摘要：This paper utilize ANNs model to predict closing price of AshakaCem Security in Nigeria Stock Market price index. In this paper, AshakaCem Security historical technical data were collected for four years trading period (2005 – 2008), the data was partition into training, cross validation and testing set in the ratio 70%:10%:20% respectively. Architectural configuration of Feed Forward Artificial Neural Networks (FFANN) with parameters of three (3) layers, four (4) input nodes, one (1) hidden layer, eighteen (18) hidden nodes, one (1) output layer and one (1) output node was obtained and FFANN was build with the calibrations, trained with 268 exemplars, cross validated with 37 exemplars and tested on 76 exemplars and evaluated on four (4) performance indicators which include Mean Square Error (MSE), Correlation Coefficient (r), Normalize Mean Square Error (NMSE) and Mean Absolute Error (MAE). The technical data were subjected to sensitivity analysis. FFANN predictor was developed and modeled historical trading data of AshakaCem Security and pattern was captured in the historical data with a trend accuracy of 80%, Efficient Market Hypothesis was contradicted in this paper and sensitivity analysis shows that previous closing price (pclose) is the most significant input on FFANN predictor output. FFANN predictor build was evaluated and result shows that MSE = 3.59739909, NMSE = 0.039391504, MAE = 1.3407 and r = 0.981331937. It is possible to trained ANNs with controlled parameters using technical data of AshakaCem Security to capture pattern in the historical data and generalize well on unseen data.
