Data Science in the NFL: Finding the Right Players and Strategies
Who are the best NFL players and why? This is a question that NFL teams want to answer, or perhaps they just want confirmation. NFL teams spend a lot of money on scouts to find the best future players, and on staff to determine if their current players are up to par. Teams already have sources to tell them who are the best players and why. So why are some teams beginning to hire data scientists to analyze a players stats to determine his value? How does data science help determine whether a player is good or not?
With football being perhaps the most popular, talked about, and drama filled sport in America, NFL teams invest a lot of money to become the team that everyone talks about, pays to watch, pays to be endorsed by, and generally just pays. The best way to do this is to win a Super Bowl (or have Tim Tebow on your team).
How do you win a Super Bowl?
To win you have to get the right players on your team. Not the best, but the right players. This has been a proven strategy. Contrary to popular opinion, It’s more about how well a team plays together to execute effective plays, even if one man’s role is perform a glory-less block that no one sees. This makes finding players more difficult than it might seem. If your goal is to make a cohesive team, full of players who complement each other rather than sign the players with the most star studded track record, you have to consider several different categories of not just a few future players, but of all your current players.
How does data science fit in with recruiting?
Good staff and good scouts are priceless, because they have rare abilities and experiences. However they are limited to evaluating only what they remember seeing. They are limited to finding simple patterns. Using mathematical models allows teams to find more complex patterns. The models not only tell them what traits the team needs, but also the players that exhibit those traits.
This would be similar to a business trying to find who they should hire next, based on who is currently on the team and what clients they would work with. Imagine working at a place were people are selected because their skills compliment the other members of the team.
Data Science in the game
Where else could the NFL use this kind of analysis? Let’s take a look at game planning shall we. Any football fan will hear at some point in time that the most prolific and successful quarterbacks in the NFL study game film FOREVERRRR (picture Squints from Sandlot), in order to find weakness’ in a defense. How does a QB find these weakness’ in defenses? They see patterns in certain players, such as: every time Troy Polamalu plans to blitz he nearly goes offsides because of impatience to take out the opposing QB. The only problem with this method is that it takes a long time to see with the human eye…hours in fact. What if your local, handy, some say “genius” data scientist could generate a report for you to show the areas of weakness in every player on the defense, based on the patterns that their mathematical model highlighted. It accomplishes the same goal, only it does it more efficiently, and without the mental strain on your most valuable player. In doing this, you could nearly predict how the defense will react to the types of plays and schemes your offense will run. Kinda similar to predicting how consumers might react to the kind of marketing you put out there, or how your current customers might react to a new product launch.
With data science, companies and organizations could save millions,some already are. The NFL is starting to implement analytics into evaluating and confirming what they hope they already know about players and personnel, but soon they will begin to see that not only is this method more efficient, it’s also more accurate. Are you ready to work smart?
Check out our other blog posts: