ABSTRACT
One of the difficulties encountered in modelling the price series of shares listed on the Johannesburg Stock Exchange is that many of the shares are only thinly traded. The market is heavily dominated by institutional investors, and if for any reason a share happens not to be an ‘institutional favourite’ there will very likely be days, or even weeks, during which no trading of that share takes place. One approach is to model the presence or absence of trading quite separately from the modelling of the price achieved when trading does take place. This is analogous to the modelling of the sequence of wet and dry days separately from the modelling of the amounts of precipitation occurring on the wet days. It is therefore natural to consider, as models for the trading pattern of one or several shares, HMMs of the kind discussed by Zucchini and Guttorp (1991), who used them to represent the presence or absence of precipitation on successive days, at one or several sites.
