I've recently become interested in possibly of developing some sort of method for ranking athletes of sports such as American football and determining which players are better than others in terms of specific statistics.

My thoughts are that there are two ways to go about doing this. The first would be some sort of mathematical formula which would take in the statistics of a given player and provide some sort of standardized score which could be compared with other players to determine which is better.

My other idea would be to have some machine learning algorithm go through historical data and determine the patterns which indicate how well a certain combination of statistics would perform in the following week of play by using the patterns it recognizes over time.

I'm not sure which approach would be more effective and so I'm hoping that someone has an idea or any advice as to which would be best to look into. Thanks!


1 Answer 1



If the main goal is predicting anything, say, the statisitcs of the player in the next game, game result, then I would not recommend to do any scoring. Better way to go is using the pure statistics data as an input to the model. Any scoring/rankning - is information loss.


If the goal is ranking itself, than you still need to have some target variable to predict. As you may want to check real predictive value of those ranks. That could be, again, playser stats in the next game or game result itself.


Sport scores prediction and RFM scoring are probably the next directions for you to look at.

  • $\begingroup$ Thanks, your answer and some extra research seems to indicate to me that pure statistical data would be better. I'm not worried much about ranking...this is more along the lines of analyzing a trade for players in a fantasy sports league to determine if trading player A for player B would be recommended or not. Hope that helps you better understand my intention :) $\endgroup$ Feb 4, 2015 at 0:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.