If you are interested in learning more about this topic or other methods of statistically analyzing the soccer transfer market, you may be interested in the following resources:
This paper tackles a question similar to the feature selection process of our research: what characteristics contribute to or correlate with a high transfer fee. While our model only utilizes game statistics, this paper also considers external factors such as fan appeal (number of Google searches) or race.
This paper dives deeper into our second question: can we predict the transfer fee of a player from their productivity. Unlike our model, however, which was a non-linear neural network, they only used linear techniques for prediction. This does simplify the problem, but the model may not generalize well to future data if there is a complicated relationship between the different game statistics and the ultimate transfer fee. For example, if a doubling in assists results in a double of transfer fee, while a doubling of goals results in 8 times the transfer fee.