By Brian Burke
The efficiency gets at least a couple teams wrong each season. For some frustrating reason some teams insist on defying the laws of yardage and turnovers, winning far more than their expected share of games. The model is, of course, far from perfect, and we should expect it to err on occasion. And it's OK when it does because we can still learn a few things about whatever might be causing the disagreement.
This season the team that most confuses the model is the Arizona Cardinals. They've been winning based on aspects of performance that don't usually persist.
First, Arizona has enjoyed a great deal of success on special teams. They have missed only one field goal all season, and their other units have performed almost as well. But special teams performance is notoriously variable and inconsistent. There are six different special teams units, and they each only see the field a few times a game, which makes them difficult to assess. We often hear analysts refer to "small sample sizes" when judging a player from only a few plays or games. Well, special teams is all small samples, even after a full season.
Second, the Cardinals quarterbacks have over-performed. Both Carson Palmer and Drew Stanton have produced more win probability added than their play would indicate. This means they happen to have played their best in situations that mattered the most and saved their low points for games that were mostly already decided. Some might call this a talent for clutch play, but it is inconsistent in the extreme. A player who has over-performed in the past is just as likely to underperform in the future.
Lastly, the real strength of the Cardinals defense is its interception rate. They are second in the league in total interceptions and second in net turnovers. Analysts are fond of referring to "forcing interceptions." Although there is certainly a significant element of truth, interceptions are mostly a function of two things: the offense and luck. We can tell because interception rate "sticks" with offenses more than with defenses, but most of it varies randomly from game to game.
Each season, there's always one problem child, and this year it's Arizona. Either that or they're just plain good.
Last week, I had four out of the 12 games picked correctly off the spread (not including Thursday night's game). Will Leitch got three out of 12. Our totals so far are neck and neck: 72 for me, 70 for Will. Here are our picks for Week 12.
Note: The Machine's outcome probabilities are based on who will win the game, not the point spread. The score predictions are Burke's own estimates when factoring in the win probability, which is based on a logistic regression model of projected core team efficiency statistics. For more on Burke's model, click here.
JAGUARS: 13 (27% win probability)
COLTS: 33 (73%)
VEGAS: Colts (-14)
It might surprise some that Jacksonville has a slightly better defense than Indianapolis. But the Jaguars struggle to move the ball and have only a small shot at an upset.
Andrew Luck will take out some frustrations in this one.
BUCCANEERS: 17 (29%)
BEARS: 31 (71%)
VEGAS: Bears (-6)
The big difference in this game is each team's running games, otherwise they appear strikingly similar statistically. Chicago runs successfully 48 percent of the time (third best), Tampa only 35 percent (fourth worst).
And just like that, Jay Cutler is on a winning streak.
JETS: 10 (31%)
BILLS: 21 (69%)
VEGAS: Bills (-4.5)
This is a very bad matchup for the Jets. They can't pass and Buffalo is exceptionally good against the pass.
This game will be played from inside a Tauntaun carcass.
CHIEFS: 23 (69%)
RAIDERS: 13 (31%)
VEGAS: Chiefs (-7.5)
Kansas City's combination of low-risk, competent offense and strong defense will travel well.
Putting the Raiders in the Thursday night game is just doubling down on misery.
PACKERS: 35 (67%)
VIKINGS: 13 (33%)
VEGAS: Packers (-10)
Minnesota has a small advantage at stopping the run, otherwise it's all about the passing game. The Packers have hit their stride and the Vikings have the second-worst net passing efficiency in the league.
Remember from now on: When Aaron Rodgers tells you to chill ... chill.
CARDINALS: 20 (34%)
SEAHAWKS: 24 (66%)
VEGAS: Seahawks (-7)
The machine still sees the Cardinals as overrated. They're winning by takeaways, which explain past wins well but are the least consistent predictor of future wins. Seattle's running can control a game if they get a lead.
The Cardinals won in Seattle last season, but eventually, the breaks have to stop falling their way. This feels like the week.
BROWNS 23 (65%)
FALCONS 20 (35%)
VEGAS: Falcons (-3)
Atlanta started the year with a red hot passing game, but were hamstrung by a weak pass defense. Now that their passing game has fallen to earth, they are really going to struggle. Cleveland's defense has steadily improved throughout the season.
This reeks of one of those games that the Falcons lose on a dumb fumbled snap on a chipshot field goal. I'm sorry: The FIRST PLACE Falcons.
WASHINGTON: 16 (38%)
49ERS: 24 (62%)
VEGAS: 49ers (-9)
Washington is very odd. They've got good stats, but can't put together a "complete" game. Special teams is part of that, but they tend to fail on high-leverage plays -- 3rd downs and late drives. San Francisco's high sack rate is the difference between the Niners of the last couple seasons and this year.
Seriously, go read Dan Steinberg's amazing piece about what it's like to watch a game in Washington, and be happy this is in San Francisco.
TITANS: 21: (38%)
EAGLES: 30 (62%)
VEGAS: Eagles (-11)
These two teams are more similar than you might think. Tennessee's weak spot is against the run. Look for LeSean McCoy and Darren Sproles to take advantage early.
Mark Sanchez sure could use a confidence boost at this point, right? If the Eagles lose here, look out.
RAMS: 21 (38%)
CHARGERS: 27 (62%)
VEGAS: Chargers (-4.5)
San Diego has an advantage in every statistical dimension, even penalties, except running and stopping the run.
Very iffy about this prediction. Feels like the Rams could jump up and bite anyone right now.
DOLPHINS: 17 (40%)
BRONCOS: 21 (60%)
VEGAS: Broncos (-7.5)
A classic matchup of a great offense and great defense. The problem for Miami is that Denver's defense is also very strong.
If I'm Denver, this game is sort of terrifying. This could turn the wrong way fast.
LIONS: 21 (41%)
PATRIOTS: 27 (59%)
VEGAS: Patriots (-7.5)
New England's defense is vulnerable to the run, but they are now moving the ball with ease.
Don't sleep on the Lions here: That defense is terrific, and this feels like a trap game for New England.
RAVENS: 27 (41%)
SAINTS: 30 (59%)
VEGAS: Saints (-3.5)
These two teams appear as statistical mirror images. New Orleans has the more efficient offense, Baltimore has the better defense. Both teams' failures to generate interceptions are holding them back. The Saints have the edge thanks to home-field advantage.
It is legitimately possible that the NFC South winner will end up with six victories.
COWBOYS: 28 (56%)
GIANTS: 24 (44%)
VEGAS: Cowboys (-3.5)
Dallas has at least a slight edge across the board -- offense, defense, running, and passing.
Eli Manning should get his interceptions down to at least three in this game, and hey, that's progress.
BENGALS: 20 (46%)
TEXANS: 23 (54%)
VEGAS: Texans (-1)
The only apparent difference between these two teams is in defensive pass efficiency, where Cincinnati has a solid advantage. But it's not enough to overcome home-field advantage.
The real question is how many touchdowns J.J. Watt will throw.
Brian Burke is a former Navy fighter pilot and tactics expert. He is the founder of Advanced Football Analytics, the site that made modern football analysis a reality.