After reading about Al Harris of the Packers possibly being done for the season due to a likely ruptured spleen, I went and checked his and his teammate, Charles Woodson’s career stats. The article said that they form one of the most formidable cornerback (CB) tandems in the NFL. If you don’t watch the Packers regularly, it’s hard to verify except to compare stats with other CBs. However, when you look at the stat categories for a defensive back, it doesn’t show the effectiveness of the player. Deion Sanders, arguably the best cornerback ever in NFL history, has a statbox similar to many CBs and thus makes it difficult to conclude his greatness. Let’s look at ESPN and NFL.com’s stats of Charles Woodson’s career:
|ESPN – Charles Woodson Career Defense Stats|
NFL.com – Charles Woodson Career Defensive Stats
|2008||Green Bay Packers||3||7||7||0||0.0||0||6||2||1||41||20.5||41T|
|2007||Green Bay Packers||14||63||52||11||0.0||0||9||4||1||48||12.0||46T|
|2006||Green Bay Packers||16||59||48||11||1.0||0||12||8||1||61||7.6||23T|
The two sets of stats are essentially the same. We can rule out several of the categories as obviously being bad indicators of how well a secondary player plays – touchdowns (TDs), sacks (SCK) safeties (SFTY), forced fumbles and recoveries (FF, REC). As a matter of fact, most of those are in either ESPN’s or NFL’s and not both, which confirms them as bad indicators. So, let’s focus on the three stat groups that have traditionally been used to assess the efficiency of CBs (and even safeties) – tackles, interceptions, and passes defended (PD, PDef).
Tackles are a tricky statistic because they can mean two things. It might mean the receiver he was covering just made a big catch against him and he had to tackle him. It might mean a running back ran into the secondary and he made a great open field tackle. If he had a large number of tackles it could mean he is very active and covers a huge swath of the field, or… it could mean that he is a bad pass coverage guy and the opponents keep targeting him. Thus, it is difficult to really put any emphasis on any of the tackle categories; total, solo, and assisted.
Interceptions have a similar problem. A CB having no interceptions in a game gives no clues in whether he played a good or bad game. Interceptions also have a lot to do with luck and situation. It really doesn’t show the skill of a secondary back. Additionally, interceptions are rare and normalizing the statistic to per game would not solve the problem. Interceptions are a worse stat than sacks for a defensive back (DB) because interceptions can only occur if a pass is directed at his coverage area, whereas a DB or a defensive tackle rushes the quarterback a majority of the game.
Passes defended is a potentially good statistic. It shows that when a pass is directed at the CB, was he successful at preventing the pass from being completed by the offense. It shows a skill that he is legally (with no penalty) in the area and provided excellent coverage to defend the pass. So what is the problem? The problem is that it is not normalized. What does it mean when a cornerback defended two passes in a game? There is nothing to compare it to. It could mean two passes were directed at him and he knocked both down, for a 100% efficiency, or it could mean 10 passes came at him and he knocked only two down, for a 20% efficiency. Two completely different results. But what do you normalize it to?
NEW STATISTIC – PASSES DEFENDED EFFICIENCY
Just like an average and ERA in baseball, or a yards after catch (YAC) and rushing average in football, we need something that normalizes the stat against all other games and players. One way would be to divide the passes defended in a game by the number of passes directed at him in that same game. This would give us a pass defended efficiency average per game (PDE). Another option would be to divide the number of total passes defended in the season by the number of total passes directed at him in the season (PDE). The latter one is a better statistic because it would remove anomalies such as games in which only a few passes are directed at him and his PDE in that game is either 0% or 100%. However, what makes this stat difficult to record is determining the number of passes directed at a specific CB. In times when a CB plays zone or steps up to tackle or defend another player, it’s hard to say is the pass directed at him. Or if there are multiple defenders in the area, what do you do? Also, some passes are difficult to determine its intended target. Maybe this can be a statistic recorded for the secondary.
NEW STATISTIC – AVERAGE PASSES DIRECTED TOWARDS
Another statistic is to determine the percentage of passes in a game directed at a CB. In many cases a quarterback will target the CB that is most vulnerable. This would show up in this statistic. However, a quarterback will also target his best receiver regardless of who the CB is. That means a less than stellar CB on a less than stellar receiver will have a low percentage of passes directed at him. That does not mean he is a good CB.
NEW STATISTIC – AVERAGE CATCH YARDS AGAINST
Just like the running back has the average yards per rush, the wide receiver has the average yards per catch, and the quarterback has the average yards per pass attempt, we need a statistic for the defense. They already have average yards per rush against and possibly an average yards per catch, but it is for the whole defense and accounts for tight ends, running backs, and wide receivers. There needs to be a statistic for average yards per catch against each specific CB or at least for the whole secondary. As a fan, it’s sometimes very difficult to watch when a timid CB plays at least 10 yards off his receiver and gives up an easy catch and run.
The biggest problem is trying to normalize a statistic. With the way defensive schemes are used, many times it’s not just one person responsible for a catch but multiple players. However, that happens on offense also, such as determining whether a pass is the fault of the quarterback or its intended target. Quarterback stats drop even when they make perfect passes but the receiver drops the ball. What we realize though is that these stats usually evens out, and the best quarterbacks, running backs, and receivers have high numbers compared to worse QBs. A similar reliable statistic must be created and tracked in order to determine how well a cornerback plays.