What simple thing can every NFL team do to win more games?

Besides being one of America's favorite pastimes, Football is big business. Nobody knows this better than the NFL. As a product that's delivered to millions of fans each week, football teams definitely want to beat the competition.


Obviously sports are unscripted, but that doesn't mean there aren't steps teams can take to be more successful. Let's start exploring this by examining which players score the most points. This chart shows the average points per game for NFL players. The structure of the chart is organized by NFL division, subdivisions, and team. See the clean division between the top and bottom half of this chart? That's because this chart first the chart divides the NFC from the AFC. It then divides by subdivision: North, East, South, West and finally by team. Each box represents a player. This chart colors boxes by group. For instance, the grey boxes in the bottom right are all a part of the group: AFC > West > Broncos. Man the Broncos are doing really well right now. Click on this set of grey boxes to see more details. It looks like the box for Prater is the largest; Prater scores the most points per game for the Broncos.


The chart above is a tree map. Tree maps are great because they show the size of different concepts while also organizing these concepts hierarchically. The additional use of color and an interactive zoom feature further enrich the analytical power of this approach for visualizing complex ideas. I created the chart above by getting some data from the NFL. Then I used Python to create a data object. Finally I graphed this object using d3 and some techniques first developed by Mike Bostok (who now does information visualization for the New York Times).


Speaking of the New York Times, they recently featured an article that discussed the use of predictive analytics to create a robot coach. The robot coach is built on a data science concept referred to as conditional probability distributions. For now I'll say that these use evidence of past success to predict future success. I'll also talk about these more in an upcoming post. The robot coach tells teams they should punt less often on 4th down and instead either go for a field goal or for a touch down. Think for a minute about this recommendation and ask the question, what's special about field goals? Take a look at this chart:


When visualizing the average points per game for players from various teams, one thing immediately stood out to me. All of the top scoring players were all kickers! You may have noticed the gigantic group of orange boxes to the right in the graph above. Well this huge set of players are all kickers. Kickers are seriously talented. Just this week Bronco's kicker Matt Prater kicked an all time record-setting 64 yd. field goal!


Prater's epic field goal- that's a huge distance. Nothing makes this distance seem larger than actually visualizing what it'd be like to kick such a distance. Here is a picture I took a while back while walking around in Century Link field. Now let's focus a bit on the important team, the Seattle Seahawks. How can they win more games? Well I'd recommend that they give Hauschka a few more plays.
High level view of Century Link field in Seattle, WA.

Where is your favorite team at? Take a look at this chart which shows total points per team. Points per game is important since the number of total points per game relates to the win-loss ratio for any given team. To win more, you need to score more points.


As mentioned, the highest scoring players are kickers. Using better kickers and using these kickers more often (as robot coach also suggests) should increase a team's win ratio. But of course a kicker on their own cannot win a game. It requires team work and lots of fans to be successful in the one and only NFL. What's the connection between win-ratio and points per game? What other key statistics can help us predict the successful of a team? I'll dig into these issues in more in the future.

Tags: