I want to estimate coefficients of the Present Value Model of a stock exchange (as defined here). The model uses just two variables, namely: "price at the beginning of period $t$" and "forthcoming dividend during period $t$". Assuming there is only one dividend per year, what the latter variable points to is quite straightforward. I am not however quite sure what is meant by the former one, prices do change daily. My first guess is to take the daily price on the very first day of the year when the stocks are traded. Would that be correct? Isn't it so that such an approach might make my model prone to some short-term imbalances?