Statistics are a powerful tool in aiding fans and the media in their coverage of the NFL. But power often comes with trouble, and statistics can be dangerous when used incorrectly.
Improper stat usage can come in many ways.
There are the stats people place a lot of importance on but are actually not that critical for success.
Sometimes, conclusions are drawn based on a very limited sample size, such as a quarterback’s five pass attempts on fourth down in a season. Completing all of those five plays does not make him the “best quarterback on fourth down.” You need more evidence to support such a claim.
Then you have the ugly semantics battles, where a stat can be used incorrectly from its intended meaning or even given a different name altogether.
We will look at each type of bad stat usage. Here are six of the very most misunderstood NFL stats that could really use some better context.
We start with the stat everyone loves to talk about but no one actually loves.
The current passer rating system was adopted in 1973 by the league. Though it has some obvious flaws, it still does a decent job at picking out winners and providing a general barometer of a quarterback’s passing success.
The problem is the way passer rating gets abused by fans and the media. For starters, many like to call it “quarterback rating” when that clearly is not the title or purpose of passer rating.
A quick search on Twitter brings the results of an NFL team and beat writer calling it quarterback/QB rating.
If it was actually a quarterback rating, it would include all the plays a quarterback participates in, including sacks, runs and fumbles. It does not help when ESPN creates its own “Total Quarterback Rating (QBR)” stat, which only encourages people to also call them “quarterback ratings.”
Passer rating simply measures the passes a quarterback attempts without being sacked in a game, and the math is built around decade-old averages that probably could use an update as the stat is useless to compare quarterbacks across eras.
Jim Kelly has the fourth-highest passer rating (84.4) among quarterbacks in the Hall of Fame, yet only ranks 26th all time as we head into 2012. You have to do an era adjustment for passer rating to mean something.
The other factor working against passer rating is the way some broadcasters treat it like it is complex calculus when the formula simply involves addition, subtraction, multiplication and division.
If the four pillars of grade school mathematics are too much to handle, that says a lot about the state of education in our country.
Passer rating is not very complex, nor is it a stat that is all-encompassing for a quarterback’s performance. But it still has value if used appropriately.
Time of Possession
Of all the stats you can look at from a simple box score after the game, one of the last to pay attention to is time of possession.
While it sounds like something that would be really important, the truth is time of possession is just a result of doing a lot of good things throughout the game (getting first downs, sustaining offense, and quick stops on defense).
The actual possession of the ball is not important if you are not scoring any points. A long drive that ends in a punt or turnover looks good for your time of possession, but what did it actually do to your win probability?
No team sets a goal of possessing the ball for 35 minutes, but any team would love to set a goal of 35 points on the scoreboard. And they will take it any way they can get it.
Possessions must be productive but it really does not matter if they last 10 minutes or 50 seconds.
Defenses need rest? No, they need a stop. Then they can rest.
The best mockery of time of possession came in 2009 when the Indianapolis Colts played on the road in Miami on Monday Night Football. Despite having the ball for only 14:53, Peyton Manning led the Colts to 27 points and a comeback win.
The 14:53 is the lowest time of possession for a winning team since 1977, which is when they started keeping track of it.
All of the Dolphins' great efforts to possess the ball for 45:07 did not matter because they did not score enough touchdowns and points at the end of the game. Their final two drives lasted 9:34 and led to a field goal and game-ending interception by Chad Pennington.
Now one might say, “Yeah, but the 2011 Colts were dead last in time of possession, and they were the worst team in the league while the top two teams were Houston and Pittsburgh.”
Okay. the Chargers were third and they went 8-8. In 2011, only half of the teams that were in the top 10 time of possession had a winning record.
No surprise they were two of the best offenses in the league. They were productive when they had the ball. That is the goal.
Time of possession has no meat to it as a stat. While all the events that matter in a game are happening, someone is just keeping track of the clock. At the end of 60 minutes one team usually comes out a few minutes ahead of the other.
At the end of the day, it is all the yards, turnovers, red zone plays, third-down conversions, returns, kicks and penalties that decide the outcome. Not the clock.
The fact that time of possession became an official stat five years ahead of the sack is disheartening.
Rushing Yards Per Carry
There are a lot of myths still associated with the running game, especially in today’s NFL.
But one of the stats that should be a good indicator of effectiveness is rushing yards per carry (YPC). Passing yards per attempt works well for passing, so why doesn’t the mathematical equivalent for rushing work, too?
The problem is most runs in the NFL are gains of two to four yards, and the YPC number is inflated by those rare, long runs. A long run is great for an offense, but it can only help you on that one drive.
This is why rushing YPC has a poor correlation with winning (0.17 since 1970).
Since the merger, 24 teams have averaged over 5.0 YPC for the season. Out of the 24, 14 failed to make the playoffs, including nine of the top 13 teams.
Of the top 15 teams in rushing YPC, only the 1998 San Francisco 49ers won a playoff game. That of course was the game where Jerry Rice fumbled the ball on the final drive, but the referees missed it. Then Steve Young found Terrell Owens for the game-winning touchdown with three seconds left. Talk about a fortunate win.
The rushing stat that best correlates with winning is the number of carries themselves. The reason is simple: Teams with the lead run out the clock, piling up the carries and getting the win. Throw in a few kneel downs at the end for extra carry stat-padding.
You can have a great day running with over 5.0 or even 6.0 YPC but if you do that on 20 or fewer carries, your team only wins about 8.2 percent of the games.
How about averaging less than 3.0 YPC? That sounds like a horribly ineffective day on the ground. Since 1970, teams that carry the ball at least 30 times with less than 3.0 YPC have won 77.4 percent of those games. Plus, all 32 teams have a winning record in such games.
Still, experts, especially those that were former running backs, get this twisted all the time, thinking the carries cause the winning, when the opposite is true.
Perhaps the most useless stat in football is “RB X’s team is [insert great record] when he gets at least 20 carries.” Of course. It has nothing to do with the quality of that running back, but everything to do with the game situation.
I caught Mark Kriegel on the second episode of NFL A.M. using the stat for DeMarco Murray. Out of 194 running backs with at least 10 career games of 20 or more carries, 179 of them had a winning record. Murray’s not special.
A team's record with a running back logging at least 20 carries may be the only rushing stat more useless than rushing YPC.
Maybe some day a person will take the time to make some incredible punting metrics, but for now, the most ignored position in the game is judged heavily on gross punting average. The bigger the better, but that is not necessarily the goal of all punts.
If you are punting from the opponent 40, then ideally you want a punt of 39 yards to stop at the 1 and set your defense up in great position. Booming it into the end zone for a touchback is just going to net you 20 yards.
Also, when you punt from the 40, the most you can get is 40 yards. The offense that performed worse would have punted from farther back, giving the punter more room to blast one.
This is why so many of the punters with great averages—the ones that make it to Pro Bowls—are on bad teams with bad offenses.
Oakland’s Shane Lechler and San Francisco’s Andy Lee have been prolific punters in recent years, but a big part of it has to do with the putrid offense and record of their teams. Except last year, when San Francisco finally got back to winning with a 13-3 season.
Lee and Mat McBriar (2006 Dallas Cowboys) are the only two punters since 1994 to lead the league in punting average for a playoff team. They are also the only two from a winning team. The list includes just as many 1-15 teams (2000 San Diego Chargers, 2001 Carolina Panthers).
Donnie Jones joined the St. Louis Rams in 2007 and currently ranks seventh all time with a 45.3 yard career punting average. In that time, the Rams have gone 15-65 (.188), the worst five-year run in NFL history.
But at least they have a great punter?
While net punting average would be a better stat, it still does not account for the “available room” problem you face with a punter on a bad offense.
By the time we actually have a good punter metric, the position may be phased out by a maverick coach that revolutionizes the game with fourth-down gambles.
And to match how most people already feel about the punting position, no one will care they are gone.
Fourth Quarter Comebacks
I can always go thousands of words on this topic alone, but you can read a definitive three-year tale of the comeback conspiracy by the Denver Broncos here.
If you are new to it all, then the problem is simple.
For decades some teams have relied on their public relations to write about the fourth-quarter comebacks their popular quarterback has had in his career. This type of material appears in media guides and gets written and talked about by the media to spread the “clutch” legacy.
The problem is some teams, starting with Dallas (Roger Staubach) and Denver (John Elway) made a grave error in tallying the comebacks by counting games where the team never even trailed in the fourth quarter (counting games when they were tied as comebacks).
How can you come back when you are never behind in the first place?
That led to inflated totals of 23 fourth-quarter comebacks for Staubach, and 47 for John Elway, which was long considered the record.
But it was all bogus. Remove the games they never trailed, and Staubach has 15 comeback wins. Elway only has 34, which ties him for third all time. He never held the record. Dan Marino currently holds it with 36 and Peyton Manning has 35.
The Broncos helped shape Elway’s legacy by fabricating a record the league chose not to standardize. You can really see the mid-90s’ fabrication of the record by Denver here.
Want to know what Elway really did 47 times in his career? That represents the number of times he had the ball in the fourth quarter, down by one score, and failed to win the game (46 losses, one tie). Just as ironic, Staubach also had 23 failed comebacks.
The numbers that falsely represented these quarterback’s triumphs actually account for the number of times they (and their team) failed.
While Denver is the biggest offender, other teams have dabbled into the semantics mess by creating their own incorrect reports for their team’s quarterback. This is why I standardized the data and made it available on Pro-Football-Reference.com in 2009.
For a statistic to have any meaning, it must be tabulated by the same standard for everyone.
Things have improved on this front, but I still routinely find errors. Consider the case of Tom Brady. I could choose any player, but let’s just go with someone who may have a shot at breaking the record some day.
Brady has 25 fourth quarter comebacks and 35 game-winning drives. His total of fourth quarter/overtime wins could be completely different depending on which team he played for. Also, what they are actually called changes.
See the problem? Multiple teams should not be allowed to have their own definition because it all goes out to the same sources who do not know any better, and that is why there is so much inconsistency.
In the 2012 Patriots media guide, the 2011 AFC Championship comeback over Baltimore as Brady’s “seventh career fourth-quarter playoff comeback.”
That is false. They are falling into the Denver trap of counting everything as a comeback. Three of Brady’s seven game-winning drives in the playoffs were not comebacks, because the Patriots never trailed in the fourth quarter.
It is not an overly complicated stat to come up with for the true comebacks, but NFL history has botched it big time.
It took baseball writer Jerome Holtzman nine years to get Major League Baseball to make his save statistic for pitchers official. I have spent three years trying to make fourth-quarter comebacks and game-winning drives standardized and official and hope to do it quicker than Holtzman’s crusade.
Expect to hear a lot more about fourth-quarter comebacks this season when the career record is up for grabs starting in Week 1.
Finally, something we are experiencing right now: the overreaction to preseason statistics.
Throw out the records, the points, the stats, and even most of the performances. All that really counts in the preseason are injuries and filling out the last few spots on your roster that were mostly pieced together anyway.
Every year we wait six months for football to return but wish the preseason would end as soon as it begins. It is not an accurate representation of the product you will see on the field next month when things count and are real again.
Preseason football is not based on game-planning and playing starters for 60 minutes. It is more like the fantasy football game you play.
When someone tries to use the preseason stats as an indicator of how his or her team will do this season, I feel like going on an Alec Baldwin-sized rant from Glengarry Glen Ross.
Only with the meaningless preseason can a team like the 2005 Colts go 0-5, but start 13-0 in the regular season. Conversely, the 2008 Lions were 4-0 in the preseason and 0-16 in the regular season.
The preseason suffers just about every stat flaw. A small sample size made even smaller with different shifts and caliber of players coming in or sitting out the entire game. Vanilla offenses and defenses that try to show as little as possible.
Kevin Kolb struggled again Thursday night, but you did not need a preseason game to expect that. Even the famous “dress rehearsal” in the third week of the preseason is only slightly less meaningless than the other preseason games every year.
The only thing worse than preseason stats is a beat writer posting training camp stats on Twitter. Apparently that is a new trend in this social media-driven world, and it is one that needs to die a quick death.
Like drugs, just say no to training camp stats. September 5 cannot come fast enough.
That way we can get back to talking about how high a player’s quarterback rating was on their fourth-quarter comeback that really wasn’t a comeback, which was a long drive because of the booming punt by the opponent, but that helped raise the team’s time of possession in the end, which was also helped out by the running back’s long run to pad his YPC against the defense that went 3-1 in the preseason but is doing nothing when it counts.
Scott Kacsmar is a football writer and researcher who has contributed large quantities of data to Pro-Football-Reference.com, including the only standardized database of fourth quarter comebacks and game-winning drives. You can visit his blog for a complete writing archive, and can follow him on Twitter at @CaptainComeback.