Barnwell: Stacking surprises on the NFL's seven 3-0 teams

Pats' win still had some teachable moments (0:51)

Tom Jackson says he wouldn't want to be in Monday's film session for the Patriots due to some errors that popped up on offense and special teams. To watch NFL Primetime, sign up here for ESPN+ http://plus.espn.com/. (0:51)

Last week, I evaluated the NFL's 0-2 teams and tried to figure out which of them had the best shot of making it to the postseason. Let's change gears and look at the most successful teams in football.

The league has seven 3-0 teams after the Rams won on Sunday night, and they should all feel good about their playoff chances. Since 2002, just under 70% of the teams that started 3-0 were able to parlay their early-season success into postseason football.

Instead of trying to identify the teams that are most likely to miss out on playoff football, though, I want to go in a different direction. Even though we've only had three weeks of action, I've seen enough in some cases to challenge -- or flat-out reject -- the notions I held heading into the season. Players and teams who I thought might start slowly or struggle have impressed. In other cases, I feel more confident about the ideas I held heading into the campaign.

So, for each of the seven 3-0 teams, I've gone through and identified something that has surprised me and another thing that hasn't really been shocking through three games. (There's one exception below.) I've sorted through these teams from the most surprising 3-0 start to the league's least surprising, which means this list begins and ends in the same division:

Jump to a team:
KC | LAR | NE | SF

7. Buffalo Bills

Not surprising: This is the best defense in football.

Most people should realize by now that the Bills have a good defense, but that has undersold the story for a while. In 2017, Sean McDermott took over what had been the league's 27th-ranked defense by DVOA under Rex Ryan and immediately pushed it to 15th. While the Bills failed to return to the playoffs and took a step backward in 2018, the defense wasn't the problem. Leslie Frazier's defense allowed 47 points to the Ravens in last year's season opener and then 31 points to the Chargers in a game most famous for Vontae Davis retiring at halftime, but the Bills have been a dominant defense ever since.

Consider that even with those two dismal games to start the season, the Bills finished second in DVOA last season, ahead of well-regarded powerhouses like the Ravens and Vikings and only behind the Bears, who were a takeaway factory in 2018. Over their last 17 games -- from Week 3 of 2018 on -- here's where the Bills rank in a few key defensive rate statistics:

No unit has things harder than the Bills' defense, which faces a ton of drives and inherits terrible starting field position from an offense that turns over the ball too frequently. The Bills have gone up against 29 drives beginning on their own side of the field, almost always after an offensive takeaway. To contrast, the Patriots -- who rank just ahead of the Bills in points allowed per drive -- have faced 14 such drives over that time frame. The Chiefs have faced just five.

To be right up there with the likes of the Bears and Patriots is downright magical. We saw another example of Buffalo's efforts on Sunday, when it held the Bengals to six punts and three turnovers on their first nine drives. If we define short fields as drives beginning with 65 yards or less to go for a touchdown, the Bills faced four short possessions Sunday. They allowed a total of seven points on those four drives, with those coming after a Josh Allen interception gave the Bengals the ball on Buffalo's 22-yard line. (One of those possessions, to be fair, came with two seconds left at the end of the first half.)

The Bills sealed the victory with Tre'Davious White's second interception of the day, which matched his total from 2018. While the best cornerback in football discussion often includes some combination of Jalen Ramsey and former Bills star Stephon Gilmore, White absolutely belongs there. His second interception involved catching a deflection, but both picks required incredible hands.

White's takeaway numbers aren't staggering because opposing offenses know he's a star and stay away. He has been targeted just 14.1% of the time since the start of 2018, per the NFL's Next Gen Stats; only Richard Sherman, Casey Hayward, and William Jackson have been targeted less frequently by opposing quarterbacks. McDermott has also been able to use White as a press corner on nearly 47% of his targets, which is remarkable for a corner who is listed at 5-foot-11 and 176 pounds. Only seven other corners in the league were in press coverage more frequently on their targets.

It's a testament to the coaching in Buffalo to see just how many players have come to town and improved on their established level of play. Castoffs like Jordan Phillips and Kevin Johnson have vital roles in this defense. Micah Hyde and Jordan Poyer were both converted cornerbacks with short track records as starters when the Bills signed them to take over as their starting safeties in the spring of 2017. Now, they're one of the best safety duos in football, both for their ability as ballhawks and their ability to eliminate big plays. The Bills have allowed just nine pass plays of 30 yards or more since the start of 2018, four fewer than any other team in football.

They are stout at all levels of defense. They don't make stupid mistakes, and they take advantage when offenses make mistakes. It would be easy to play into the small-market story about how the Bills have a no-name defense and like it that way, but put that aside. This is a unit every bit as good as the teams we laud for great defense, and while they haven't played an incredibly difficult slate of opposing offenses this season, no defense in the league has had to shoulder a larger portion of its team's workload over the past year than the Bills.

Surprising: Josh Allen's lack of deep passes

The book on Josh Allen coming out of Wyoming was pretty simple. Quarterbacks need strong arms to succeed, and he had one of the strongest arms coming out of college since Matthew Stafford nearly a decade earlier. The Bills obviously loved Allen's intangibles and clearly valued his running ability, which kept him afloat at times during an uneven rookie season, but even the most sympathetic Allen backer would admit that accuracy was his biggest problem.

During that rookie season, he completed just 52.8% of his passes. That number would have been fine in the 1970s, but after you adjust for era and look at passers with 300 attempts or more, it was the 11th-worst completion percentage since the 1970 merger. He didn't have great receivers, but his pass map was also something out of the '70s. An even 20% of his passes traveled 20 yards or more downfield. To put that in context, no other starting quarterback topped 15% last season, and the only other quarterback over the past decade to top 20% in a season was Tim Tebow. (Tyrod Taylor, Allen's predecessor in Buffalo, was previously the one behind Tebow at 18.4%.)

The Bills came out of training camp preaching accuracy as the most important trait for him to master in 2019. His completion percentage is up to 64.1% through three games, and while he has shown more consistent mechanics and been more aggressive with looking upfield while scrambling as opposed to putting his head down to run into space, the Bills have also made Allen's life easier. Through three weeks, just 14.6% of his passes have traveled 20 yards or more in the air. That ranks 13th in the NFL.

The NFL's Next Gen Stats do the best job of putting this in context. In 2018, a typical quarterback given Allen's range of throws and receivers would have completed 60.5% of his passes, which ranked 28th out of 30 quarterbacks. Allen then completed 52.8% of his throws, with the resulting difference of 7.7 percentage points ranking as the worst in the league. Allen was given a tough slate of passes and didn't do a good job with them.

Through three weeks, though, Allen's expected completion percentage is 63.1%, suggesting he's been given a much easier range of throws. His actual percentage is 64.1%, which is slightly above that expectation. The second-year passer has still turned the ball over too frequently and hasn't yet fully gotten past his habit of making breathtakingly bad decisions when throwing on the run, but he has grown as a passer. He has also been given easier throws to make.

6. San Francisco 49ers

Not surprising: The running game is going even without big-name backs.

The Mike Shanahan offense has been turning mid-round picks and little-known backs into stars for more than two decades now. You know the names. Terrell Davis. Mike Anderson. Alfred Morris. Arian Foster. Even in Atlanta, Devonta Freeman was far more productive under Kyle Shanahan than he was before or after the offensive coordinator left for San Francisco.

All of this made it more surprising that the 49ers seemed to focus on acquiring running backs lately. During John Lynch's first draft as general manager, Kyle Shanahan reportedly beat the table for the Niners to move up and grab Joe Williams, who was cut without ever playing for the team. (The Colts, who traded down with the 49ers as part of that deal, took Marlon Mack with the selection they got from the 49ers.)

Over the past two offseasons, Shanahan has dived into free agency. He gave Jerick McKinnon a four-year, $30 million deal before the 2018 season, only for McKinnon to tear his ACL in camp and aggravate the injury this summer. The Niners will likely pay McKinnon $16 million without him ever taking a regular-season snap for the team. They also added Tevin Coleman on a one-year, $5 million deal this spring, but Coleman suffered a high ankle sprain in the opener and is likely to miss about a month of action.

The 49ers, you might have noticed, have not missed a beat. After running the ball 40 times for 168 yards and two scores against the Steelers on Sunday, San Francisco's backs have carried the ball 114 times for 525 yards. The Niners are fourth in rushing yards and 12th in rushing average, and 43% of their runs have improved their offense's chances of scoring on the drive in question, which ranks 14th in the league. It's not exactly Davis or Foster, but this is a comfortable improvement for a team that ranked last in rushing DVOA in 2018.

Shanahan has built a useful running back rotation out of his third, fourth and fifth options. Matt Breida averaged a gaudy 5.3 yards per carry last season, but he ranked 30th out of 47 backs in Football Outsiders' Success Rate statistic, which measures how reliably a back keeps his offense on schedule. Breida was at 46% last season; through three weeks this season, he has been successful on nearly 59% of his carries.

Raheem Mostert, a fellow undrafted free agent who was signed to a small extension and expected to contribute on special teams in 2019, has turned his 34 carries into 202 yards. Most teams don't have a goal-line specialist back these days, but the 49ers signed back Jeff Wilson onto the roster from their practice squad and have turned their goal-line carries over to him. Wilson -- a third undrafted free agent -- has four touchdowns on eight attempts inside the 10-yard line.

It was promising to see the 49ers run the ball effectively against a Steelers front that looks good on paper and was 10th in rushing DVOA heading into the game. The Niners were able to get by in the running game without star left tackle Joe Staley, who will miss six to eight weeks with a fractured fibula. After their bye, Shanahan & Co. will get a matchup with the scuffling Browns before a Week 6 game with the Rams. The winner will likely be in first place in the NFC West afterward.

Surprising: The 49ers are second in the league in interceptions.

When I suggested that the 49ers were likely to improve in 2019, I pointed out one factor that was extremely unlikely to reoccur. The 49ers racked up just two interceptions in 2018. All season. Two. Unsurprisingly, no team in NFL history had ever failed to intercept at least three passes over a full campaign before the 2018 49ers.

Now, I write a lot about teams regressing to the mean. Sixteen games just isn't a big sample, and so when a team does something at an egregiously high or low rate relative to the rest of the league, it's usually a product of that compressed season. History told us that the 49ers weren't likely to intercept one out of every 271 passes again this season. The average team intercepted just over 13 passes last season, and that would have been the simplest projection for the 49ers in 2019.


Should Chiefs' defense be a concern?

Chris Berman and Tom Jackson give their takeaways from the Chiefs' win over the Ravens. To watch NFL PrimeTime, sign up here for ESPN+ http://plus.espn.com/.

The 49ers weren't "due" to intercept some crazy number of passes just because they barely intercepted any last season. That's what's known as the "gambler's fallacy." And yet, through three games, the 49ers are like that roulette player who bet all of his money on black because the last five numbers were all red, and he hit anyway. They have five interceptions in their three victories, which is second in the league behind the Patriots. The Niners took two of those interceptions -- both picks of Jameis Winston -- to the house for scores.

Is there some brilliant strategy they have suddenly employed to create interceptions? No. The defense is creating more pressure after adding Dee Ford and Nick Bosa this offseason, and pressure helped create this terrible throw from Mason Rudolph, but the 49ers have only jumped from 20th a year ago to 13th in pressure rate.

Instead, they've caught the sort of breaks they didn't catch a year ago. In the Bucs game, O.J. Howard had a pass bounce off of his hands for one interception. A miscommunication on a Peyton Barber hot route led to a Sherman pick-six, and a terrible decision to force a screen pass from Winston led to another 49ers touchdown. Last year, those plays might have resulted in incompletions or dropped picks. That's just bad luck. The only interception that's materially different is Kwon Alexander's interception of Andy Dalton; while Dalton was pressured, Alexander showed uncommon range to track Tyler Eifert all the way to the sideline for a pick. (Alexander also dropped a would-be pick earlier on the drive.)

This season, the 49ers have picked off one out of every 21 passes they've seen as a defense. I would not count on that continuing, either. Robert Saleh's defense looks much improved after the offseason additions and a return to health for several key contributors, but the team wasn't due for a run on interceptions to start the year. The best projection for interceptions over the rest of their season would be to, well, regress toward the mean.

5. Green Bay Packers

Not surprising: The new coaching staff hasn't turned Aaron Jones into a bell cow back.

One of the many reasons Packers fans grew frustrated with Mike McCarthy before the longtime coach was fired last season was his usage of Jones. After years of cycling through disappointing running backs, the Packers finally seemed to stumble on something exciting with their 2017 fifth-round pick out of UTEP.

Jones averaged 5.5 yards per carry as a rookie and ran for 346 yards over a four-game stretch early in the season, but after a 131-yard performance against the Saints in Week 7, he carried the ball just 19 times over the remainder of the season amid injuries and pass-protection problems. He was expected to be the guy after returning from a two-game suspension last season, but he averaged just 13.3 touches per game for an offense that sorely needed someone to take some of the load off Aaron Rodgers.

New staff. New scheme. This is Aaron Jones' year. Right? He cut his body fat in half during the offseason to stay in better shape. In Week 2, he runs the ball a career-high 23 times and racks up 150 yards from scrimmage in a 21-16 win over the division rival Vikings. Afterward, new coach Matt LaFleur responds to the big performance by saying he wants to even up the touches between Jones and Jamaal Williams? And then he gives Williams more carries (12) and snaps (32) than Jones (10 and 20, respectively) in Sunday's win over the Broncos?

I'm not surprised, though I don't think it's optimal. For one, while LaFleur isn't the same as McCarthy, he inherited a situation in which the organization thought of Jones as a part-time back. There are times when a new coach comes in and frees a back who had been stifled by putting him in a much larger role, but more often than not, new coaches come in and use those backs in the same way. Think about guys like Kenyan Drake, Duke Johnson and even David Johnson's role as a receiving back with the Cardinals in 2018. New coaches and coordinators often come in and use the backs they inherited the same way the old coaches did.

Coaches also generally establish their preferred back usage in one job and then continue on with that pattern in their next stop. LaFleur was the offensive coordinator in Tennessee last season with a classic thunder-and-lightning combination of Derrick Henry and Dion Lewis. After a pair of early middling games, the Titans began to squeeze Henry's carries. Over a nine-game stretch from Week 4 through Week 13, Henry averaged just nine carries per game. Finally, with Lewis predictably scuffling outside of New England, LaFleur turned the offense over to Henry and saw the former Alabama star rack up 585 yards on 87 carries over the final month.

Maybe LaFleur was just playing the hot hand. Williams was more effective than Jones on Sunday, though Jones did score two touchdowns. I'm not sure playing the hot hand off a handful of carries is a great idea, though, and Jones sure seemed to have the hot hand a week ago. If the Packers are trying to keep him healthy for a late-season push, maybe this will work out great. Given that Rodgers has been off to an inconsistent start while getting used to the new LaFleur scheme, a heavier dose of Jones would probably be best.

Surprising: This is the best defense in football.

I know what I said earlier about the Bills. They're great. But have you seen the Packers through three weeks? Mitchell Trubisky, Kirk Cousins and Joe Flacco aren't exactly Patrick Mahomes, but the Packers have faced 38 drives through three games, which is tied for the second most in football. They're allowing less than one point per possession, which is good for the second-best mark in football.

With the offense struggling for consistent production, though, the defense has to make leads hold up for long stretches. They've also created opportunities for the offense with takeaways. The Packers have turned opponents over on a league-leading 21.1% of possessions, and the same Mike Pettine defense that forced 15 takeaways in 16 games a year ago now has eight after just three contests.

While Pettine is around again as defensive coordinator, it's fair to note that many of the faces are new imports after a busy offseason from general manager Brian Gutekunst. And while I took issue with several of the contracts Gutekunst handed out -- deals that rise dramatically starting next season -- I can't argue with the results so far. The Packers refreshed their pass rush overnight by signing Preston Smith and Za'Darius Smith, and they made Joe Flacco & Co. miserable on Sunday. The Smiths racked up five sacks and six knockdowns, bringing their combined totals through three games to 7.5 sacks and 14 knockdowns. Clay Matthews and Nick Perry combined for five sacks and 15 knockdowns over the entire 2018 season. First-round pick Rashan Gary also picked up his first career sack against the overmatched Broncos line.

Safety, long a Packers problem, has become a strength. Adrian Amos has quickly settled in as a leader and made the key interception of Trubisky in the Week 1 win over the Bears. His partner, Darnell Savage, is a Defensive Rookie of the Year candidate through three weeks. The first-rounder made a spectacular play to pick off an errant Flacco pass on Sunday.

Cornerback has been a revolving door, too, but the Packers appear to have a full-fledged superstar on their hands in Jaire Alexander. I was impressed with Alexander as a rookie last season, although he didn't have much help in a secondary that was riddled both by injuries and players stuck out of position. With more support, Alexander looks like he might be one of the best cornerbacks in the NFC. He was spotted ripping the ball out of Noah Fant's hands Sunday. Kevin King has also played better in a small sample this season, pushing 2018 second-rounder Josh Jackson into what has primarily been a special-teams role.

Through three weeks, the Packers have scored seven touchdowns. Two have come on the opening drives of games, when they were likely running plays they had scripted before the game began. Of the other five scores, three came off takeaways. For years, the Packers would go as far as Rodgers could carry them. Now the defense is carrying Green Bay.

4. Los Angeles Rams

Not surprising: Wade Phillips has managed to get the most out of his defenders.

Most coaches who make it to the Pro Football Hall of Fame do so on the back of a great head-coaching career. It's probably about time to start thinking about great coordinators and positional coaches as potential Hall candidates, too. Dante Scarnecchia is an obvious choice after spending most of the past two decades molding a dominant offensive line in front of Tom Brady in New England. If I had to pick a coordinator, Phillips is the first guy on the list. Even though he won only one playoff game across 8½ seasons as a head coach, he remains one of the league's finest defensive minds at age 72.

If the Rams' offense had held up its end of the bargain last February, we would be lauding Phillips for holding the Patriots to 13 points. Through three weeks, while the Rams have given up 49 points, 14 of those points came off of short fields against the Panthers in Week 1. On Sunday night, the Browns started drives from their own 38-, 43- and 49-yard lines, as well as a brief drive before the end of the first half from the Los Angeles 18. Those drives produced a total of three points.

Phillips got so far into Freddie Kitchens' head that he somehow convinced the struggling Browns coach to call for a draw on fourth-and-9, something I don't believe an NFL team has done on purpose since the Dolphins attempted a fourth-and-10 draw with Bernie Parmalee in 1997. Phillips saw Mayfield struggling to deal with pressure and repeatedly reacting to even the threat of pressure by rolling away from the pocket to his right to try to make a desperate play. By the end of the game, Phillips was lining up his front four in wide splits to try to isolate Cleveland's tackles and then twisting his linemen to both create quick interior pressure while having someone waiting when Mayfield panicked and ran outside. While Mayfield stuck in the pocket and made a pair of great throws on the final drive, the game-sealing interception is a clear example of what Mayfield was doing wrong.

While the Rams have a competitive advantage with Aaron Donald collapsing pockets, this isn't a one-man show. Phillips' defense has improved on 2018 despite losing big names such as Ndamukong Suh and Lamarcus Joyner. Eric Weddle has stepped in at safety for Joyner, but another big name has made his presence felt. Remember when I mentioned Clay Matthews as a disappointment in the Packers section? Rushing the passer on more than 86% of passing plays, Matthews has four sacks in his first three games with the Rams, including two on Sunday night. The second of those sacks would qualify as a coverage sack, but he did a great job of shedding what appeared to be a stable block to take Mayfield down and prevent him from scrambling.

Phillips also has a habit of developing little-known inside linebackers into stars, and Cory Littleton continues to improve as one of Phillips' star pupils. I mentioned this last week, but how many inside linebackers could hold their own while covering Michael Thomas on a drag route? Littleton had a monster game against the Panthers in Week 1 and would be a Pro Bowler if we were casting ballots after three weeks.

Should anyone be surprised here? The most important thing the Rams have done over the past decade is hire Sean McVay. The second-most important thing they've done, realistically, is convince Phillips to work alongside him.

Surprising: Jared Goff is struggling.

In general, Goff is doing great. He just pocketed a $25 million signing bonus. He has about as much job security as any young quarterback in football. He's 24 and living in Los Angeles and has a brilliant coach who helps him unlock defenses at the line of scrimmage.

Over the first three weeks of this season, though, Goff hasn't played well. Opposing defenses have emulated the Patriots' game plan from the Super Bowl and played what amount to six-man fronts to try to force the Rams away from their outside zone game. Todd Gurley hasn't been healthy enough to play his usual workload. The Rams are rebuilding the interior of their offensive line. This was supposed to be the point in which Goff could shoulder a larger portion of the workload, but that hasn't happened.

Earlier in his career, the Rams spent money on pieces around Goff to surround their cheap young quarterback with talent. Now he is the expensive one. His cap hit doesn't rise from $10.6 million to $36 million until next season, but the Rams have to expect him to play like a franchise quarterback now that he's beginning a franchise quarterback caliber deal. He just hasn't been that guy.

Remember the expected completion percentage stat I mentioned about Allen? A quarterback making Goff's throws would be expected to complete 67.7% of his passes, the eighth-friendliest rate in the league in 2019. Goff is completing only 62.9%, and the only passers with a larger gap are either injured (Cam Newton, Ben Roethlisberger), chum (Josh Rosen) or subject to major criticism (Mitchell Trubisky, Andy Dalton). Goff is not supposed to be in a group with Dalton and Trubisky.

Plenty of quarterbacks get off to slow starts and recover just fine, and I'm not particularly concerned about his completion percentage being five points below expectation after three weeks. What does strike me as something to look out for, though, is how the Rams seem to be struggling with play-action. From 2017 to '18, Goff averaged more than 10.1 yards per play-action pass, posted a passer rating of 112.3, and threw 21 touchdowns against three picks on 335 attempts. Through 40 play-action throws this season, he is averaging 7.9 yards per play-fake with a passer rating of 53.5. He has thrown three interceptions on play-action in three weeks, including both of his picks on Sunday.

The interceptions were throws Goff would like to take back. On the first pass, Brandin Cooks has a step on reserve cornerback T.J. Carrie, but the throw is late and in a place where Carrie can make a play. It required an impressive diving pick, but you'll notice in this animation from NFL Next Gen Stats that Robert Woods (17) might also have been open across the middle of the field:

I can't fault Goff for not squeezing that pass into Woods, in part because his second interception was a similar decision. Again, Woods has a step on his defender, but this pass is thrown in the wrong spot and gives Joe Schobert a chance to tip it up in the air. Goff needed to put more loft on it and fit it in the triangle between Jermaine Whitehead (35), Eric Murray (22), and Juston Burris (41), with the latter player eventually bringing in the tipped pick.

It wasn't just the interceptions. McVay was visibly frustrated with Goff for what he seemed to consider a subpar decision on third-and-1 during the second half. He missed a wide-open out route to Cooks in the first half on a pass that Next Gen Stats estimated to have a 72.6 percent chance of completion.

Goff's numbers on Sunday look fine apart from the interceptions -- 24-of-38 for 269 yards with two touchdown passes to Cooper Kupp -- but he was facing a Browns secondary missing all four of its starters and its best linebacker, Christian Kirksey. The five starting defensive backs played every snap; they included a pair of backups (Carrie and Terrance Mitchell), a special-teamer (Murray), a defensive back they claimed off waivers from the Raiders earlier this month (Burris), and another who was claimed off the Packers waiver wire last November (Whitehead). Isn't this the sort of spot Goff is supposed to smash?

One more thing to worry about and then I'll move on: Goff has traditionally been best in the warmest month of the NFL season under McVay. From 2017 to '18, he posted a passer rating of 123.8 and averaged more than 10.3 yards per attempt in September. Over the ensuing three months of those seasons, Goff posted a passer rating of 94.6 while averaging 7.6 yards per pass. I don't expect that sort of drop-off to occur again in 2019, but with Goff currently sporting a passer rating of 84.5 while averaging 7.0 yards per attempt and posting a Total QBR in between that of Josh Rosen and Eli Manning, I do know that we're going to need to see a better Goff for the Rams to continue on their undefeated run.

3. Kansas City Chiefs

Not surprising: There's still life in LeSean McCoy's legs.

We often underestimate just how much context matters in evaluating skill position players. When the Bills cut Shady McCoy this summer, I joked on Twitter that McCoy would lead the Chiefs in rushing. The responses, as you might suspect, are from Chiefs fans laughing at the idea that their team would even sign McCoy, let alone see the former Eagles standout play ahead of Damien Williams or Darwin Thompson.

I was surprised hours later when the Chiefs won a bidding war with the Chargers and gave McCoy a one-year deal worth $3 million with another million in incentives. Signing McCoy made sense, but the price tag was far more than I would have expected for a back who averaged just 3.2 yards per carry with more than 2,400 career carries on his odometer.

Freed to play in his old coach's offense, though, McCoy has looked impressive. Williams was awful to start the season, carrying the ball 22 times for just 34 yards before going down with a knee injury. McCoy, dealing with an ankle knock, has rushed 29 times for 158 yards and been a dangerous threat as a receiver. He scored twice in Sunday's win over the Ravens, contributing 80 yards and five first downs or touchdowns on 11 touches.

It's reminiscent of Williams himself, who looked to be an anonymous, replacement-level back in Miami before starring over the last month of the season in Kareem Hunt's absence in 2018. What looks like a washed-up back somewhere else looks like a star in Kansas City.

Surprising: The Chiefs still can't stop the run.

The Chiefs couldn't stop opposing teams from running the football last season. In Bob Sutton's final year as defensive coordinator, the Chiefs finished 32nd in rush defense DVOA and allowed 5.0 yards per carry, and 51.5% of opposing carries increased their expected chances of scoring, the highest rate for any team in the NFL.

Over the offseason, the Chiefs underwent a defensive overhaul. Out went Sutton, Justin Houston, Dee Ford and oft-injured star safety Eric Berry. They brought in new coordinator Steve Spagnuolo, signed Alex Okafor and Tyrann Mathieu, and traded a first-round pick for Frank Clark, who is a better run-defender than Ford. Former starters like Reggie Ragland and Daniel Sorensen have been kicked out of the lineup for Cowboys import Damien Wilson and rookie safety Juan Thornhill. A lot has changed.

And yet, nothing has changed. The Chiefs are allowing opposing offenses to average an eye-watering 6.2 yards per carry this season. No other team even hits a rushing average of 5.5 yards per carry. The Chiefs are also last in average yards after first contact at 2.5 yards per attempt, and their 34.3% first-down rate is 31st in the NFL. I didn't think they were going to turn into the '85 Bears overnight, but run defenses this bad usually improve just by sheer chance the following season, let alone after making significant changes.

Of course, Kansas City is also 3-0, so you can make the case that its run defense might not matter. On Sunday, though, it might have ended up extending what should have been a comfortable win over the Ravens. Even given an early lead, it allowed a couple of prayers from Lamar Jackson to become completions and let the Ravens run for 203 yards and four touchdowns on 32 attempts. The Chiefs were able to stop the Ravens on all three of their two-point tries, including a pair of rush attempts, but the game shouldn't have been that close.

2. Dallas Cowboys

Not surprising: The Cowboys are 3-0.

Yes, I did predict that the Cowboys would decline in 2019, and I stand by my prediction. I've been impressed with how the offense has looked for stretches this season, especially in the second half of games, but there are still 13 games to go. No one is right or wrong about anything after three weeks.

Given Dallas' schedule to start the year, though, I don't see how anybody could have expected much less than a 3-0 start. It has faced what very well might be three of the five worst teams in football with the Eli Manning version of the Giants, Washington and the Dolphins. Two of those three games were at home. The Cowboys won all three games handily, which is good for a team that rode its luck in close victories last season, but 3-0 was the most plausible outcome for the Cowboys heading into the year.

Week 4 looked like it would begin the tougher part of their slate, but even after the Saints impressed in Seattle on Sunday, the Cowboys have to feel better about facing Teddy Bridgewater in New Orleans than Drew Brees. There's a chance they may also get another backup quarterback two weeks later if Sam Darnold isn't ready by Week 6. We'll get a better idea of where this team stands once its schedule gets tougher.

Surprising: The Cowboys haven't paid Dak Prescott.

Is Jerry Jones waiting to sell some of the stadium art before paying his star quarterback? The price is only going to get more expensive, and if the Cowboys can't sign Prescott before Patrick Mahomes signs his mega-extension next offseason, it might make the prices we're seeing mooted for a possible Prescott deal seem like a bargain.

Is it foolish to put Prescott and the reigning MVP in the same sentence? Before this season, maybe. Through three weeks, though, Prescott has been every bit as good as Mahomes. To go back to that expected completion percentage stat from NFL Next Gen Stats, Prescott's options would have typically generated a 64.6% completion rate. Even after an uneven game against the lowly Dolphins, he has hit on a whopping 74.5% of his attempts. The resulting difference of 9.9 percentage points is the most in football, ahead of Russell Wilson (8.7 percentage points) and Mahomes (5.7 points). And if you think that is related to Prescott throwing shorter passes, the Cowboys signal-caller's average attempt has traveled 9.7 yards in the air, further than that of Mahomes, at 9.2 yards per throw. Prescott's percentage of receivers who are open and/or wide open are both below league average. He's playing like a legitimate superstar.

Prescott will be tested in the weeks to come. Over the next two weeks, the Cowboys face the Saints and the Packers, who are each tied for the league lead in pressure rate at 36.9%. If he keeps these numbers up and leads the Cowboys to 5-0, Jones might have to sell the scoreboard to finance Dak's new deal.

1. New England Patriots

Surprising: Sony Michel has been one of the league's least productive backs.

I'm going to cheat and sneak two surprises in here, because the Patriots going 3-0 and looking like the best team in football isn't anything new. The negative surprise is about Michel, who I thought might have a viable shot at leading the league in rushing and/or rushing touchdowns as the primary back on a top-three offense. The Patriots have been down as many as three of their five starting offensive linemen for parts of this season and will be without center David Andrews for the year and left tackle Isaiah Wynn until midseason, but Michel has been a mess.

After carrying the ball nine times for 11 yards in Sunday's win over the Jets, Michel now has 45 carries for 108 yards, an average of merely 2.4 yards per carry. The 2018 first-round pick has two rushing touchdowns, but those scores have required eight carries and a lone target inside the 10-yard line. I was expecting more big plays from Michel given his explosiveness at Georgia, but he has shown virtually no burst or any propensity to get more than what's blocked during the first three weeks of the season.

What makes Michel so interesting, in part, is how he goes against many of the habits Bill Belichick typically seems to deploy at the running back position. Belichick traditionally prefers to fill his backfield on the cheap with mid-round picks or players signed off of the waiver wire. Michel was a first-round pick. Belichick loves backs who are versatile and don't reveal whether you intend to run or pass before the snap, although he's also found a role for LeGarrette Blount and Benjarvus Green-Ellis in recent years. Since the start of 2018, the Pats have run the ball 75% of the time when Michel is on the field and just 32% of the time when he's on the sideline.

Belichick seemed to grow frustrated with Michel on Sunday. With James White out and Michel struggling, both Rex Burkhead (53 snaps) and Brandon Bolden (21 snaps) saw the field more frequently than Michel, who played only 17 snaps. It's way too early to give up on the Georgia product, and I suspect he'll play a big role for the Pats at some point during the season. That time just might not be this Sunday in Buffalo.

Also surprising: This is the best defense in football.

It has to be the Patriots, right? As good as the Bills are under difficult conditions, and as dominant as the Packers have looked taking away the football this season, the Patriots have yet to allow an offensive touchdown all season. The 14 points the Jets scored Sunday came on a muffed punt and a pick-six from backup quarterback Jarrett Stidham. The Steelers kicked a field goal in the opener. The Dolphins didn't advance past midfield until the fourth quarter in Week 2.

Through three weeks, the Patriots are allowing 0.27 points per possession. That is no typo. Factoring in the Super Bowl win over the Rams, the Pats' defense has now gone four games without allowing a passing or rushing touchdown by the opposing offense.

That has happened just eight times since the AFL-NFL merger, and in these score-happy times, we haven't seen a team piece together a four-game scoreless streak since the Steelers pulled it off over a five-game run during the 2000 campaign. Those Steelers are the only team to even make it to four in a row since 1991. With Kordell Stewart and Kent Graham failing to combine for a completion percentage of even 50%, that Steelers team went 9-7 and missed the playoffs. The Patriots should be just fine on the completion percentage front with Tom Brady.

I wrote about the Patriots at length last week, but even given the Super Bowl performance and a relatively meek start to the season, it's difficult to fathom that they would have been this good on defense. They're mostly returning the core of a defense that finished 16th in DVOA in 2018 and 31st the previous season, having swapped out Trey Flowers for Michael Bennett while adding Jamie Collins. For all of Belichick's well-earned reputation as a defensive genius, the Pats haven't ranked in the top 10 in DVOA since 2006. I'm never shocked when a Belichick defense plays well, but to put together the best four-game stretch in nearly two decades?

My logic in talking about a 16-0 Pats season had more to do with a 2007-esque receiving corps than that dominant defense. Of course, a week later, the receiving corps is no more. At times on Sunday, not one of them was on the field; Antonio Brown was cut, Julian Edelman injured his rib, and Josh Gordon came off the field with hip and finger injuries. It wasn't going to matter against the Jets.

Look at the DVOA stat I just mentioned, though. Since Brady leveled up and turned into the greatest quarterback of all time in 2007, the Patriots haven't produced a single top-10 defense to play opposite their star quarterback. They've ranked in the top 10 in scoring defense many times, but that's a product of great field position and having to face a low number of drives against teams that were desperate to throw and catch up. Since 2007, the average Pats defensive drive has come with 73.9 yards to go for a touchdown, which is a little over a yard more per drive than any other defense in football.

Brady and the receiving corps of doom lasted one week. Brady and the best defense he has had in 15 years against the league's easiest schedule? I wouldn't quite rule out the 16-0 dream just yet.