By Brian Burke
With three weeks of games the family portrait is quickly coming into focus, and we can get a reasonably accurate picture of how teams are going to look this year. At this point it's important to keep two things in mind. First, teams with very good or very bad performance to date will likely regress, meaning that the tendency for any measure of performance is to gravitate toward the average. Put simply, take any individual or team stat with a heavy grain of salt. Second, opponent strength is especially important this early in the season. Over a three-game stretch it's not unusual for some teams to have a much softer or harder schedule than others, and the first three weeks of the season are no exception.
There's an unusually high number of evenly matched games this week, so prediction accuracy will largely be a dice roll. But it should make for a fun Sunday.
Last week, I had seven out of the 15 games picked correctly off the spread (no pick for the Thursday night contest). Will Leitch got nine out of 15 games. Our totals so far are neck-and-neck: 18 for me, 21 for Will. Here are our picks for Week 4.
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.
BUCCANEERS: 13 (27% win probability)
STEELERS: 35 (73%)
VEGAS: Steelers (-8)
Tampa Bay can't pass the ball and they can't stop the pass. Not a good combination to be taking on the road.
Did anyone see the Bucs being quite THIS bad? Maybe I need to go through that infamous #longread again.
JAGUARS: 17 (30%)
CHARGERS: 38 (70%)
VEGAS: Chargers (-14)
Of all Jacksonville's deficiencies, their poor running ability stands out as the weakest. San Diego is struggling to run the ball too, but they can throw it all day.
Don't look now, but schedule sets up for San Diego to be 6-1 heading into Denver on Oct. 23.
TITANS: 20 (35%)
COLTS: 31 (65%)
VEGAS: Colts (-7.5)
The stats really like the Titans. Although they've thrown a bunch of picks, they have forced just as many. With one or two notable exceptions, their pass defense has played very well. Indy's pass defense is very vulnerable, but with Locker out and Whitehurst in Indy has the advantage.
It still feels like the Colts are going to win this division by four games.
DOLPHINS: 20 (58%)
RAIDERS: 13 (42%)
VEGAS: Dolphins (-4)
Miami is averaging 4.3 net yards per attempt on offense. That's really, really bad. They'd actually have to improve a lot just to get back to really bad. Fortunately for them, they can run the ball … and they're playing the Raiders who won't have the benefit of home field in London.
This week's designated "Red Zone Channel keeps forgetting it's still going on" game.
PANTHERS: 24 (42%)
RAVENS: 27 (58%)
VEGAS: Ravens (-3)
Baltimore's defense might be the weakest it's been in 15 years. They own the league's third-worst net passing efficiency. But overall these are statistically even teams, so Carolina is a slight underdog on the road.
The way it usually works is that Cam Newton inexplicably has a monster game right after a stinker like last Sunday night.
FALCONS: 31 (57%)
VIKINGS: 21 (43%)
VEGAS: Falcons (-3)
It's hard to know what to expect with Teddy Bridgewater in the mix now for Minnesota. He might fair well against a below-average Atlanta defense, but the Viking's defense will be very lucky to slow down the Falcon's passing attack.
At least Bridgewater will be fun to watch in defeat. (Looking your way, Cleveland.)
EAGLES: 24 (44%)
49ERS: 30 (56%)
VEGAS: 49ers (-5)
Penalties are a big problem for San Francisco. They're averaging nearly one full penalty yard per snap. (League average is around 0.4). Philadelphia's running game is a shadow of its 2013 self.
The Eagles are better right now, but the 49ers, understandably, are desperate. They wouldn't dare christen their fancy new stadium with two losses, would they?
BILLS: 17 (44%)
TEXANS: 23 (56%)
VEGAS: Texans (-3)
Houston looks solid but is statistically perplexing. They're a very high on the risk-reward spectrum, throwing for lots of yards but getting picked off on one out of 25 drop backs. The Bills look like a competent but not exceptionally good team so far.
The Bills bandwagon emptied, like, super fast.
SAINTS: 27 (53%)
COWBOYS: 24 (47%)
VEGAS: Saints (-3)
The Saints are better than their record shows. They have yet to grab an interception on defense, something driven largely, but not completely, by luck. I expect their fortunes to change versus the turnover machine that is the 2014 Cowboys.
Time for the first national television public depantsing and pile on of Tony Romo. How did it take this long?
PACKERS: 24 (48%)
BEARS: 26 (52%)
VEGAS: Packers (-1.5)
Both offenses have been underperforming. Despite a depleted secondary, Chicago has been the beneficiary of a heap of interceptions. Green Bay looks like a slightly better team but Chicago gets the edge at home.
It's sort of disappointing when these particular franchises have teams that score a lot of points, and also when they play when it's not that cold outside.
PATRIOTS: 16 (48%)
CHIEFS: 17 (52%)
VEGAS: Patriots (-3.5)
The Patriots might be the biggest surprise all year. Although slow starts are nothing new for Brady's offense, 4.8 net yards per attempt so far is woeful, especially against some weaker opponents.
If the Royals and Tigers end up tied for the AL Central title, they'll be playing -- alas, in Detroit -- at the exact same time of this game.
LIONS: 21 (51%)
JETS: 20 (49%)
VEGAS: Lions (-1.5)
Both teams have strong defenses, but Detroit has the better passing attack. The Lions may have cured their penalty problems that dogged them in recent seasons.
This feels like one of those games that leads, inexorably, to Rex Ryan being fantastic on NBC next year.
Editor's Note: Both the Man and the Machine incorrectly picked the Thursday night game between the Giants and Redskins.
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.