The Performance of Grey System Agent and ANN Agent it determines the winning bid and announces the winner. 7. THE GREY SYSTEM PERDICTOR AGENT ALGORITHM Based on the algorithm of GM (1, 1) which we discuss in Section 4, the Grey System agent al- gorithm is designed to make use of the simulated marketplace which proposed by Anthony (2003). Firstly, the Grey System Predictor Agent col- lects the observational data from the simulated marketplace after a particular run. The last eight observational data are reserved for the testing purpose (calculation for the residual error). Pre- dictor Agent then utilizes a number of the latest remaining observational data to generate two constant numbers, a and b in Formula 4 (which are used to build the prediction equation) for GM data are used to predict the second future value (twelfth data). This process is repeated for the third future value and so forth. This is because in the real-time auctions, there will be auctions that close at each time step. Hence, the involve- ment of the latest or updated observational data is important in order to increase the accuracy of the prediction. In the fixed situation, the Grey System Agent utilized four latest remaining fixed observations to predict the five future closing prices. The num- ber of fixed observation is increased until eight observations for each particular run. However in the moving stage, four moving observational data are utilized in order to predict the eight future closing prices. As in the previous case, the prediction will loop for seven times. Eight predicted data from each approach would then be compared to the original observational data in order to calculate their residual error. The aver-