The average cost for the dynamics lot-sizing model is defined as sum of all costs associated with feasible solutions, divided by the number of these solutions. This benchmark is used to estimate the MCA heuristic (Groff) and the Silver-Meal-heuristic. The first heuristic is found to produce solutions with cost not higher than average in all inventory periods and some ordering periods while the second one is shown to provide solutions with cost higher than average in the worst case.