Welcome to Week Three's edition of selected game recaps, where we take an objective look at the subjective best of the week's games, including:
- New England at Buffalo, in which the Bills make their presence as contenders official.
And
- Detroit at Minnesota, in which the Vikings prove allergic to leads again.
Buffalo 34, New England 31
Tom Brady thew four interceptions last season. He threw four interceptions Sunday against the Buffalo Bills.
Brady wasn't going to go another season with an interception rate below one percent. Only Steve DeBerg's 1990 campaign with the Kansas City Chiefs (four intereceptions in 444 attempts) joins Brady's 2010 as seasons with 400 or more attempts and fewer than 1% interceptions. And we know Brady is going to throw 400 passes this year if he stays healthy -- he's on pace to hit 399 by week nine and on pace for the first 700 attempt season of all time.
But Sunday wasn't regression to the mean. Sunday was regression above and beyond the mean. For Brady this season, five interceptions hardly represents the disaster it would for most players. Matt Cassel also has five picks, but they've come in 51 fewer passing attempts. According to INT+, which normalizes interception rates such that 100 is average (and higher means better/fewer interceptions), Cassel sits at a dismal 65. Brady isn't where Brady should be, but his mark of 91 is merely below average, not awful nor outlandish for a three game sample.
The problem is in the timing.
Sunday's interception binge began in the second quarter with the Patriots just 13 yards away from a touchdown with a chance to go up by three scores yet again in the second. The second ended the Patriots first drive of the second half around midfield. The third squelched a drive within three yards of the red zone with a mere seven point lead. The fourth and final proved too sharp a dagger, as a tie game became a seven-point deficit once Drayton Florence carried it over the plane and into the end zone.
Ryan Fitzpatrick threw two interceptions of his own, but they weren't nearly as costly. One came on a 4th down from the New England 35 and was basically an incomplete pass. The other one hurt -- it killed an early Buffalo drive with the Bills already trailing by seven. One of the Four Truths of football from Bill Connelly of SB Nation and Football Outsiders is that you can't win a game in the first quarter, but you can lose one. This game would've been the perfect illustration, had it not been for Brady's interceptions illustrating point number two: big plays win games. Observe, the massive impact of these picks, in terms of WPA and EPA lost:
Click to slightly embiggen
In a three point game, Brady's interceptions cost the Patriots a total of 18.8 expected points and a net of 12.3 expected points against Ryan Fitzpatrick's interceptions. Brady played very well outside of these four plays, ending with 11.2 total EPA and a +0.36 WPA. But on this given Sunday, an afternoon with uncharacteristically poor ball security was a crucial reason why Brady's Patriots are now looking up at Fitzpatrick's Bills.
Detroit 26, Minnesota 23 (OT)
As a Numbers Guy™, I am very wary of trends based on small sample sizes. This week, the trend du jour circa 1:30 PM Central Time was the Vikings' inability to hold leads: through two games, the Vikings blew two huge leads, taking win probabilities of at least 82% into half time. Come halftime against Detroit, and the Vikings held a 20-0 lead and a 95% win probability. This Numbers Guy™ wasn't buying into the trend.
That's the thing about probability though. The chances of the Vikings going out and blowing these three leads in a row is 0.00063 (1 in 1587, roughly), but the chances of any team doing it over three games? There are enough three game stretches played over the course of a season where we at least get into the realm of possibility. The Vikings just happen to be in that one special three-game stretch.
The Vikings coaching staff will likely be panned for going away from the running game in the second half. -- Adrian Peterson said he "can't explain it" himself on Tuesday. After picking up 114 yards and a score on 18 rushes in the first half (6.33 yards per carry), the Vikings abandoned the run in the second half, only rushing eight times against 19 passing plays as the Lions stormed back.
But let's not ignore just how ineffective those eight rushes were. The Vikings lost two yards on these plays, racked up a whopping -5.4 EPA, and let 21 points of win probability slip through their fingers. Although it's difficult to envision any team, let alone one with Adrian Peterson, fail so thoroughly in the running game, we should have seen a second half drop-off coming. The Vikings managed two big runs (43 and 39 respectively) when Peterson hit the second level and beyond, but he couldn't consistently get past the first level. On no other occasion did the Vikings' run attack gain more than eight yards, and overall they finished with a 39% success rate in the first half.
That's not usually a precursor to a 1-for-8 success rate in the second half, and a cogent argument could certainly be made for sticking to the run despite the repeated failure -- the Vikings could still run clock, Adrian Peterson is probably too good for such a poor streak to be sustained, the Vikings pass attack isn't great, et cetera -- but the ineffectiveness of the running game was as much to blame as the lack of attempts it was given.
The Vikings' pass attack wasn't successful either, compiling -6.5 EPA and -0.15 WPA over its 23 attempts in the second half. Credit is certainly due to the Lions' offense as well -- Matthew Stafford continued to impress, putting up +.36 WPA, +11.6 EPA, and 6.6 AYPA, and, not coincidentally, Calvin Johnson (9.8 yards per target) and Brandon Pettigrew (8.6 yards per target, non-Stafford game high 8.8 EPA) impressed with their duties in the air attack.
At this point, the Vikings are pretty clearly better than an 0-3 record -- a team simply doesn't put together the kinds of first halves Minnesota has by accident. But with the Lions atop the NFC North at 3-0 along with the Green Bay Packers, it might already be a case of too late, regardless of how much or how little we see from Minnesota in weeks four through sixteen.
That's the thing about probability though. The chances of the Vikings going out and blowing these three leads in a row is 0.00063 (1 in 1587, roughly), but the chances of any team doing it over three games? There are enough three game stretches played over the course of a season where we at least get into the realm of possibility. The Vikings just happen to be in that one special three-game stretch.
The Vikings coaching staff will likely be panned for going away from the running game in the second half. -- Adrian Peterson said he "can't explain it" himself on Tuesday. After picking up 114 yards and a score on 18 rushes in the first half (6.33 yards per carry), the Vikings abandoned the run in the second half, only rushing eight times against 19 passing plays as the Lions stormed back.
But let's not ignore just how ineffective those eight rushes were. The Vikings lost two yards on these plays, racked up a whopping -5.4 EPA, and let 21 points of win probability slip through their fingers. Although it's difficult to envision any team, let alone one with Adrian Peterson, fail so thoroughly in the running game, we should have seen a second half drop-off coming. The Vikings managed two big runs (43 and 39 respectively) when Peterson hit the second level and beyond, but he couldn't consistently get past the first level. On no other occasion did the Vikings' run attack gain more than eight yards, and overall they finished with a 39% success rate in the first half.
That's not usually a precursor to a 1-for-8 success rate in the second half, and a cogent argument could certainly be made for sticking to the run despite the repeated failure -- the Vikings could still run clock, Adrian Peterson is probably too good for such a poor streak to be sustained, the Vikings pass attack isn't great, et cetera -- but the ineffectiveness of the running game was as much to blame as the lack of attempts it was given.
The Vikings' pass attack wasn't successful either, compiling -6.5 EPA and -0.15 WPA over its 23 attempts in the second half. Credit is certainly due to the Lions' offense as well -- Matthew Stafford continued to impress, putting up +.36 WPA, +11.6 EPA, and 6.6 AYPA, and, not coincidentally, Calvin Johnson (9.8 yards per target) and Brandon Pettigrew (8.6 yards per target, non-Stafford game high 8.8 EPA) impressed with their duties in the air attack.
At this point, the Vikings are pretty clearly better than an 0-3 record -- a team simply doesn't put together the kinds of first halves Minnesota has by accident. But with the Lions atop the NFC North at 3-0 along with the Green Bay Packers, it might already be a case of too late, regardless of how much or how little we see from Minnesota in weeks four through sixteen.
Wow cool thanks for breaking down the Vikings game. I've been wondering what happened in the second half, and why the run game was abandoned. Is that the same issue in their last 3 games? Defenses adjust and their run game becomes ineffective?