≡ Menu

Something was missing from the 2017 season: a quarterback who threw a lot of interceptions but also threw for a lot of yards. At a basic level, you might assume that these two statistics would be inversely related. After all, a bad quarterback would throw a lot of interceptions and not throw for a lot of yards, while a good quarterback shouldn’t throw many interceptions but should throw for a lot of yards.

But in a competitive environment, such absolutes rarely hold up. Some quarterbacks will have to be aggressive to be effective (high average yards per pass but also a high interception rate), while some will choose to be conservative (low average yards per pass but also a low interception rate). But in 2017, we missed that. For example:

  • Marcus Mariota ranked had the 4th worst (i.e., highest) interception rate and ranked 14th in net yards per attempt. By 2017 standards, that stands out as aggressive.
  • The quarterbacks who 5th-through-8th in interception rate all ranked below-average in net yards per attempt.  Again, these quarterbacks were not necessarily very aggressive, just not very good.
  • Ben Roethlisberger (12th worst interception rate, 7th best net yards per attempt average), Derek Carr (10th, 15th), Blake Bortles (14th, 12th), and Jameis Winston (13th, 8th) join Mariota as the only passers to rank in the bottom 15 in interception rate and top 15 in net yards per attempt.

Conversely, Tom Brady, Jared Goff, Drew Brees, and Alex Smith all had very low interception rates and very high net yards per attempt averages, with all four ranking in the top 7 of both metrics. Philip Rivers ranked 2nd in NY/A and 9th in interception rate. Case Keenum ranked 4th in interception rate and 10th in NY/A.

Really, the only two outliers when it came to great interception rates were Tyrod Taylor (1st in interception rate, 25th in NY/A) and Jacoby Brissett (tied for 6th in interception rate, 28th in NY/A).

The chart below shows the 32 qualifying passers in 2017. The Y-Axis displays net yards per attempt; the X-Axis displays interception rate.  You’ll notice a lack of any dots in the upper right section of the graph, which is where the very aggressive passers would go:

Here’s the full version which includes player labels:

In the comments to Wednesday’s interesting guest post post from Adam Steele, a lot of discussion centered on Lynn Dickey and his unusual 1983 season. Dickey had the worst interception rate of any passer that year at 6.0%, but he also averaged an insanely high 7.92 net yards per attempt. Dan Fouts led the league at 8.10 NY/A, but 1983 Fouts and 1983 Dickey both ranked at the time in the top 5 of all passers since the merger in net yards per pass attempt.

In 1980, Steve Grogan led all passers with an atrocious 7.2% interception rate; Fouts led the NFL in NY/A at 7.25 by the narrowest of margins: Grogan averaged 7.24 net yards per attempt that season.

In 2000, Kurt Warner was one of just three passers to have an interception rate north of 5.0%; the other two were Troy Aikman (in the final season of his career) and Ryan Leaf (in an averaged Ryan Leaf season). Meanwhile, Warner set a post-merger record by averaging 9.03 net yards per pass attempt, a mark that hasn’t been challenged since.

To recap: This season, zero quarterbacks ranked in the top 10 in net yards per attempt and in the bottom 10 in interception rate (with more interceptions placing a quarterback closer to the bottom).  Mariota, Roethlisberger, and Winston stand out, to me, as the most aggressive passers, but really none were particularly aggressive by historical standards.

As recently as 2016, Ryan Tannehill and Philip Rivers each ranked in the top 10 in NY/A and bottom 10 in interception rate; in 2015,  Roethlisberger and Winston did it (Roethlisberger ranked in the top 3/bottom 3 in a particularly notable gunslinger type of season); in 2014, Mark Sanchez in 8 starts with the Eagles qualified.

In 2013, Andy Dalton was top 10/bottom 10. Josh Freeman and Tony Romo did it in 2012, Carson Palmer and Rivers in 2011, and Eli Manning, Jon Kitna, and Brees in 2010.

I’m not so sure that the lack of a gunslinger in 2017 is a trend, but I do think teams are too focused on avoiding interceptions. Even Roethlisberger, who was perhaps the second most aggressive passer this season by these two stats, threw 5 of his 14 interceptions in one game!

That said, there may still be hope in the form of Deshaun Watson.  The Texans rookie averaged 7.10 NY/A this season, the same as Brady and Smith. But he also had a 3.9% interception rate, which would have ranked 3rd in the NFL had he qualified.  He was also responsible for the gunslinger game of the year: against the Seahawks, he threw 3 interceptions but averaged 13.4 yards per attempt; he is joined by two Hall of Famers as the only passers in the last 30 years to average 12+ yards per attempt in a game where he threw 3+ interceptions.  Watson started just six games in 2017, but he was responsible for just 2 of the 7 games in 2017 where a passer threw 2+ interceptions and averaged 9.0+ yards per attempt on 15+ attempts.

So maybe the gunslinger didn’t disappear in 2017; maybe he only got hurt.

  • Watson just missed the qualifying threshold for attempts, but he seemed to be the definition of a gunslinger in his brief stint. You pointed out his bad int% and high NY/A, but he also led the league in intended air yards per attempt (11.3). Winston and Palmer are also interesting cases. They had low interception rates in a historical sense, but they did rank in the bottom ten at 2.5 and 2.6 percent. Winston ranked first in IAY among qualifying passers (10.7), while Palmer ranked fourth (9.7). Kizer actually ranked fifth in IAY (9.6), which indicates that Jackson probably could have put less pressure on him to convert yardage.

  • Deacon Drake

    Andy Dalton has been victim of one of the worst OC trains since Gruden left town. He may still be fixable. However, most coaches don’t want a gunslingler. They want a vanilla statue that makes their play calling look genius. Aaron Rodgers (and his absences) makes McCarthy and look like a damn fool because he turns garbage plays into success.

    Bottom line: The whole Kaep thing proved that teams and coaches are really only out there for entertainment purposes, not to actually try and make plays to win games. Basically, implement a scheme, pick some guys, and let chips fall. A coach makes an aggressive personnel decision, it’s on him. Coach goes the conservative route- oh the other team made plays and beat us.

    Honestly, Watson fell in the draft because he was a proven playmaker that didn’t fit a specific system. The fact O’Brien is still in Houston running spread concepts and had the stones to go get him is fortunate. Meanwhile, teams were throwing themselves at Trubisky, because they thought… Christ, I don’t know what they were thinking. I guess because he was just good enough to be a QB, but could be groomed into a system guy.

    Borderline teams need a gunslinger to make plays. Alex Smith can dink and dunk all day, but regression eventually catches up, usually in the red zone, and you need a guy to thread the needle rather than checkdown and kick. An aggressive QB steals chunks of yards, takes some of the pressure and burden of skill guys.

    • mrh

      Besides Watson, I think Mahomes fits the future gunslinger model (in his one game, his NY/A was 7.27 and INT rate 2.9%) – notably, the Chiefs traded up to get Mahomes and took him over Watson. Just watching him play, he looks like the anti-Alex Smith.

    • Adam

      Relevant study:


      Long story short, gunslinger QBs have historically been far more successful than risk averse QBs. Unfortunately, front offices, coaches, and the media have yet to figure this out, leading to immense pressure on modern QBs to avoid mistakes at all costs.

    • Richie

      ” The fact O’Brien is still in Houston running spread concepts and had the stones to go get him is fortunate.”

      I think it’s a borderline fireable offense that O’Brien started Savage over Watson in week 1. How could he watch those 2 in practice and not see that Watson was significantly more talented?

      • Deacon Drake

        I completely agree. I understand trying to ease a you player into the game, but, uh, Tom Savage.

      • LightsOut85

        I recall reading earlier this season, that O’Brien knew Savage, personally, before he was on the roster. (Nepotism in the NFL?! Color me shocked).

      • Wolverine

        This is exhibit A for why I think Bill O’Brien is a bad/mediocre coach. This supposed “quarterback whisperer” has had a good to great defense (except this year), and some excellent skill position talent. Yet, he’s subjected the Texans and their fans to a parade of Ryan Fitzpatricks, Ryan Malletts, Brock Osweilers, and Tom Savages.

    • Joseph Holley

      Now, maybe this is true–but when Brees, Brady, Rivers, Roethlisberger, Ryan, and (rechecks chart) Goff, rank highly on this graph with Alex Smith, I think that teams can ask their QB to make plays without getting picked off. 4 of these 7 are playing this weekend, and Goff got knocked out by Ryan. Alex Smith could be playing if his kicker didn’t miss a FG, and Rivers missed the playoffs b/c his kicker couldn’t make FG’s in the 1st 2 games this year.
      So–when all 7 of these QB’s should have made the playoffs (6 did), and 4 make the final 8 (a 5th should have, 1 couldn’t b/c of matchup, and 1 might have if given the opportunity), maybe coaches should be saying, “A downfield passing game with low INT’s is the ticket to play in January” to their QB’s–oh, wait–everyone already knows this.

      • Adam

        Except most of those are not downfield psssing games. Brees and Goff had more YAC than air yards, while Alex, Ben, and Rivers were barely above 50% air yards. These guys weren’t necessarily being aggressive – their receivers just made a lot of plays for them after the catch. That’s great from the team’s perspective, but misses the point of this post. Deshaun Watson was the only QB in 2017 to remotely resemble a gunslinger.

        • Also didn’t help that Andrew Luck didn’t play a down in 2017. He has always had strong gunslinger tendencies

          • Wolverine

            “Also didn’t help that Andrew Luck didn’t play a down in 2017. He has always had strong gunslinger tendencies”

            I’m not even a Cots fan, but one of tne of the things that annoys me to no end is when people criticize Andrew Luck. The Colts roster has been dreadful, and he always knew he had to take risks to pull them up to competence.

        • ammek

          The diminution of the high-risk, high-reward element of throwing a pass is the development that I dislike most in the modern NFL. It can be overstated – Lynn Dickey’s running backs caught about five passes for every six caught by his receivers in 1983 – but I think that by taking much of the risk out of downfield throwing, offenses have removed much of the suspense too.
          It would be interesting to resurrect the old Excitement Index and find a way to apply it back into the 1980s and beyond to see if this is a real phenomenon, and if it’s connected to gunslinging, or if I’m just getting grouchy (and spoiled by too many televised games).

          • Adam

            I’m with you, man. The high risk downfield throw is the most exciting play in football, and I hate how many teams are scared to even attempt such throws anymore. The suspense of watching the QB take a deep drop, not knowing what’s happening down the field, the ball arcing through the air as we hold our breath waiting to see who catches it…that’s mostly absent in today’s NFL, and it makes me sad.

            • Wolverine

              I would argue that the “long bomb” isn’t that high risk, because even if it’s intercepted, it’s (usually) no worse than a punt. Teams really should be trying it more often, as it’s a really low risk/high reward play.

              It’s the intermediate throws into the middle of the field (digs/seams) that are risky as far as interceptions. However, these plays are definitely chain-movers. Instead, we get to watch constant check-downs and screens short of the sticks, with the hope that the receiver breaks a tackle to make a first down.

              • I agree. I’m not sure how you would could incentivize more vertical passing with rule changes or anything, but I wonder if some coaches would change their mind if you lumped punts in with turnovers. Take a team like Indianapolis with Brissett this year. At first glance, it looked like they were very careful with the football – they had the fourth fewest turnovers. But they had the 10th most punts. They would’ve ranked 15th most Punts+Turnovers, so while good in reducing turnovers, below average in the “Wasted Drive” index. Seattle is another one. They were tied for 7th fewest turnovers, but had the 4th most punts, making them 11th Highest on the Punt+TO Wasted Drive index. I think it would perhaps shed more light on the fact that in may ways punts are as bad as turnovers – they both are wasted drives that fail to produce points.

                • Richie

                  I guess they could allow defenders to have more pre-pass contact within 20 yards of scrimmage.

                  • Four Touchdowns

                    Richie, I think if they made the rules more favorable to offensive line play, then you’d see more deep shots. One of the advantages of short passing is that the ball gets out before the QB eats a sack or gets hit.

                    • Richie

                      Maybe. But I think if the line was allowed an advantage, coaches would still prefer to throw the ball short with a higher chance of completion.

                    • Four Touchdowns

                      Perhaps. I think it also depends on the QB and personnel you have. If you’ve got a strong-armed QB and a legit deep threat, you’ll probably want to throw deep more often.

  • Also, Jimmy G finished the year at 2.8% interception rate and a league-high 8.08 NY/A. Not as much of a gunslinger as Watson, but probably would have been the #2 gunslinger if we lowered the cut-offs.

    • Josh Sanford

      I think it’s just a line graph from year to year, unadjusted for era, but it should probably be two lines: the average of the top 3 from each season, with the lower line being the average of the bottom 25% of starters, or something like that.

  • Here’s the same chart but for 1977.

    The X-Axis, instead of going from 0% to 5.5% for interceptions, goes from 2% to 7.5%, since that’s a more appropriate range (Roger Staubach had the best INT rate that season at 2.5%).

    The Y-Axis, instead of going from 4.5 to 8.0, goes from 3.5 to 7.0. Greg Landry had the lowest NY/A at 3.86, while Bradshaw had the highest at 6.73.

    So this chart shifts the axes a little bit, but otherwise shows the same data. Shifting the axes should be irrelevant on how the dots are spread out. But you can see how different 1977 NFL was from 2017.


    • Wolverine

      That’s great stuff. Really illustrates how much the passing game has changed, in a visual form.

  • Josh Sanford

    This is a fantastic article.

    Is there a way you could graph or chart each year’s most gunslinger QB in a single graphic, so we could look at a progession through time (if there is such a progression )?

    • Thanks, Josh. Yeah, I can do something like that. Not sure what the best way to present it is. Do you do era adjusted or no?

  • Wolverine

    This post reminds me of how sad I am that the Bruce Arians retired. When his Cardinals offenses were healthy, they were damn fun to watch.

  • ChasInNJ

    Rex Grossman was run out of the NFL way too soon.

    • Wolverine

      “Fuck it, I’m going deep…”

  • Von Cameron

    very interesting. what does it mean if youre low/low or high/high? is one good or bad? maybe a bubble for each guy sized by their completion percentage (or, in the gunslinger case, the number of pass attempts) would show some separation

  • Pingback: Weekly Sports Analytics News Roundup - January 16th, 2018 - StatSheetStuffer()