Htydjtk

I have a variety of NFL datasets that I think might make a good side-project, but I haven't done anything with them just yet.

Coming to this site made me think of machine learning algorithms and I wondering how good they might be at either predicting the outcome of football games or even the next play.

It seems to me that there would be some trends that could be identified - on 3rd down and 1, a team with a strong running back theoretically should have a tendency to run the ball in that situation.

Scoring might be more difficult to predict, but the winning team might be.

My question is whether these are good questions to throw at a machine learning algorithm. It could be that a thousand people have tried it before, but the nature of sports makes it an unreliable topic.

There are a lot of good questions about Football (and sports, in general) that would be awesome to throw to an algorithm and see what comes out. The tricky part is to know what to throw to the algorithm.

A team with a good RB could just pass on 3rd-and-short just because the opponents would probably expect run, for instance. So, in order to actually produce some worthy results, I'd break the problem in smaller pieces and analyse them statistically while throwing them to the machines.

There are a few (good) websites that try to do the same, you should check'em out and use whatever they found to help you out:

• Football Outsiders


• Advanced Football Analytics

And if you truly want to explore Sports Data Analysis, you should definitely check the Sloan Sports Conference videos. There's a lot of them spread on Youtube.

Yes. Why not?!
With so much of data being recorded in each sport in each game, smart use of data could lead us to obtain important insights regarding player performance.

Some examples:

• 
  Baseball: In the movie Moneyball (which is an adaptation of the Moneyball book), Brad Pitt plays a character who analyses player statistics to come up with a team that performs tremendously well! It was a depiction of the real-life story of Oakland Athletics baseball team. For more info, http://www.theatlantic.com/entertainment/archive/2013/09/forget-2002-this-years-oakland-as-are-the-real-em-moneyball-em-team/279927/
  


• 
  Cricket: SAP Labs has come up with an auction analytics tool that has given insights about impact players to buy in the 2014 Indian Premier League auction for the Kolkata Knight Riders team, which eventually went on to win the 2014 IPL Championship. For more info, http://scn.sap.com/community/hana-in-memory/blog/2014/06/10/sap-hana-academy-cricket-demo--how-sap-hana-powered-the-kolkata-knight-riders-to-ipl-championship
  

So, yes, statistical analysis of the player records can give us insights about which players are more likely to perform but not which players will perform. So, machine learning, a close cousin of statistical analysis will be proving to be a game changer.

Definitely they can.
I can target you to a nice paper. Once I used it for soccer league results prediction algorithm implementation, primarily aiming at having some value against bookmakers.

From paper's abstract:

a Bayesian dynamic generalized model to estimate the time dependent skills of all teams in a league, and to predict next weekend's soccer matches.

Keywords:

Dynamic Models, Generalized Linear Models, Graphical Models, Markov
Chain Monte Carlo Methods, Prediction of Soccer Matches

Citation:

Rue, Havard, and Oyvind Salvesen. "Prediction and retrospective analysis of soccer matches in a league." Journal of the Royal Statistical Society: Series D (The Statistician) 49.3 (2000): 399-418.

Machine learning and statistical techniques can improve the forecast, but nobody can predict the real result.

There was a kaggle competition a few month ago about predicting the 2014 NCAA Tournament. You can read the Competition Forum to get a better idea on what people did and what results did they achieve.

It has been shown before that machine learning techniques can be applied for predicting sport results. Simple google search should give you a bunch of results.

However, it has also been showed (for NFL btw) that very complex predictive models, simple predictive models, questioning people, or crowd knowledge by utilising betting info, they all perform more or less the same. Source: "Everything is obvious once you know the answer - How common sense Fails", Chapter 7, by Duncan Watts.

Michael Maouboussin, in his book, "The Success Equation," looks at differentiating luck from skill in various endeavors, including sports. He actually ranks sports by the amount of luck that contributes to performance in the different sports (p. 23) and about 2/3 of performance in football is attributable to skill. By contrast, I used MM's technique to analyze performance in Formula 1 racing, and found that 60% is attributable to skill (less than I was expecting.)

That said, it seems this kind of analysis would imply that a sufficiently detailed and crafted feature set would allow ML algorithms to predict performance of NFL teams, perhaps even to the play level, with the caveat that significant variance will still exist because of the influence of luck in the game.

I have read some about it and I had the following blog in mind:

http://fellgernon.tumblr.com/post/46117939292/predicting-who-will-win-a-nfl-match-at-half-time#.UtehM7TWtQg

This blog deals with the prediction of a NFL match after the half time is already over. The prediction is 80% accurate with simple GLM model.

I do not know if that is suitable for soccer.

I've done some research in this area. I've found first order Markov chains work well for predicting within game scoring dynamics across a variety of sports.

You can read in more detail here:
http://www.epjdatascience.com/content/3/1/4