns such as, in this case, the four rankings ACD, ADC, DAC, DCA.