Sorting through the NFL's stats explosion
In the NFL's burgeoning statistical age, Dom Capers still has a quick, relatively easy way to gauge NFL teams.
It's passer-rating differential, or the difference between a team's passer rating and its opponents' passer rating.
In a quarterback-driven game, it's a great way to measure what most often determines winning and losing: How efficiently a team passes the ball relative to how difficult it makes life for the opposing quarterback.
It has a way of cutting through statistical noise because the most important parts of the game — big-play capability, quarterback decision making and throwing accuracy, pass-rush pressure and coverage — all find their way into the number, some way, somehow.
"I think about when the Packers won the Super Bowl (in the 1996 season) and beat the (Carolina) Panthers out here, they were plus-41 or -42 that year," said Capers, who was Carolina's coach in that game and has been the Packers' defensive coordinator since 2009.
"They were No. 1 in (passer-rating) offense, No. 1 in (passer-rating) defense, which rarely happens. If you have a guy like Brett Favre, you had a high quarterback rating, and their defense was hard on opposing quarterbacks. Look at when we won it here (in the 2010 season). If we weren't the top (in differential), we were one of the top two."
The Packers in fact were No. 1 in the NFL in passer differential in '10 at plus-31.7.
Capers didn't invent passer-rating differential, and to NFL stat heads it's nothing new. The website Cold Hard Football Facts has been compiling it weekly and touting it as a superior indicator of team quality since 2009. According to a study by CHFF, 45 of the 74 NFL champions (61 percent) since the 1940 season finished either first or second in passer-rating differential, and all but four champs finished in the top 10.
Still, as interesting as it is, it's just one simple way of evaluating NFL teams in a statistical age that is just starting to explode among clubs and the public at large.
NFL teams have been working with analytics for a while now — the Packers hired Mike Eayrs as director of research and development in 2001 — but the level of sophistication is just beginning to take off. Aside from the work teams do in their research departments, there now are league-wide databases available to fans and teams alike at websites such as Pro Football Focus, Football Outsiders, Advanced Football Analytics, Stats and CHFF, among others.
"In football, we're still learning to count," said Mike Tanier, a blogger for bleacherreport.com who previously worked for 10 years at Football Outsiders. "One of the first things that Pro Football Focus and Football Outsiders are doing is actually counting and disseminating events that weren't counted for decades and decades."
Tanier is talking about statistics such as receiver targets and missed tackles, which in the past couple of seasons have become almost mainstream even though the NFL doesn't compile them. But the analytics go much, much deeper than that, especially among teams but even among outside entities.
The different companies have their own methods of compiling and using data, and thus different strengths. The information available from each of the aforementioned websites is enormous, and combined it's overwhelming for all but the most statistically obsessed fans, and of course teams.
Pro Football Focus, at least for fans and media, specializes in grading individuals on every snap of every game. According to a company spokesman, analysts spend a combined 40 hours on each game tabulating and cross-checking their data, which range from pass targets and drops, to missed tackles, to sacks and pressures allowed; to more advanced stats for teams such as offensive and defensive personnel groupings and formations, blitzes, coverages and pass routes. The company has 350 data points for each play.
The entire database is available to the 13 teams that subscribe to full service. They use it primarily to augment advance scouting of opponent tendencies, though they also can do database studies of their own. For instance, a team can take a player trait, such as height, weight, arm length or hand size, and see if it correlates with performance in a specific category, such as sacks allowed, dropped passes, fumbles or the like.
"They also have it linked to their video system," said Bryan Hall, the PFF spokesman. "Say I want to see all times a team is in (one receiver, two tight end and two running back) personnel, at this down-and-distance, and other variables. (The PFF stat) says they're going to pass X amount, run X amount, in this or that direction. That's not the public side of what we do, but that's how teams are using us."
Football Outsiders and Cold Hard Football Facts, on the other hand, don't do as much individual grading, but rely heavily on proprietary algorithms based on accumulated play-by-play data. Grades are based on how much the result of a given play helps or hurts a team's chances of winning, with the historical data incorporating game situation (down-and-distance, score, etc.) and opponent. The grade for a player, unit or team for a game or season is called DVOA (don't ask what that stands for).
"Almost every conventional stat has a caveat," Tanier said. "If you look at a cornerback and see 18 passes defensed, first of all he might be getting picked on, and second we've discovered with passes defensed, different teams' scorers do them different ways. The ball will be 10 yards over the guy's head and the team will give him a pass defensed. If you're a stat savvy fan you have to start with Football Outsiders and Pro Football Focus and kind of learn from there."
If the data available to fans from these web sites collectively is immense, the volume of data available to teams is preposterous and growing. Along with Eayrs, the Packers this year hired an analytics intern, so their research department is growing.
The main purpose of teams' analytics departments is to convert the exploding amount of raw data into something manageable for the coaches, who in turn reduce it to something manageable for players. Packers assistant coaches also regularly ask Eayrs to research projects for them, which can be as simple as comparing the number of checkdown passes their quarterbacks threw in team drills in training camp in 2013 compared to this year.
"You can get overloaded on that stuff," Capers said of all the data available.
So can even the most serious stats fan. While there's real value in all the statistical analysis available online, it's also easy to get lost in the morass of grades, rankings and blogs.
But it's also true that most conventional NFL stats are limited in value at best, and extremely misleading at worst. Yards per rush, for instance, is an OK indicator of a running back's effectiveness, but it has serious limitations. Total tackles by a player, especially for linebackers, is almost useless, because it tells nothing about where on the field the tackles occur.
There are limitations in football analytics compared to baseball for obvious reason. There's far more objective data in baseball, and the standard data goes back to the early years of the professional game. Football has more interdependent parts, and much of the data (what counts as a good block?) is far more subjective. Still, teams are delving deeper into analytics, and subscribing to these new services, for a reason.
But the excess of data for interested but not obsessed fans can be too much. That's why there's beauty in passer-rating differential: It's relatively simple yet informative.
In that light, the graphic accompanying this column offers another quick and dirty stat for ranking offenses and defenses in the NFL.
The standard measure is yards. If a team is called the top-rated offense (or defense) in the league, that means it leads the league in total yards (or fewest yards allowed). But that's not a particularly good measure of how good an offense or defense is. For instance, teams can put up big yardage numbers in the second halves of blowout losses, but that probably doesn't say much about the quality of its offense or the opponent's defense.
Points are a better measure but still can leave out a big part of the story. Maybe a team has a good defense that sets up a mediocre offense for a lot of easy scores. Or an offense moves the ball up and down the field but has had some bad breaks or fluky turnovers in the red zone that lower its scoring. The ability to move the ball has to count for something.
An easy way to mitigate extenuating factors is to steal an idea from Rick Gosselin of The Dallas Morning News. After every season, Gosselin combines every team's rankings in the 20-plus special teams categories to calculate a composite special teams ranking.
Along the same lines, but in simpler form, there's value in combining teams' rankings in points and yards to better measure how they truly rate on either side of the ball. So, for instance, this week the Packers' offense ranks No. 28 in yards but No. 7 in points. Those rankings add up to 35, which is 19th best in the league.
It's still too early in the season to read too much into any statistic — the Packers at 3-2 also have the league's best passer-rating differential at plus-40.1. But the combined scoring and yards rating is a little more, or should I say less, food for thought.
Check the accompanying graphic to see the league's combined rankings.
— firstname.lastname@example.org and follow him on Twitter @PeteDougherty
WHERE THEY RANK
Following are NFL offensive and defensive rankings this week based on combined rankings in yards and scoring. For instance, Indianapolis' offense ranks No. 2 in yards and No. 1 in scoring, which totals 3 rankings points. The lower the total, the higher the combined ranking. The number in parentheses is the team's ranking points.
1. Indianapolis (3)
2. Atlanta (6)
3. Denver (10)
4. (tie) Dallas, New Orleans (11)
6. Philadelphia (15)
7. Seattle (17)
8. San Diego (19)
9. Cleveland (20)
10. Pittsburgh (22)
11. (tie) Baltimore, Cincinnati, New York Giants, Washington (26)
15. Miami (29)
16. St. Louis (30)
17. (tie) New England, Chicago (33)
19. Green Bay (35)
20. San Francisco (36)
21. Kansas City (39)
22. Houston (45)
23. Detroit (46)
24. (tie) Tennessee, Carolina (48)
26. Minnesota (49)
27. Arizona (50)
28. Buffalo (54)
29. Tampa Bay (55)
30. New York Jets (57)
31. Jacksonville (62)
32. Oakland (64)
1. Detroit (3)
2. San Diego (4)
3. Buffalo (8)
4. San Francisco (12)
5. Seattle (14)
6. New England (16)
7. Kansas City (18)
8. Baltimore (19)
9. Pittsburgh (23)
10. (tie) Indianapolis, Miami (26)
12. New York Jets (27)
13. Dallas (29)
14. (tie) Denver, Houston (30)
16. Green Bay (32)
17. Cincinnati (33)
18. (tie) New York Giants, Minnesota (35)
20. Washington (36)
21. Arizona (37)
22. Carolina (38)
23. Chicago (40)
24. Oakland (41)
25. St. Louis (43)
26. Tennessee (50)
27. (tie) New Orleans, Philadelphia (53)
29. Cleveland (55)
30. Atlanta (58)
31. Tampa Bay (61)
32. Jacksonville (64)