A novel algorithm for prediction of crude oil price variation based on soft computing

In this paper a method based on soft computing approaches is developed to predict the daily variation of the crude oil price of the West Texas Intermediate (WTI). The predicted daily oil price variation is compared with the actual daily variation of the oil price and the difference is implemented to activate the learning algorithms. In order to reduce the effect of unpredictable short term disturbances, a data filtering algorithm is used. In this paper, the prediction is called "true" if the predicted variation of the oil price has the same sign as the actual variation, otherwise the prediction is "false". It is shown that for several randomly selected durations, the true prediction is considerably higher than the result of most recent published prediction algorithms. To ensure the accuracy and reliability of the algorithm, several on line predictions are executed during one complete month. The on line results indicate that the true predictions are consistently the same percentage for periods of one month.