Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Defense and Rest Time

Do defenses really wear out over the course of a game? Do defenses benefit from long drives that give them more time to rest on the sideline? Guest columnist Ben Baldwin investigates.

21 Nov 2017

Any Given Sunday: Giants Over Chiefs

by Rivers McCown

The Giants are terrible.

Odell Beckham was lost for the year early, removing any pretense of offensive explosiveness. For the fifth straight year, the Giants are on pace to have a bottom-12 run offense DVOA, and they have already rung the Orleans Darkwa emergency bell. Bob McAdoo gets fired by his team's fan base on Twitter every quarter.

But to go back to the point of contention, so to speak, the real issue for the Giants has been the defense. We projected the Giants as a wild-card team this offseason. But even though Beckham is out, the offense has mostly kept its end of the bargain. Our -0.9% DVOA projection for the offense has been answered by a -2.5% DVOA in reality. The passing offense, led by Evan Engram's quick emergence into a star tight end, has kept the Giants close to even.

The defense, on the other hand, has been downright ugly. The Giants had our fourth-best projected defense in the preseason. Yet even after this performance, they're at 28th at defensive DVOA, at 11.3%. We expected some defensive regression from the Giants because teams that leap from the bottom 20 to second overall in one year average a gain of 12.1% in DVOA the next year. (Remember, higher DVOA means worse defense.) We didn't know that the defense would turn completely back into a pumpkin.

It's rare to see a collapse this stark without major injuries, and we've seen our share of those on the Giants defense. Jason Pierre-Paul and Olivier Vernon made the pass rush top-heavy last year. Pierre-Paul was slowed by surgery in the offseason, and Vernon's absence from Weeks 4 to 9 left the Giants without much of any pass rush. Last year, the two combined for 94 hurries in 1,832 snaps. This year, through Week 10, they have 32.5 hurries in 847 snaps. The Giants haven't fallen much in adjusted sack rate from last season, because they weren't great then either, but they aren't getting the same pressure they got last year. They've fallen from eighth in Sports Info Solutions' pressure rate last year to 24th in 2017.

The Giants have also seen their broken-tackle rate skyrocket from 9.0 percent in 2016 to 12.4 percent in 2017, a rate that puts them in the bottom four. Jonathan Casillas had as many broken tackles through Week 10 as he did all last season. You can say the same about Landon Collins, Vernon, Keenan Robinson, and Eli Apple as well.

Regression also hit the secondary hard. Dominique Rodgers-Cromartie and Janoris Jenkins finished No. 1 and No. 2 in success rate among qualifying cornerbacks this year. Jenkins is 12th this year through Week 10, despite some, uh, let's say questionable effort at times:

But in suspending Rodgers-Cromartie, the team put a spotlight on 2016 first-rounder Apple. Apple has proven through two years that he is nothing if not combustible, as he's fourth from the bottom in success rate this year after posting a 41 percent success rate in 2016. Now, the Giants picked up Ross Cockrell on the cheap after Pittsburgh unexpectedly let him go, and he has played well when he's been on the field, but he only played 214 snaps until getting 74 this week. Apple was a healthy scratch. Cockrell, like Trevin Wade last year, is stepping in to be better than the original plan.

All of which is to say that while the Giants have played poorly, the key cogs are on the field for them to rebound defensively. They just need to stop playing Apple, find actual linebacker help (B.J. Goodson has been solid when healthy), and surround the 2017 versions of their best 2016 players with non-black holes. They're not getting back to the No. 2 defense in the league, but they can be much better than they have been so far this year. And perhaps this game can be their first step to it.

Where the Game Swung

As you would expect from a low-scoring, close game, the WPA chart squiggles all over the place. Let's specifically cover Kansas City's last drive to end regulation, and New York's to end the game in overtime.

The Chiefs got the ball back with one timeout left and 1:30 to drive the length of the field to tie or win the game. It almost ended before it began, as Alex Smith was picked on the first play of the drive. However, he was bailed out by a pass interference call on Jenkins.

(Click here if you are having trouble loading the image.)

The Chiefs followed that with a middle-of-the-field bomb to Travis Kelce, picking up a Giants zone blitz. The Chiefs were set up in range of a long field goal, at the New York 34, with a minute to play. From here on, Kansas City ran seven plays in two-minute mode, with mostly slants, flats, and quarterback runs, and advanced the ball just 29 more yards. The only pass targeted in the end zone was a throw to Kelce that was so far off-target it was impossible to call pass interference on the man who bumped him. Andy Reid seemed to take the air out of the ball and play for the tie.

New York got the ball back after another limp Chiefs drive and Eli Manning quickly carved up a pair of simple slants against single-high that went for 10 and 13 yards before the offense stalled. A huge blitz on third-and-5 saw Manning's back-shoulder ball flutter a bit too much for Engram. Then, on fourth-and-5 at the Kansas City 36, with no point in punting and no trust in Aldrick Rosas, the Giants went for it. Leading to this:

(Click here if you are having trouble loading the image.)

That play was that last upwards line you see in the graph: it swung the win probabilities from 34.4% NYG to 83.3% NYG. Ridiculous catches are fun, and it's always awesome when a receiver gets to posterize a corner to the point where he draws a DPI and still catches the ball.

By the VOA

KC -20.9% -1.6% 7.5% -11.8%
NYG -11.9% -29.2% -0.3% 17.0%
KC -10.4% -8.4% 7.5% 5.5%
NYG -1.8% -16.0% -0.3% 13.9%

Well, one unit showed up big, so you know I felt obligated to cover the Giants defense first. Now for the depressing part...

The Eye of the Beholder

Alex Smith was the reason the Chiefs got off to a great start. Alex Smith is the reason the Chiefs have faltered since Week 6.

After starting out with a pass offense DVOA of more than 50.0% in four of their first five games, the Chiefs have been above 23.1% DVOA once in their last five games. The surprising thing about it, though, is that the difference isn't where you would expect it to be. I think the generic talking point about Smith's drop-off is something like "well, the deep ball stopped working." Except, the deep ball hasn't declined all that much, either in usage or effectiveness. The Chiefs were first in deep-pass DVOA over the first five weeks, and have been fourth from Week 6 to 11. It's the short passes that are going nowhere.

Kansas City’s Passing Offense by Week, 2017
Weeks Passing DVOA Passing DVOA
on Deep Balls
Deep Ball
Passing DVOA
on Short Balls
Weeks 1-5 59.8% 183.3% 29 70.2%
Weeks 6-11 2.7% 137.4% 37 -5.8%

That's one of the four lowest short-pass DVOAs in the league since Week 6, ahead of only the Jets, 49ers, and Browns. They're the only team with a negative DVOA on short passes that doesn't have five or more turnovers from them in that span.

It's not exactly the answer I expected to find when I started this piece, because it doesn't actually seem to make much sense. Why would a team with Kareem Hunt's tackle-breaking ability, the slippery Tyreek Hill, and all those lovely option pitches to Travis Kelce tumble so far? Is it just a matter of defenses adjusting and daring Smith to throw deep even more?

Kansas City’s Short Passing Offense by Weeks
Weeks 1-5
Weeks 6-11
Player Targets DVOA Player Targets DVOA
Travis Kelce 27 1.2% Travis Kelce 32 2.1%
Tyreek Hill 26 11.3% Tyreek Hill 25 -5.3%
Albert Wilson 17 21.1% Kareem Hunt 22 -5.2%
Kareem Hunt 16 3.6% Demarcus Robinson 15 -25.1%
Chris Conley 10 29.6% Charcandrick West 12 -25.4%
Charcandrick West 10 -1.3%

Or, perhaps, is it a matter of us not respecting Kansas City's neglected receivers in the offense? With Albert Wilson and Chris Conley both out, the Chiefs have been forced to focus the workload more on Hunt dumpoffs and screens. It's hard to imagine that this alone has been so demonstratively bad for the Chiefs, but perhaps Reid's options and ball-could-go-anywhere idea is less tested when defenses don't actually care about Demarcus Robinson.

Either way, you see that the decline has hit every receiver except for Kelce, and it has hit them hard. Hill has been dealing with hamstring injuries so it's perhaps not ridiculous to suggest that he has been less than himself because of that. But what else is there to blame a decline this steep on? Lots of teams lose lower-rung options in their passing game and find an easy next man up.

Smith is one of my favorite quarterbacks simply because you can project so much on him, and what you put on him says a lot about you. To some he's a "good enough to win with" guy, and to some he's a known mediocrity who will never carry a franchise anywhere because quarterbacks are eventually the trump card in any rational analysis of a team. This year, Smith has had a little bit of red meat for either crowd.

Ultimately, my takeaway is that the Chiefs simply haven't been effective enough in the run game since Week 5 to give Smith the same kind of freedom he had early on in the year. Kansas City's rush offense DVOA has gone from a league-leading 24.2% DVOA in Weeks 1 to 5 to a pedestrian -14.0% over their last five games. That's good for 18th over that stretch. It's something we covered earlier in the year that just hasn't been fixed yet despite the return of healthy offensive linemen.

But in saying that, it blows me away how much of a difference that can make. What a stark difference it is, from effortless to putrid, all because the run game has been stymied.

Posted by: Rivers McCown on 21 Nov 2017

7 comments, Last at 27 Nov 2017, 10:37am by nat


by Travis :: Tue, 11/21/2017 - 4:43pm

That play was that last upwards line you see in the graph: it swung the win probabilities from 34.4% NYG to 83.3% NYG.

Teams with the ball on the opposition 2 with the score tied on not-the-first-drive of overtime only win five-sixths of the time?

by RickD :: Tue, 11/21/2017 - 4:59pm

I don't know about generic teams, but these are the 2017 Giants after all.

The only thing I can think is Aldrick Rosas is not a particularly good kicker. He's 12 for 17 this season on field goal attempts. But most of the misses have been over 40 yards, with only 1 less than that, and it was over 30 yards.

If I had to assign a probability there, I'd say it was at least 95%.

I wonder if the WP calculator isn't handling overtime properly.

by Bryan Knowles :: Tue, 11/21/2017 - 6:05pm

PFR's WP calculator does some odd things in niche situations late in close games. They probably could stand to do a bit of housecleaning in it this offseason.

by Vincent Verhei :: Tue, 11/21/2017 - 6:28pm

I suspect there have been exactly six teams in that specific scenario in their database, and one of them fouled things up.

by BJR :: Tue, 11/21/2017 - 6:37pm

In which case, how can they possibly publish their numbers with any degree of integrity?

by nat :: Mon, 11/27/2017 - 10:37am

Since the almost-by-not-quite sudden-death variant of the OT rules went into place for the regular season in 2012, there have been seven cases of a team having a first down at the five or closer. 5 were wins. 1 was a loss. 1 was a tie.

It's reasonable (understandable, although not right) to just ignore tied games when working up a win probability database for OT. That would make teams 5-1 in the database for this situation.

Both the tie and the loss were cases where the first drive got inside the five, but only resulted in a field goal. Applying those samples to this situation is clearly a mistake, since the win probabilities are very different depending on whether it's the first drive of OT. After the first drive, a field goal to take the lead takes the win, too.

Tiny data set, AND bad categorization of the situations.

This is not the first weirdness in the various win probability calculators. It won't be the last. It's one of the problems with just crunching the numbers without applying common sense.

by Mike W :: Tue, 11/21/2017 - 5:20pm

I agree with the takeaway. The KC run blocking has been pretty futile since the first few games.