This website is best viewed at 1024 x 768 resolution, full-screen.

Purpose and Concepts

PURPOSE: The purpose of this statistical analysis is to "explain" the statistical reasons why teams win. It is not, repeat not, designed to predict winners against the line. Bud does not believe anyone is good enough to beat the line.

Reason: There is too much game-to-game variation around the average performance. History shows that on important stats like intercepts, fumbles, and sacks, every team will have their best and worst effort in back-to-back weeks. But we don't know if these performances will be in games 1 and 2, 4 and 5, or 7 and 8. Without that knowledge we can't predict winning margins with accuracy.

"Explaining" the statistical reasons why teams win is different than "predicting." So this research becomes an intellectual exercise in comparing winning statistical profiles with losers.

"Point Value:" Each stat contributes some point value to the winning margin. Some of the most important stats and their points contributions are: 1) Intercepts on offense and defense add or subtract about 5 points per game to the margin; 2) fumbles lost and recovered are worth 2-to-4 points, depending on the year; and 3) a first down rushing is worth about 2 points while a first down passing is worth less than half that.

There are reasons: A first down rushing, for example, suggests a team is ahead on the scoreboard and stays on the ground to run the clock down and protect the lead; a first down passing suggests the team is behind and is in the air to catch the leader.

Counter intuitive: While the NFL lumps interceptions and fumbles together, one is more important than another. There have been seasons when the last ranked team on fumbles lost or recovered went to the Super Bowl.

Reason: Since running teams are generally ahead on the scoreboard (because they want to keep the clock rolling) and because there is a correlation between rushing plays and fumbles lost, the fumble (for a strong running team) may be considered a sign of strength. Counter intuitive? Right on.

Each week we will be adding some explanatory notes regarding the game, and we will tie these notes to statistical events from the recent games. Stay tuned.

To enjoy these notes you will want to keep in mind the definition of a "statistical correlation." In Paul Vogt's Dictionary of Statistics and Methods, he defines the correlation as follows: "Correlation: The extent to which two or more things are related ('co-related') to one another. This is usually expressed as a correlation coefficient."

OK. It simply means that two stats move up and down together from game to game. If you don't throw any intercepts on offense but steal 3 INTs on defense, there is a strong correlation with the winning margin. They go hand-in-hand. They are related, i.e., they are co-related; they are linked; they are associated.

So the bulk of this research is devoted to finding the stats linked to winning. For example, smoking and lung cancer are correlated. Your education is correlated to your annual income, so school dropouts lose educational chances and ding their future incomes. A television show's success in the ratings is correlated to the rating the show earned last week; it is also correlated to the success of the show before its time slot (which in TV slanguage is called the "lead in.") If a student wants to go to college, it is important to have a friend who went to college. If you have one college friend your chances of making it are doubled. That is a key correlation for kids from the center city.

In football there are big correlations between the following six factors and winning: 1) Passing Efficiency; 2) Clock control (running game/clock stats); 3) Field Position; 4) Interceptions; 5) Fumbles; and 6) QB Sacks on offense and defense.

There are a number of stats which correlate with these "ideas" or "factors." The "interception percentage" for example, is a measure of passing efficiency. The "sack percentage" is a stat correlated with the sack itself.

First downs rushing, rushing yards, and rushing TDs all correlate. They all help keep the clock rolling. So they are all clock control stats.

As the season progresses, you will be learning about the 6 clusters of stats that go to make up the 6 key factors of the game.

Stat methods are simply "ideas" about numbers. In addition to the idea of a correlation, there are three other ideas. First, the "average" is the single number that represents a team's ability in each game--the team's average performance.

For example, teams average about 4 yards per rush, but 6 yards per pass attempt (Dallas leads the league on defense allowing only 4.8 opponent yards per toss). That defensive stat is one reason they are winners.

Second, the "idea" of a "difference between averages." When QB Dan Marino joined the Dolphins, the average yards per pass jumped from 5+ to almost 7.

That is a humungous "difference" between averages. Since a change of 1-yard from a prior average is worth more than 3 points in the winning margin, this difference, with Marino at the helm, added a TD to the Dolphin's winning margin.

Finally, the third statistical "idea" is that of variation. Teams run hot and cold. Their game performances vary around their averages from week to week. They play over their heads, below their averages and near their averages. But they seldom play on their averages.

That is one ruling reason why it is impossible to beat the line.

Math and statistics teachers will find this weekly updated research helpful in motivating students. Sports are interesting to young people and motivate them to do work on their own. At a center city high school in Los Angeles, 90% Hispanic and 10% Asian, students have written semester projects built around sports statistics and the relationship of statistics in our daily decision making.

Is there a statistical "difference", for example, between buying a new car every 3 years as opposed to 10 years? Yes, indeedy. You can retire 3 years sooner if you are satisfied with a workable older car. Is that an important decision to make when you are first employed?