This paper presents a new methodology to overcome issues with proprietary, limited, or censored data. The methodology uses an artificial neural network (ANN) to translate a player`s rating values to NFL Combine values. Though the Combine values typically are recorded once during a professional`s career, the rating values exist well past a player`s rookie year in professional football. After the ANN is trained, additional Combine performance values are forecasted using the rating database. Once the forecasts are made, performance-aging curves are applied to the aggregated data to investigate the impact that aging has on a player`s expected Combine performance. The results of this methodology are somewhat consistent with other related literature in term of determining peak-performance age and rates of improvement; however, as expected, the rates of decline after peak are higher in football than other sports that are far less physically natured.
© Copyright 2008 International Journal of Sports Science and Engineering. World Academic Press. All rights reserved.