Last Year's Close Games Can Help Predict This Season's Records

In one of the Football Know It All's earliest posts, I discussed how year-to-year records have a faily weak correlation. One of the reasons for this is parity in the league. Turth be told, aside from the elite five or six teams and the five or sixe teams residing at the the bottom of the NFL barrel, there is not a tremendous amount of variance between the rest of the NFL teams.

This relative equality in the NFL leads to a large amount of Close Games, defined as games decided by 8 points or less. In fact, over the past five seasons, 48%, or roughly half, of the NFL's regular season's games have been Close Games. What decides the outcomes of these close games? There are many factors, and in fact, in my next post, I will talk about the kind of running game that helps teams tremendously in close games.

One factor, however, is random events. When teams are within a few points of each other, a number of factors, from a gust of wind, a penalty, or a dropped pass can affect the outcome. Supporting this assertion is that the standard deviation between NFL teams is significantly greater for overall record than for record in Close Games. This figure implies that in Close Games, the difference in the teams' qualities starts becoming less important while the impact of chance creeps in.

These things make victims of even the best teams in football. For instance, the Colts have had a 79% winning percentage over the past five seasons. However, in Close Games in that time, their winning pecetnage has been only 63%. Now, 63% is still very good, but it is substantially worse than 79%. One thing that happens is that, when they let teams stay close late into a game, they allow for the possibility of random events having more of an impact. The mark of good teams, however, has been that they rarely allow inferior teams to close in and make it a game.

The thrust of all of these figures is that random events do play a role in the outcomes of close games. An interesting notion to consider is that almost all teams play between 7-9 Close Games per year. So, each year, the random events that occur in close games can have a small, yet noticeable, impact on a team's record. As we've mentioned before, random events even out over time. It has been demonstrated that a team that loses a disproportionate number of Close Games one year will often improve in its Close Games record, and often in its record overall.

Thus, teams that lost a disproportionate amount of Close Games in 2007 may, by the simple power of probabilities, see an improvement this year. To keep an eye on this trend, here are a collection of teams that lost a large percentage of Close Games last year:

Pittsburgh Steelers
San Diego Chargers
Indianapolis Colts
Minnesota Vikings
Seattle Seahawks

Some teams that fared better in Close Games than they did overall and thus may not fare as well this year include:

Carolina Panthers
Denver Broncos
Buffalo Bills
Tennessee Titans
San Francisco 49ers

Remember, though- Just because random events take more of a role in close games does not mean that they are the sole predictors. Far from it. Even in close games, the odds still overwhelmingly favor the teams that are designed to win such games. Next post, I'll discuss what type of running game is needed to improve a team's chances in close games.