Pricing Football Players using Neural Networks

TL;DR

This paper presents a neural network model trained on FIFA 2017 data to accurately predict football player prices, exploring various training parameters and achieving high accuracy in categorization and price estimation.

Contribution

The study introduces a neural network approach for football player valuation and systematically investigates training techniques and hyperparameters for optimal performance.

Findings

Top-5 accuracy of 87.2% in price category prediction

Average price estimation within 6.32% of actual prices

Effective neural network configuration for sports player valuation

Abstract

We designed a multilayer perceptron neural network to predict the price of a football (soccer) player using data on more than 15,000 players from the football simulation video game FIFA 2017. The network was optimized by experimenting with different activation functions, number of neurons and layers, learning rate and its decay, Nesterov momentum based stochastic gradient descent, L2 regularization, and early stopping. Simultaneous exploration of various aspects of neural network training is performed and their trade-offs are investigated. Our final model achieves a top-5 accuracy of 87.2% among 119 pricing categories, and places any footballer within 6.32% of his actual price on average.