Abstract: The forecasting precision with the method of moving average can be improved by the optimal moving step length n. The authors put forward a way to accurately get the optimal moving step length n with the aid of computer. The forecasting precision should be predefined based on the actual problem firstly, and the optimal step length n can be found by adjusting the value of n to make the error rate between actual value and forecasting value minimal. The same way can be used to find the value of n in the method of multi-time moving average. If the forecasting value of n passes the test, this model can be employed in short-time forecasting of actual problem.