Sunday, September 13, 2009

Refchat Index: Analyzing Strength of Schedule for the 2009 NFL Season

In our last blog posting, we explained that the visiting team's offense has a surprisingly smaller impact on the result of the game (both when you account for scoring margin and when you focus only on which team won) than either the visiting team's defense, the home team's defense, or the home team's offense.

We can use this to make an adjustment in predicting a game or analyzing which team has a tougher schedule. We'll make an adjustment that puts the most weight on the home team's offense, a good amount of weight on each team's defense, and a small weight on the visiting team's offense. Let's call this the Refchat Index.

Let's look at which teams have the strongest and weakest schedules in the 2009 NFL season based on how their opponents did in 2008. We'll focus first on the scoring margin for each team. Next, we'll roll out the Refchat Index and analyze the schedules using our weighted analysis.

Under raw scoring margin, the teams with the easiest schedules based on how the opponents fared in 2008 are: Green Bay Packers, Arizona Cardinals, Minnesota Vikings, Baltimore Ravens, and the Seattle Seahawks. The teams with the hardest schedules based on raw 2008 scoring margin are: Miami Dolphins, Carolina Panthers, Tampa Bay Bucs, Atlanta Falcons, and the New England Patriots.

But using the Refchat Index, the teams with the easiest 2009 NFL schedules using 2008 data are: Arizona Cardinals, Green Bay Packers, Pittsburgh Steelers, St. Louis Rams, and the Chicago Bears. The teams with the hardest schedules are: Carolina Panthers, Miami Dolphins, Atlanta Falcons, New England Patriots, and the Tampa Bay Bucs.

Comparing the Refchat Index to the raw scoring data, the teams whose schedules are much easier using the Refchat Index are: St. Louis Rams (4th easiest rather than 11th easiest), Pittsburgh Steelers (3rd easiest rather than 7th easiest), and the Washington Redskins (17th easiest rather than 21st easiest).

The teams whose schedules look much harder when you apply the Refchat Index are: Cincinnati Bengals (21st easiest rather than 14th easiest), Detroit Lions (18th rather than 13th), and the Baltimore Ravens (9th rather than 4th).

So if you use the rough yardstick of how opponents did in 2008, then the Refchat Index suggests that the Rams, Steelers, and Redskins may do better than most people predict (although the Redskins still have a tough schedule) and that the Bengals, Lions, and Ravens may do worse than most people predict (although the Ravens still have an easier than average schedule).

Of course, a big problem with this analysis is that NFL teams do not stay consistent from year to year, but if you want an overview of the upcoming season, this is how the Refchat Index would influence the analysis.

Saturday, September 12, 2009

Predicting NFL Games: Visiting Team's Offenses Are Overrated

Although we focus on referee statistics, as long as we are logging data about the 2008 NFL season, we can look at other pieces of information, such as how to predict the results of NFL games regardless of what referee is covering the game.

Looking over the 2008 NFL season, we come up with a theory -- the strength of the visiting team's offense is overrated. You might take advantage of this by downgrading the importance of the visiting team's offense. If the visiting team has a great offense, that team might be a bit overrated. If the visiting team has a terrible offense, that team might be a bit underrated for that game.

Let's take a look at the 2008 NFL season. If we use points scored by the offense and points given up by the defense, it gives us a rough way to put a value on how good the team's offense and defense are. This rough statistic is not exact -- for example, it doesn't account for the strength of schedule -- but over the course of the 16-game season perhaps it evens out to a good degree.

We took a look at the correlation of the offense and the defense of the visiting teams and of the home teams to the final score and then to which team won (regardless of the scoring margin). In each case, we found that the biggest influence is the home team's offense. A bit further back is the strenth of each team's defense. And much further back is the strength of the visiting team's defense.

For example, in the 2008 NFL season, the correlation to the scoring margin was:
  • Home team's offense (0.375)
  • Home team's defense (0.276)
  • Visiting team's defense (0.207)
  • Visiting team's offense (0.128)
There is a similar disparity when you compare the teams' offense and defense to the game result (ignoring scoring margin).

So the new theory is: the Visiting Team's Offense Is Overrated.

And you can take advantage of this by adjusting the traditional predictions when the visiting team's offense is particularly strong or particularly weak.

We can use this theory to modify other analyses, such as strength of schedule. We'll do that in another posting.

Friday, September 04, 2009

Any Requests for the 2009 NFL Season?

Any requests or ideas for what to try to track for the upcoming 2009 NFL season for referees and their statistics?

With limited time, I can't track a ton of statistics, but I will try to track what I hope will be the most interesting and useful statistics for analysis. Let me know what ideas you have -- if I don't have time, maybe some other reader will pick up on your ideas.

Tracking referee statistics really becomes interesting during the regular season, not beforehand. The problem with looking at referee statistics while planning your fantasy football draft or projecting a team's season is that the NFL does not release the upcoming referee schedule for the year. So even if it turns out that your favorite running back will have an offense-favorable referee schedule, you won't know that when it's time to do your fantasy football draft or make your projections for the year.

When the season starts, I'll discuss some non-referee statistical analysis I've done on the past season, including the trend that the visiting team's offense is not as important for predicting the result of the game as other factors.