Last week, we went 3-3 on our plays (all “Other Plays”). This has been a year with two big outliers. The Jaguars offense has been historically inept. The numbers are staggeringly awful. Jacksonville’s play success rate is only 24.8% (league average is 44.6%) and they average only 3.3 adj yards per play. The Jaguars don’t just have the worst historical MP Rating (going back to 2000) through week 4–they do, and it’s more than 3.5 points worse than the second-worst team–they have the worst rating of any team at any point in the season, and they’ve only played 4 games!
On the flip side, the Denver offense continues to be prolific, but they have not been as historically good as everyone seems to think. In terms of outcome, yes, no team has outscored the Broncos through four games. But at Massey-Peabody, we like to talk about process. Going back to 2000, Denver’s adjusted yards per play average of 7.41 ranks just 12th-best of NFL teams through four weeks, their 58.1% adjusted play success average ranks 4th, and their average offensive scoring efficiency ranks 3rd. Surprisingly, they’re only the 21st rank pass offense through four weeks. In fact, three teams last season had better passing efficiency numbers at this point. So are the Broncos the best team in the NFL? Undoubtedly. But is their offense the best in the last decade? Based on what we’ve seen so far, not even close. Is their team the best in the past decade? Looking at the historical Massey-Peabody ratings, definitely not. Through four weeks, the 2007 New England Patriots were a full two points superior and the 2011 Pats were a point better. And Massey-Peabody rates the Broncos’ offense only 6th best through four weeks, behind NE 2011, NE 2007, STL 2001, IND 2007, and NE 2012. The Broncos host the Jaguars in Week 6; right now we’d make the Broncos a 25.8-point favorite. There will be more discussion of the Broncos to come, but let’s move onto the picks.
As usual, injures (except to the QB) are not accounted for. Lines are widely available as of 1:30pm EDT on Wednesday. In cases where lines vary widely (this is generally on games lines between 7.5-9.5 and 1-2.5, and is due to differences in vig) I’m using the Bookmaker line.
Big Plays (3-1-1 YTD)
Tennessee +3 (-120) vs. Kansas City [MP= -2.0]
This line has been moving around a bit. It’s still a big play at +2.5. We’re still not buying the Chiefs. Yes, they’re 4-0. But their opponents are a combined 3-9 in their other games this year. We’re also not really buying the Titans either (more on that later). A main reason why we’re short the Chiefs is, surprisingly, due to their quarterback play. The way we rate quarterbacks is a bit complex, but we use a technique that allows us to estimate a QB’s value based on number of starts, years pro, draft position, etc. We call this the quarterback’s fixed effect. From here, it gets a bit more complex and messy. For previous seasons, we’re able to estimate the credit/blame for surplus performance (performance above/below opponent’s baseline + QB fixed effect) that’s assigned to the quarterback (the QB random effect) and to the team offense in general. The QB’s random effect is cumulative, and changes over the course of his career. Alex Smith is a little unusual; he looked to follow in the footsteps of fellow #1 overall pick David Carr, but since Jim Harbaugh took over in San Francisco, Smith has improved leaps and bounds. If we looked just at the last two-plus seasons, Alex Smith’s ratings would be much much better, but looking at his entire body of work, his rating is substantially worse. We rate him more than a point worse than an average starting QB.
Kansas City’s strength is their defense, especially the secondary, which has held opponents to just an average adj pass efficiency of just 5.7. Their offense is not explosive, but dinks and dunks its way down the field courtesy of an above-average 45.5% play success rate. The real reason the Chiefs are 4-0 is scoring efficiency. This captures how well a teams is able to convert yards into points on offense, and prevent yards from being converted into points on defense. It is the least important, and most noisy, of the stats the Massey-Peabody model uses. Kansas City has excelled in scoring efficiency on both sides of the ball (offense: .072, defense: .028, league average: .060). There is skill inherent in scoring efficiency–it is a vital part of our model–but while it’s probably the best stat at explaining game outcomes, it’s the least predictive of our metrics. We don’t think Kansas City will be able to keep over-performing their more fundamental numbers.
I won’t go into as much detail on the Titans, but suffice it to say they’ve also been overachievers. They have been good in scoring efficiency and rush defense, but little else. The Jake Locker injury makes this game a little more interesting from a handicapping perspective. It certainly hurts the Titans, but not as much as one might think. Ryan Fitzpatrick (as a backup) rates only 0.65 points per game worse than Locker. Kansas City is a better team, but only by about a half-point. The Titans should be a small favorite here, and I’ll gladly take Tennessee plus the points.
Houston +6.5 at San Francisco [MP= +3.5]
The market is generally at +6.5, but there are a few juiced +7s available. This line is a clear-cut reaction to outcome rather than the process behind it. Matt Schaub has thrown a pick-six in three straight games, and while the interception last week was a particularly egregious mistake, a mistake doesn’t normally cost a team more than a touchdown in expected points, as that one did. Drew Brees’s touchdown pass to Darren Sproles at the the end of the first half last Monday was a half-step from being a 98-yard interception return. While there is some skill in throwing and avoiding interceptions, there is a lot of randomness. Randomness can have a huge impact on a game, as these two examples show.
Houston has certainly underperformed in some fundamental areas this season, averaging only 5.91 adj yards/play, but San Francisco is even worse at 5.89. Houston has been more consistent in moving the ball, as their 46.2% play success rate bests SF’s 40.5%. Houston’s defense has been underrated, allowing only 5.04 adj yards/play, 35.7% play success, 6.05 yards/pass and 3.9 yards/rush (all stats are adjusted for context, so they won’t match up with raw box scores). Houston’s weakness has been in scoring efficiency, and while we do remove fumble luck from this metric, it can be unduly influenced by pick-6’s. Houston’s defensive scoring efficiency (.098) is far-and-away the worst in the league, while last year, they were around average. Expect some regression to the mean.
Other Plays (11-9 YTD)
Dallas +7.5 vs. Denver [MP= +4.5]
This line was 7 earlier but has moved to 7.5 at about half the big shops, and even +8.5 at a few. This is still a play at +7. Denver’s success, as mentioned earlier, has been blown out of proportion, in our view. Denver’s opponents are a combined 4-8 in games not against the Broncos, and that includes wins over the Redskins and Jaguars.
We’re a believer in Dallas. They have been superb in offensive play success, our most important stat predictively (52.0%), and have finally found a running game (4.7 yds per rush). They have struggled against the pass in Monte Kiffin’s new Tampa-2 defense–which Peyton Manning is intimately familiar with from his days with Tony Dungy in Indy. That is not a good omen when you’re facing a quarterback as good as Manning and a supporting cast as good as Denver’s, but the numbers say that Dallas at home should only be getting 4.5 points, not 7.
New York Jets +9.5 at Atlanta [MP= +6.7] (some +10s are out there)
Carolina -1.5 at Arizona [MP= -3.6]
New England +1.5 at Cincinnati [MP= -0.5]
(Note: This will not be included in our ledger: Green Bay is a play at -6.5 [MP= -8.7] but not at -7, where the market is at the time I’m typing this.)
