How Football ChangedThe long arc

A Quarter Fewer NFL Deep Throws Go Down the Middle

In 2006–12, 27% of NFL deep passes went down the middle. Since 2021, 20% have. The ones still thrown there are worth more than deep passes outside, probably partly because quarterbacks pick their moments. They are also intercepted more often. Whether a pass counts as “middle” also depends on the home team’s stat crew, which charts every pass.

Verdict: Data story

By Edgehalla ResearchPublished 15 min read

Post on X

On Thursday, Sept. 8, 2011, in week 1 at Lambeau Field, Drew Brees threw nine passes that traveled 15 or more yards in the air down the middle of the field. His receivers caught eight.

Green Bay won 42–34, and Brees kept throwing there. In 2012 he sent 49 of his 131 deep passes down the middle.

By 2018 it was 14 of 90.

Quarterbacks throw about a quarter fewer deep passes down the middle

Brees was not the only one who backed off. Across the league, quarterbacks went deep down the middle less often. Whether that was a mistake is a harder question.

Every pass in the NFL’s play-by-play is logged as going left, middle or right. That label is a judgment call by the stat crew working each game, and the play-by-play never says where “the middle” starts. Every one of those Brees throws at Lambeau flew at least 15 yards past the line of scrimmage before anyone touched it. The stat sheet records that distance as air yards. In this story, any pass with 15 or more air yards is a deep pass, so a throw the play text calls “short” still counts if it flew 15 yards.

From 2006 to 2012, about 27% of deep passes went down the middle, a little more than one in four. From 2021 to 2025, about 20% did, one in five. That is about a quarter fewer. With more than 42,000 deep passes in those two spans, the drop is far too big to be luck.

About a quarter of the deep passes down the middle went away

Share of NFL deep passes (15+ air yards) logged as thrown down the middle, 2006–2025. The shaded band shows how far luck alone could move each season.

Show data
The middle’s share of deep passes was 27.2% in 2006–12 (6,779 of 24,933) and 20.2% in 2021–25 (3,641 of 18,051), a change of −7.0 points, most likely between −8.0 and −6.2.
YearAll deep passesAll deep passes (range)Early downs, game in doubt
200627.2%25.7% to 28.7%25.2%
200727.8%26.3% to 29.3%27.7%
200826.5%25.1% to 28.1%26.1%
200928.8%27.3% to 30.3%29.4%
201027.0%25.6% to 28.4%27.4%
201127.1%25.8% to 28.6%26.4%
201226.0%24.6% to 27.4%24.6%
201324.4%23.1% to 25.8%24.0%
201422.3%21.0% to 23.7%21.3%
201522.0%20.7% to 23.3%22.3%
201623.8%22.5% to 25.2%23.2%
201722.3%21.0% to 23.7%22.0%
201822.3%21.0% to 23.7%21.2%
201922.6%21.3% to 23.9%22.3%
202022.4%21.1% to 23.8%19.8%
202119.8%18.5% to 21.1%19.1%
202221.6%20.3% to 23.0%22.6%
202318.6%17.3% to 19.8%17.2%
202421.3%20.0% to 22.6%20.0%
202519.7%18.4% to 21.0%18.8%
72,665 deep passes, regular season. The second line counts only first and second downs with the game still in doubt, outside the last two minutes of either half.

It fell in steps. Through 2012 the share sat between about 26% and 29% each season. It dropped in 2013 and 2014, stayed around 22–24% through 2020, and has been about 19–22% since 2021. Any one season can swing a point or so by chance, so the 2023 low and the bounce after it are not worth reading into.

Draw one straight line through all 20 seasons and most years sit a point or so off it. 2009 and 2014 strayed furthest, about 2 points above and below the line. A typical season misses the line by about 1.2 points, well more than the 0.7 that the luck of which passes get thrown would explain. So we tested the fall before trusting it. A straight line drawn through the first ten seasons, 2006–15, forecast about 20% for the next ten. The next ten actually came in at about 21%. No change at all would have left them near 26%. The fall kept going, just more slowly than the early line promised. Split the seasons into even and odd years, or the games into even and odd weeks, and each half falls on its own.

The fall passes every split we tried. Shares of deep passes thrown down the middle, regular season.
CheckResult
A straight line through 2006–15 forecast 2016–25 at19.6%
2016–25 actually came in at21.4%
No change after 2015 would have meant25.9%
Odd-numbered weeks only, 2006–12 → 2021–2527.8% → 20.1%
Even-numbered weeks only, 2006–12 → 2021–2526.5% → 20.2%
Even-year seasons only, change per season−0.4 points
Odd-year seasons only, change per season−0.6 points
All 20 seasons, change per season−0.5 points (most likely between −0.6 and −0.4)

Because “middle” is a scorer’s call, we also checked whether the scorers changed rather than the passing. Counting only visiting teams’ passes, the middle’s share fell with 28 of the 32 home teams’ stat crews. One crew’s habits can’t make that happen. A change in how crews across the whole league log the middle could, and the data can’t rule that out. The crews get their own section below.

It happened in every game situation we checked

The first suspect is the game around the pass. Maybe teams just found themselves in fewer spots where a deep middle throw makes sense. They didn’t. On first and second down with the game still in doubt, the share went from about 27% to about 20%, the same fall as overall.

One slice could mislead, so we checked every deep pass at once. We sorted each one by down, distance, score, field position and time left. Then we asked what recent teams would have done if they had faced exactly the old mix of situations. Recent teams would still have thrown about 20% down the middle. They faced much the same downs, scores and clocks as before. Handing them the old mix barely moves their figure, from about 20.1% to 20.3%, out of a drop of roughly 7. The other 7 or so points came from teams throwing elsewhere on the same downs and scores.

The drop shows up whether teams were ahead or behind. Teams trailing by 9 or more went from about 28% to 23%. Teams leading by 9 or more fell too, from about 26% to 18%. The shotgun doesn’t explain it either. Its share of deep passes jumped from about half to four in five, but passers in the shotgun and under center both threw down the middle less, by about the same amount: from about 28% to 21% in the shotgun, and from about 27% to 19% under center. The shotgun story has the formation’s own arc.

Read down the table and every row lost about 6 to 8 points.

Every situation lost about the same share of deep middle throws. Share of deep passes thrown down the middle, by game situation, regular season.
Situation2006–122021–25
All deep passes27.2%20.2%

Continues below

Rows 2–7 of 7
Situation2006–122021–25
First and second down, game in doubt26.7%19.5%
Trailing by 9 or more28.3%22.6%
Leading by 9 or more25.7%17.6%
In the shotgun27.6%20.5%
Under center26.8%18.8%
Recent teams, if they had faced the 2006–12 mix of situations—20.3%

The same quarterbacks changed their habits

If it wasn’t the situations, maybe it was the people: a new generation of quarterbacks who simply throw elsewhere. That may be part of it, but the veterans changed too. Fifteen quarterbacks threw at least 150 deep passes in both 2006–12 and 2016–25. Twelve of them threw a smaller share down the middle in the later years. On average their share fell about 4 points, against about 6 for the league over the same years.

Twelve of fifteen long-serving quarterbacks threw less down the middle later in their careers

Share of each passer’s deep passes thrown down the middle, 2006–12 against 2016–25. The 15 passers with 150+ deep passes in both spans.

Show data
Twelve of the fifteen threw a smaller share down the middle later.
2006–122016–25Change
Carson Palmer32.4%21.7%−10.7 pts
Ben Roethlisberger30.8%20.7%−10.1 pts
Eli Manning29.5%19.5%−10.0 pts
Drew Brees31.5%21.6%−9.9 pts
Aaron Rodgers28.0%22.0%−5.9 pts
Alex Smith27.4%21.9%−5.5 pts
Tom Brady31.0%26.2%−4.8 pts
Philip Rivers28.2%23.8%−4.4 pts
Joe Flacco25.7%21.5%−4.2 pts
Matthew Stafford25.2%21.8%−3.4 pts
Andy Dalton26.1%24.2%−1.9 pts
Matt Ryan22.1%21.9%−0.2 pts
Cam Newton21.6%23.8%+2.2 pts
Ryan Fitzpatrick23.7%26.9%+3.2 pts
Andrew Luck20.8%29.5%+8.7 pts
League: 27.2% in 2006–12, 21.4% in 2016–25.

Brees went from about 31% to 22%. Ben Roethlisberger went from about 31% to 21%, Eli Manning from 30% to 20%, and Carson Palmer from 32% to 22%. Three went the other way, led by Andrew Luck (21% to 29%). Counting only road games, where a quarterback’s own home crew isn’t keeping the book, eleven of the fifteen still fell, by about 4 points on average. Brees on the road went from about 35% to 16%.

A second check followed every quarterback from one season to the next, whenever he threw at least 20 deep passes in both. Add up those season-to-season changes and the same passers had moved about 12 points away from the middle by 2025. That is more than the league’s drop of about 7. This check stacks 19 small year-to-year steps, each with its own luck, and the luck piles up: a total like this typically wanders about 5 points by chance, so 12 and 7 are within luck’s reach of each other. Trust the direction more than the size.

Most of those deep middle throws weren’t replaced

Quarterbacks who stopped throwing deep down the middle might have thrown deep outside instead. Mostly they didn’t. Per team per game, deep passes down the middle fell from about 1.9 to about 1.3. Deep passes outside held at about five, and 10- to 14-yard passes to the middle rose only from about 1.2 to 1.3. How far passes travel overall is the air-yards story.

Fewer deep passes go down the middle; outside throws and shorter middle throws rose only slightly

Passes per team per game, by how far they traveled in the air and which way they went. Regular season.

Deep passes down the middle fell from 1.90 a team-game to 1.34; deep passes outside rose from 5.08 to 5.30.

Deep, middle
1.9
1.7
1.3
Deep, outside
5.1
5.6
5.3
Medium, middle
1.2
1.4
1.3
Medium, outside
3.1
3.2
2.8
  • 2006–12
  • 2013–20
  • 2021–25
Outside = left or right. Deep = 15+ yards in the air; medium = 10–14.

The middle did gain ground among those shorter throws. It took about 28% of 10- to 14-yard passes in 2006–12 and about 32% in 2021–25, while in 2021–25 it took only about 22% of 15- to 24-yard passes. Counted per game, the gain is small. Those shorter middle throws went up by about 0.1 of a pass per team per game. The deep middle lost about 0.6.

Defenses changed in the same years. Most defenses play either one safety deep in the middle (“single-high”) or two deep safeties splitting the field (“two-high”, or a “shell”). ESPN reported in 2024, using NFL tracking data that goes back to 2016, that defenses were keeping two safeties deep more and more often. Two deep safeties guard the deep middle. Our play-by-play doesn’t record where safeties line up, so the safety numbers are ESPN’s, and our data can’t say whether the safeties pushed quarterbacks off the middle.

“Quarterbacks have faced coverages featuring two deep safeties on 45.9% of dropbacks this season, according to ESPN Analytics/NFL Next Gen Stats. That’s the highest rate since the data was first tracked in 2016. It continues a steady increase in this area; six seasons ago, in 2018, the rate was 35%.”
— Stephen Holder, ESPN, Oct. 1, 2024

Quarterbacks in that piece described the same shift. Trevor Lawrence: “Even the teams that have typically been single-high [safety] teams are playing more shell, which just changes your passing attack a little bit.” Joe Flacco: “You can still throw the ball down the field versus two-high. But I think it maybe affects the mindset of the playcaller.”

The drop survives a check on who keeps score

If the people charting the passes changed how they draw “the middle”, the whole drop could be a bookkeeping change. Mostly, it isn’t, and one afternoon in 2010 shows how we checked. On Sept. 19, 2010, Matt Schaub’s Houston Texans won 30–27 at Washington’s FedExField. The play-by-play logged nine of Schaub’s deep passes as going down the middle. The people logging them worked for Washington.

Each home team hires its own stat crew. PFF described them in 2013, in a piece about how crews count tackles differently.

“Bias stems from the differences in definition of solo tackles, assists and passes defensed.” … “Rachel Cohen of ESPN detailed how crews are hired by each NFL team’s public relations staff and how the size of the crew varies.”
— PFF, “Home Cookin’ – How NFL Stats Crews Affect IDP, Pt. 1”, April 18, 2013

We checked pass direction for the same kind of home bias, one step at a time:

  1. The home team’s crew logs the direction of every pass in its home games, for both teams.
  2. Each home team plays a different set of visiting teams every year.
  3. Some home teams’ scorers logged more visitors’ passes as middle, year after year. Most of the visitors changed every year and the scorers didn’t, so the habit was the scorers’.
  4. An offense’s own habit repeated less from one year to the next than a home crew’s did.
  5. Teams do have a habit, but it’s weak. Counted only in road games, where other scorers keep the book, an offense’s share repeats from one year to the next only a little more than luck would.
  6. Count the same offense only in its home games, where its own crew keeps the book, and the habit repeats about as strongly as a home crew’s does.

Put numbers on it. A crew that logged lots of visitors’ passes as middle one year usually did again the next. That link scores about 0.4, on a scale from −1 to 1 where 0 means one year tells you nothing about the next. An offense on the road scores barely above 0.1. Hand out the “middle” labels at random within each season, as if drawn from a hat, and you get anything from about −0.06 to 0.07. Offenses do differ on the road, though: the 49ers threw about 31% of their road deep passes down the middle in 2021–25, the Eagles about 12%. Labels from a hat would spread offenses out too, just less. Across all seasons, the real road spread between offenses is about a fifth wider than the hat’s, where repeated draws from the hat land within about 5% of it either way.

Home crews’ habits carry over from year to year; offenses’ road habits barely do. Each figure runs from −1 to 1: 0 means one season tells you nothing about the next, 1 a perfect link. Handing out the “middle” labels at random within each season, as if drawn from a hat, gives −0.05 to 0.07 for offenses (−0.06 to 0.07 on the road), so that is what luck alone looks like.
Whose habitRepeat (−1 to 1)Most likely betweenSeason pairs
Home team’s crew, logging visiting offenses (offense and crew sorted out together)0.430.36 and 0.49605
Home team’s crew, second check (visitors’ own habits taken out)0.390.31 and 0.45563
Offense (offense and crew sorted out together)0.230.16 and 0.31606
Offense, home games only (its own crew)0.380.31 and 0.45607
Offense, road games only (other crews)0.130.05 and 0.21608
Offense, all games0.310.24 and 0.38608
Quarterback, all games0.330.25 and 0.41473

The crews moved closer together, and the drop shows up with nearly all of them

Those crew habits were big. In 2006–12, visitors’ deep passes were logged as middle about 39% of the time at Washington, 38% at Dallas and 37% at the Giants. At Indianapolis and the Jets it was about 17–18%.

Some home teams’ crews logged far more “middle” passes, and almost every one logs fewer now

Each dot is one home team’s crew (three teams changed cities between the two spans): the share of visiting teams’ deep passes it logged as middle, 2006–12 (across) against 2021–25 (up). Neutral-site games left out.

Show data
Visitors at Washington, Dallas and the Giants were logged down the middle on 39%, 38% and 37% of deep passes in 2006–12; at Indianapolis and the Jets, 17% and 18%.
Visitors logged middle, 2006–12Visitors logged middle, 2021–25
Cardinals18%19%
Falcons23%19%
Ravens35%28%
Bills28%23%
Panthers24%15%
Bears23%27%
Bengals35%22%
Browns26%16%
Cowboys39%20%
Broncos26%22%
Lions23%20%
Packers29%30%
Texans29%18%
Colts17%21%
Jaguars28%27%
Chiefs28%15%
Rams27%24%
Chargers24%20%
Raiders28%19%
Dolphins29%19%
Vikings28%23%
Patriots27%16%
Saints22%15%
Giants37%17%
Jets18%17%
Eagles25%21%
Steelers29%27%
Seahawks26%19%
49ers28%18%
Buccaneers26%22%
Titans30%18%
Washington39%24%
  • No change: from (15%, 15%) to (40%, 40%)
Below the dashed line, the share fell; 28 of 32 home teams’ crews sit below it. Each dot rests on 228–458 passes per span.

Since then the crews have moved closer together. Washington, Dallas and the Giants now log visitors down the middle about 24%, 20% and 17% of the time, and the gap between the highest and lowest home crews has narrowed. In 2006–12 a typical home crew sat about 5 points from the average crew, where luck alone would have put it about 2 points away. By 2021–25 the typical distance was about 4 points, against about 2 from luck.

Those three crews explain only a slice of the league’s drop. Leave them out and the share still fell with 25 of the other 29, from about 26% to 21%. So the drop is probably real passing. Still, the data can’t tell a real change in passing apart from every crew redrawing the line at the same time, because both would look the same in the play-by-play.

The middle still pays more when it’s thrown, and costs more

That leaves the second half of the question, whether giving up the deep middle was a mistake. Start with the best deep middle play in our data, Ben Roethlisberger’s 97-yard touchdown to JuJu Smith-Schuster at Ford Field on Oct. 29, 2017. The snap came at Pittsburgh’s own 3-yard line and ended in the end zone. Pittsburgh won that game 20–15. Analysts grade every play by how much it raised or lowered the points the offense could expect from where the drive stood. By that grade, expected points added, the throw was worth about 9 points. An incompletion or an interception costs points. On average, deep passes down the middle were worth about half a point each in every era, and deep passes outside about a third of a point. The middle was ahead in every one of the 20 seasons. Its smallest leads, in 2022 and 2023, were about 0.1 of a point.

When thrown, deep passes down the middle are worth more than those outside

Expected points added per deep pass: roughly how many points a throw is worth to the offense, judged by where it leaves the drive. Incompletions and interceptions count against it. Regular season.

Deep passes down the middle averaged 0.53–0.57 expected points per throw by era; deep passes outside averaged 0.30–0.35.

2006–12
0.5
0.3
2013–20
0.6
0.4
2021–25
0.6
0.4
  • Down the middle
  • Outside
Quarterbacks choose when to throw down the middle, so this describes the throws that were made, not what the middle would give on every play.

That does not mean teams should throw there more. A quarterback throws deep down the middle when the window is open. When it’s closed, he throws somewhere else, checks down or takes a sack, and that play never counts as a deep middle pass. So the middle’s average is built from its best moments. It also costs more when it goes wrong. About 8 deep middle passes in 100 were intercepted in every era, against about 5 in 100 outside (6 in 2006–12). The middle had the higher interception rate in every one of the 20 seasons.

Era by era, the table shows both sides of the bet. Each era the middle is worth more per throw and gets picked off more often, and the ranges around the two averages never overlap.

Expected points added per deep pass, with the range the average most likely sits in and the number of passes, plus interceptions per 100 deep passes. Regular season. Across 593 team-seasons (15+ deep middle and 40+ deep outside passes), a team’s middle share and its middle-minus-outside edge link at −0.10 (95% range −0.18 to −0.02).
SeasonsMiddle: points per pass (passes)Outside: points per pass (passes)Middle: picks per 100Outside: picks per 100
2006–120.53 ±0.05 (6,779)0.30 ±0.03 (18,154)7.8%5.9%
2013–200.57 ±0.05 (6,758)0.35 ±0.03 (22,923)7.6%4.7%
2021–250.55 ±0.07 (3,641)0.35 ±0.03 (14,410)8.2%4.6%

Justin Herbert shows the pattern over a whole career. He rarely throws deep down the middle, about 13% of his 666 deep passes, and the few he does throw have been worth more than his throws outside: about a point each, against about half a point outside. That fits a passer who waits for the window: his middle average comes only from the throws he chose to make there.

One more clue points to quarterbacks picking their moments, though it is faint. Teams that threw down the middle more often got a smaller edge from those throws over their outside ones, as if, like Herbert, the teams that went there rarely went only when it was wide open. We checked 593 team-seasons, one team’s year apiece, like Schaub’s 2009 Texans. On a scale from −1 (more middle throws always means a smaller edge) to 1 (always a bigger one), with 0 for no link at all, the link comes out at about −0.1. The true figure very likely sits between about −0.2 and barely below zero.

On the road, the 49ers throw down the middle most and the Eagles least

Some teams still throw there far more than others. To rank them fairly, count only road games, so no team is helped by its own crew. In 2021–25, Kyle Shanahan’s San Francisco 49ers threw about 31% of their road deep passes down the middle, the most in the league. The Philadelphia Eagles threw about 12%, the fewest. With 32 teams, someone always looks extreme by luck, but both of these are too far out to be luck.

The table shows how far out. Take the 49ers’ 259 road deep passes. Luck alone typically moves a share built on that many throws by about 2.5 points, so call that one step. The 49ers sat about 11 points above the 20% the league average would predict for their seasons, about 4.5 steps. Luck alone usually leaves the most extreme of 32 teams somewhere between about 1.4 and 3.0 steps from average. The Eagles sit about 3.4 steps below average. Luck produces a top team as far out as the 49ers less than once in 1,000 tries, and a bottom team as far out as the Eagles about once in 100 to 200 tries.

Road games only, 2021–25: the five teams that threw the most and the five that threw the least of their deep passes down the middle. “Steps from average” counts how far a team sits from the league, in steps the size of a normal luck swing for its number of passes. With 32 teams, luck alone usually leaves the most extreme team 1.4 to 3.0 steps out.
TeamRoad deep passesDown the middleMost likely betweenSteps from average
49ers25931.3%25.9–37.2+4.5
Packers28326.5%21.7–31.9+2.7
Lions22326.5%21.1–32.6+2.3
Texans28826.0%21.3–31.4+2.5
Dolphins27425.9%21.1–31.4+2.3

Continues below

Rows 6–10 of 10
TeamRoad deep passesDown the middleMost likely betweenSteps from average
Eagles29212.0%8.7–16.2−3.4
Jaguars27813.3%9.8–17.8−2.8
Chargers29913.7%10.3–18.1−2.8
Bills34814.7%11.3–18.8−2.6
Giants25216.3%12.2–21.3−1.5

Jalen Hurts, who threw about nine in ten of the Eagles’ deep passes in those seasons, threw about 13% down the middle on the road over his career. Herbert threw about 15% on the road and about 13% overall. His 2023 season, 5 down the middle in 99 deep passes, is the lowest share of any passer with at least 60 deep passes in a season in our data. Among the 60 quarterbacks with at least 400 deep passes since 2006, only Tyrod Taylor threw a smaller career share down the middle than Herbert, about 11% of 409. Hurts, at about 15% for his career, is third.

The other end belongs to the late 2000s. David Garrard of the Jacksonville Jaguars threw about 41% of his deep passes down the middle, the most of any quarterback with at least 400 deep passes, and about the same on the road. In 2009 alone he threw 58 of 122 there. Schaub’s 36% falls to about 31% on the road, so his home games ran higher. Tom Brady, who threw more deep passes than anyone else in our data, sent about 29% down the middle, home or away.

In 2025, Jared Goff of the Detroit Lions sent 40 of his 97 deep passes down the middle, about 41%. Among passers with 60 or more deep passes in a season, it is the only season after 2013 in the dozen highest in our data. On about 100 passes, a share can swing 10 percentage points either way by chance (Goff’s true rate is most likely between about 32% and 51%). One season is too few to call a comeback for the middle. It is a season to watch.

So far in 2026, the lower level is holding

Goff’s season is one reason to check the new one, because a trend found in old seasons should hold up in new ones. Through week 6 of 2026 (data as of Oct. 27, 2026), about 21% of deep passes went down the middle. That is in line with recent seasons and below every season of 2006–12. It is not falling further, either.

A straight line through 2006–25 forecast about 19% for 2026, most likely between 16% and 22%. With 1,257 deep passes so far, 2026’s true figure is most likely between about 19% and 24%. Every season of 2006–12 sat at 26% or higher.

The 2026 season so far looks like the recent seasons, not the old ones

Share of deep passes thrown down the middle in weeks 1–6 of each season, so 2026 can sit beside the others.

Show data
In weeks 1–6 of 2026, 21.3% of deep passes went down the middle, against 20.5% in the same weeks of 2021–25.
YearWeeks 1–6
200628.2%
200726.5%
200826.3%
200929.4%
201029.3%
201126.2%
201226.1%
201324.8%
201422.9%
201521.6%
201623.1%
201720.4%
201822.0%
201922.4%
202021.4%
202120.2%
202223.2%
202319.3%
202421.2%
202518.5%
202621.3%
  • 2026: 2026, weeks 1–6
2026: the complete weeks in the play-by-play file pulled Oct. 27, 2026.

Six weeks of passes is not many. Some names are high so far, including Brock Purdy (15 of 41), Josh Allen (14 of 38) and Patrick Mahomes (10 of 25). Counts that small swing a lot by chance, so those could be luck until the season fills in.

The payoff pattern has held in 2026 too. Deep passes down the middle have been worth about 0.6 of a point each, against about 0.4 outside. Fourteen of the 268 down the middle were intercepted, against 40 of the 989 outside.

Six weeks in, 2026 carries the same limits as the 20 seasons before it.

Back at Lambeau Field

Fifteen years after that opener, on Oct. 11, 2026, Jordan Love threw a pass 53 yards in the air down the middle to Christian Watson for a touchdown at Lambeau Field. Green Bay beat Chicago 31–7.

Love’s throw was one of 268 deep passes down the middle across the league in the first six weeks of 2026. Of the ones caught, none has traveled farther in the air.

Share this article

Post on X

Quarterback tools

Live Edgehalla tools on the same subject.

All NFL tools

Show the work

Tests written down
Not in advance (exploratory)
Results computed

nflverse play-by-play, regular season 2006–2025, with 2026 weeks 1–6 (the complete weeks in the file pulled Oct. 27, 2026) held out. A deep pass has 15+ air yards and a logged direction; the middle share is middle over left + middle + right. Situation, same-quarterback, stat-crew and luck checks are the “How we measured it” rows below. Names come from the nflverse players file, matched by player id. Descriptive only: no causal claim.

Every test we graded
TestResultVerdict
How we measured it: Passesnflverse play-by-play, regular season 2006–2025: plays of type pass with a pass attempt, not sacks, spikes or two-point tries. Deep = 15+ air yards with a logged direction (left, middle, right). Direction is blank on 6.8% of 2006 passes, 2.9% of 2007 and under 1% from 2008; before 2006 it is blank on 98% or more. The play text calls 10–13% of 15+ yard passes “short” each season, with no drift.Method

More tests below

Rows 2–3 of 25
TestResultVerdict
How we measured it: The drop27.2% (6,779 of 24,933) in 2006–12 against 20.2% (3,641 of 18,051) in 2021–25: −7.0 points, 95% range −8.0 to −6.2 (game-level bootstrap), 25.8% relative. One season’s binomial noise is about ±1.4 points; seasons wander around the trend line by 1.2 points (SD), more than chance alone.Method
How we measured it: SituationsEarly downs with win probability 0.2–0.8, outside the last two minutes of a half: 26.7% (9,030) to 19.5% (6,252). Standardized to the 2006–12 mix of down × distance × score × last two minutes × field zone (241 cells, 94% of passes): 20.3% against 20.1% raw; Kitagawa split mix −0.2, rate −6.9 points. Shotgun share of deep passes 48.7% to 80.0%; middle share in the shotgun 27.6% to 20.5%, under center 26.8% to 18.8%.Method

Continues below

Rows 4–6 of 25
TestResultVerdict
How we measured it: Same quarterbacks15 passers with 150+ deep passes in both 2006–12 and 2016–25: 12 fell, mean −3.8 points (league −5.8); away games only 11 fell, mean −4.0 (Brees away 35.2% to 15.9%). A chain of consecutive seasons (20+ deep passes in both) gives −11.6 points by 2025 with a binomial SE of 5.2: direction only, not size.Method
How we measured it: Where the passes went(per team-game, 2006–12 → 2021–25): deep middle 1.90 → 1.34; deep outside 5.08 → 5.30; 10–14 yards middle 1.18 → 1.28, the middle’s share of those 27.8% → 31.8% (against 22.3% of 15–24 yard passes in 2021–25).Method
How we measured it: Stat crewsAway throws exclude neutral-site games (850 deep passes). A two-way model (offense-season and stadium-season effects together, linear probability, alternating means) gives year-to-year r = 0.43 (95% range 0.36 to 0.49) for stadiums and 0.23 (95% range 0.16 to 0.31) for offenses. An offense counted in away games only repeats at r = 0.13 (95% range 0.05 to 0.21), against −0.06 to 0.07 under luck (middle labels shuffled within each season, 200 runs). Across stadiums, visitors’ middle share spread SD 5.3 points in 2006–12 (noise 2.3) and 3.9 in 2021–25 (noise 2.4). It fell at 28 of 32 stadiums; without Washington, Dallas and the Giants at 25 of 29 (26.1% → 20.6%).Method

Continues below

Rows 7–8 of 25
TestResultVerdict
How we measured it: PayoffExpected points added (EPA) from nflverse, per deep pass. The selection hint: across 593 team-seasons, a team’s middle share and its middle-minus-outside EPA gap correlate at r = −0.10 (95% range −0.18 to −0.02). Herbert: 13.1% of 666 deep passes down the middle, 0.97 EPA a throw there against 0.52 outside. Ben Roethlisberger to JuJu Smith-Schuster: 25 air yards, 97 yards, 8.79 EPA.Method
How we measured it: RankingsTeam rankings use road games only, 2021–25; z compares a team with the league average of its seasons, and the luck range comes from 1,000 redraws of every team. 49ers 31.3% (259, 25.9–37.2, z +4.5); Eagles 12.0% (292, 8.7–16.2, z −3.4); luck’s top team +1.4 to +3.0, bottom −2.9 to −1.4 (90%). Careers: Herbert 13.1% (road 15.3%, 327), Hurts 14.7% (road 13.3%, 300), Tyrod Taylor 10.8% of 409 (lowest of the passers with 400+), Garrard 41.3% (road 40.6%), Schaub 36.2% (road 30.6%), Brady 28.9% (road 29.4%).Method

Continues below

Rows 9–11 of 25
TestResultVerdict
How we measured it: 2026 holdoutWeeks 1–6 (file pulled Oct. 27, 2026): 21.3% (268 of 1,257, 95% range 19.1–23.7); same weeks of 2021–25 20.5%; lowest full 2006–12 season 26.0%. A straight line through 2006–25 predicted 18.8% (95% band 16.0–21.6%). 2026 EPA: 0.60 middle, 0.38 outside; interceptions 14 of 268 and 40 of 989.Method
Middle share of deep passes, 2006–12 vs 2021–2527.2% (24,933) vs 20.2% (18,051); change −7.0 points (−8.0 to −6.2, game bootstrap); 25.8% relativeA quarter fewer
Trend and pre-registered splitslope −0.5 a season (−0.6 to −0.4); line fitted to 2006–15 predicted 19.6% for 2016–25, actual 21.4%, no change 25.9%Fall continued, slower

Continues below

Rows 12–16 of 25
TestResultVerdict
Split halveseven-year seasons −0.4 a season, odd-year −0.6; odd weeks 27.8% → 20.1%, even weeks 26.5% → 20.2%Both halves fall
Season noisebinomial ±1.4 points (95%) per season; residual SD around the trend 1.2 pointsSingle seasons are noise
Situations: neutral downs; standardized mixneutral 26.7% → 19.5%; 2021–25 on the 2006–12 mix 20.3% (raw 20.1%); mix −0.2, rate −6.9 pointsMix explains none
Same passers (150+ deep in both spans)12 of 15 fell (mean −3.8); away only 11 of 15 (mean −4.0); chain index −11.6 by 2025 (binomial SE 5.2)Fell within passers; size uncertain
EPA per deep pass, middle vs outside2006–12 0.53 vs 0.30; 2013–20 0.57 vs 0.35; 2021–25 0.55 vs 0.35; middle ahead every season, smallest gap 0.10 (2023)Gap in every season; selection

Continues below

Rows 17–20 of 25
TestResultVerdict
Interception rate, middle vs outside2006–12 7.8% vs 5.9%; 2013–20 7.6% vs 4.7%; 2021–25 8.2% vs 4.6%; higher in the middle every seasonHigher in the middle
Selection hint: team middle share against its middle-minus-outside EPA gapr −0.10 (95% range −0.18 to −0.02), 593 team-seasons with 15+ middle and 40+ outside deep passesWeak; fragile
Stat crew: stadium vs offense effects (two-way model)stadium r 0.43 (95% range 0.36 to 0.49); offense r 0.23 (95% range 0.16 to 0.31); visitor leave-out residual r 0.39 (95% range 0.31 to 0.45)Crew effect is real
Offense share, away games onlyr 0.13 (95% range 0.05 to 0.21); luck −0.06 to 0.07; spread 1.51× noise (luck 0.90–1.08)Weak but real

Continues below

Rows 21–25 of 25
TestResultVerdict
Home teams’ crews (visiting offenses)SD 5.3 points vs noise 2.3 (2006–12); 3.9 vs 2.4 (2021–25); fell at 28 of 32; without WAS/DAL/NYG 25 of 29Converged; fall is broad
Road team extremes 2021–2549ers z +4.5, Eagles −3.4; luck top +1.4 to +3.0, bottom −2.9 to −1.4; p < 0.001 and 0.01Beyond luck
Holdout: 2026 weeks 1–621.3% (19.1–23.7, 1,257 passes) vs 20.5% same weeks 2021–25; trend predicted 18.8% (16.0–21.6)Lower level holds
Highest passer-seasons (60+ deep passes)the 12 highest run 2007–2013 apart from Jared Goff 2025: 40 of 97 (41.2%, 95% range 32.0–51.2); Herbert 2023 lowest, 5 of 99One season; noise-sized
Charting completenessdirection blank 6.8% (2006), 2.9% (2007), under 1% from 2008; play text “deep” on 87–90% of 15+ yard passes each seasonStable after 2007

Frequently asked questions

Do NFL teams throw deep down the middle less than they used to?

Yes. About 27% of passes of 15 or more yards in the air went down the middle in 2006–12, and about 20% in 2021–25. The drop shows up in every game situation we checked and with 28 of 32 home teams’ stat crews, though a league-wide change in how stat crews log the middle can’t be ruled out.

Which NFL quarterback throws deep down the middle the least?

Among quarterbacks with at least 400 deep passes since 2006, Tyrod Taylor (about 11%), Justin Herbert (about 13%) and Jalen Hurts (about 15%) threw the smallest shares down the middle. Herbert’s 2023, 5 of 99, is the lowest single season in our data among passers with 60 or more deep passes.

Are deep passes over the middle more effective?

They are worth more on average, about half a point each against about a third of a point outside, but they are also intercepted more often. And quarterbacks mostly throw there when it is open, so the average does not show that throwing there more would work.

Who records NFL play-by-play stats?

Each home team’s stat crew. PFF, citing ESPN’s Rachel Cohen, reported in 2013 that the crews are hired by each team’s public relations staff. In our data, the home crew’s judgment shapes how often a pass is logged as “middle”.

More in The long arc

All How Football Changed research

Sources and method

  1. nflverse play-by-play and players file (CC-BY 4.0)
  2. nflverse games file (nflverse/nfldata): dates, stadiums, scores
  3. nflreadr play-by-play data dictionary (pass_location, pass_length, air_yards)
  4. Stephen Holder, “Why passing yards are declining in the NFL”, ESPN, Oct. 1, 2024
  5. “Home Cookin’ – How NFL Stats Crews Affect IDP, Pt. 1”, PFF, April 18, 2013

Research and analysis, not betting advice. Past results do not guarantee future returns; bet only what you can afford to lose.