Abstract: A computerized system and method for forecasting future performance values based on past performance data, the system training one or more neural networks on the past performance data, with the training facilitated by a training scheduler that extracts data points from a time series data array and prepares training jobs to be processed by a training worker utilizing the neural networks to process the training jobs to create a number of predicted values that when aggregated can be used to calculate a mean predicted value and a confidence metric.