This paper proposes a bi-directional inferencing artificial neural network model for the effective formulation of short-term, medium-term, and long-term strategic management plans and for effectively resolving the trade-offs that may arise among these strategies. Bi-directional inferencing refers to forward inference and backward inference: forward inference enables the formulation of short-term, medium-term, and long-term strategies, while backward inference enables comparison among the strategies. Additionally, this study proposes a decision-making mechanism using linear programming to rationally resolve the trade-offs among short-term, medium-term, and long-term strategies that commonly arise during strategic management planning. In this study, linear programming is used to determine optimal production quantities by product and corresponding expected profits while considering competitive conditions. The bi-directional inferencing artificial neural network model proposed in this study is named SPBINN, an acronym for Strategic Planning Bi-directional Inferencing Neural Network. The learning model applied to SPBINN is the widely known backpropagation learning algorithm, and the training data were structured to enable bi-directional inferencing. The strategic planning simulation methodology based on SPBINN proposed in this study can be summarized as follows. First, training data and the artificial neural network model are constructed to enable both forward and backward inferencing. Second, separate artificial neural network models are built for short-term, medium-term, and long-term strategies. Third, production quantities by product are determined through linear programming, representing optimal short-term production quantities. Fourth, these production quantities are used as inputs to run the forward inferencing neural network model to derive appropriate short-term, medium-term, and long-term strategies. Fifth, if these three strategies are consistent with one another, the process stops and the corresponding strategy is regarded as the optimal strategy; otherwise, the process proceeds to the next step. Sixth, if the three strategies are inconsistent, the backward inferencing neural network model is run based on the corresponding strategies. Seventh, the results of backward inferencing are input into the linear programming model to calculate expected profits for each strategy. Eighth, the strategy with the highest expected profit is selected. This strategic planning simulation method was applied to the competitive situation in the Korean cosmetics market, yielding highly useful results.