The Second Year Leap

Through nine weeks there has been a lot that has happened in the NFL, but no story has consistently dominated like the play of the second year quarterbacks. As veterans like Ben Roethlisberger, Eli Manning, and Cam Newton have struggled, three hot young quarterbacks have stepped forward to fill their shoes.

Last week we saw the pinnacle of this performance, as Dak Prescott, Carson Wentz, and Jared Goff combined to throw for 759 yards and 10 touchdowns with no interceptions. All three led their teams to victories, a consistent trend for the league leading Eagles at 8-1, the closely following Rams at 6-2, and the up and down but still competitive Cowboys at 5-3. Wentz is receiving MVP buzz, and Goff is leading one of the most explosive offenses in the NFL, a year after Prescott won Offensive Rookie of the Year.

The most remarkable thing about these performances is that we really didn’t see them coming. Entering the 2017 season the reviews of last year’s quarterback class were decidedly less than positive. Wentz was divisive, after a rookie year with a few glowing ups intermixed with many downs. Goff was largely similar, just without the high points. Even Prescott was met with some skepticism, seen by some as a product of the talent surrounding him.

A lot of people wrote Wentz and Goff off coming into the year. I didn’t quite go that far, but I came close. It’s common to hear people argue not to dismiss a player after their rookie season, but it’s just as common for that advice to be wrong. Players do improve in the NFL, but they rarely go from bad to good in the way that Wentz and Goff have this year.

Quite a few people are sitting on high horses now about sticking with these young passers through their struggles a year ago. And after hearing enough about this, I decided to put some numbers to it. I wanted to take a look at the improvements experienced by Wentz and Goff, to compare them to see just how unusual these sort of jumps are.

Thanks to Pro Football Reference, I was able to pull together a list of every quarterback since the merger that threw at least 100 passes in each of their first two seasons. This gave me 126 players to work with, starting from Dennis Shaw and Terry Bradshaw in 1970 and working up to the three drafted last year.

There are a lot of metrics out there to measure quarterback play, but for something like this I think the best option is to use ANY/A+. If that means nothing to you, I’ll break it down piece by piece. Y/A stands for “yards per attempt”, which is as straightforward as it sounds. The N stands for “net”, which means that sacks are thrown into both the numerator and the denominator. The A at the very beginning is for adjusted, which is where things start getting complicated. This formula adds 20 yards for each touchdown and subtracts 45 for each interception to give an overall value for each dropback the quarterback takes. Finally, the + just means that this number is adjusted based on the league average that year. A value of 100 is league average, while higher numbers indicate better performance. This makes it easier to compare players across eras.

For all 126 quarterbacks I plotted their ANY/A+ in their rookie years versus their ANY/A+ in their second seasons. Take a glance at the chart below and you’ll see almost immediately that there is a pretty good relationship between the two. Generally speaking, quarterbacks who perform better in their rookie years also perform better the following season. This isn’t surprising, but it does lend credence to the theory that what we are seeing from Wentz and Goff is unusual.

 

I’m sure you’ve figured out by now that the orange circles represent the three quarterbacks drafted in 2016. Prescott is the one on the far right, an excellent rookie year that is surpassed only by three others in history (Pat Haden, Dan Marino, and Marc Bulger if you’re curious). Wentz is squarely in the middle of the pack, and Goff is far to the left, beating out only the likes of Bert Jones and Alex Smith.

These three quarterbacks were spread extremely wide over the course of their rookie seasons, but they’re bunched pretty tightly this year. Through nine weeks Goff leads the way with an ANY/A+ of 123 with Wentz a little ways back at 117 and Prescott bringing up the rear with “only” 112. I like ANY/A+, but I think gaps of this size over a nine game sample are mostly meaningless, and it’s fair to say that all three are performing at statistically a very similar level.

It’s clear from the chart that both Wentz and Goff have experienced remarkable improvements. Wentz’s ANY/A+ has jumped by 32 points, in the 92^nd percentile among all second year quarterbacks. Goff has jumped by 69 points, by a significant margin the biggest improvement since the merger. Even when you take into consideration that someone with a bad performance is likely to improve simply by reversion to the mean, Goff still exceeds expectations by 51 points, the second highest of all time.

The improvement by Wentz this year is remarkable. The improvement by Goff is quite literally unprecedented. It’s enough to get me to dig into the components of ANY/A+ to see where this improvement is coming from.

In the table below I’ve laid out for each passer the main statistics that go into ANY/A+. For each I created a similar chart to the one above and calculated how their performance this year stacks up against what we would have expected from their rookie seasons. In each case the number given is a percentile value, with higher numbers suggesting that they outperformed previous rookie to sophomore growths.

	ANY/A+	Y/A+	TD%+	INT%+	SACK%+
Carson Wentz	94%	96%	98%	57%	14%
Jared Goff	98%	97%	66%	72%	98%
Dak Prescott	57%	25%	90%	58%	74%

There are a few things that we can take from this table. First of all, it is impressive to see just how across the board the improvement has been. Both Wentz and Prescott improve by more than the median quarterback in 3 of the 4 categories, and Goff is better than the median in all four. He has increased his touchdowns and his yards per attempt while decreasing his interceptions and his sacks, and he has done it more impressively than the majority of prior rookie quarterbacks.

It's even more interesting to see where players have struggled. Wentz has actually gotten worse from last season to this season when it comes to taking sacks. After being slightly above average with a SACK%+ of 107 last year, he’s putting up a well below average value of 88 this season. But this has come with a sharp increase in Y/A+ and TD%+, suggesting that he is making more big plays deeper down the field. After a very conservative rookie year, Wentz has started attacking defenses, and his play has leapt as a result. His interceptions haven’t improved as much as Goff’s but they’ve still dropped more than the average passer from their first to their second seasons.

Goff’s sack percentage tells an interesting story as well. His touchdowns and interceptions have both turned around fairly drastically, but they pale in comparison to his sack numbers. Goff is not being hit nearly as much as he was a year ago, and it isn’t hard to see why. This offseason the Rams went out and added a pair of veteran free agency linemen, center John Sullivan and tackle Andrew Whitworth, one of the best pass blockers of the past decade.

Both players have stepped up their games, but it’s hard to say how much of that is development on their own merits and how much is the talent brought in around them. In addition to the two linemen, Goff’s receiving corps was completely rebuilt, bringing in Robert Woods via free agency, Sammy Watkins via a trade, and Cooper Kupp via the draft. It’s rare to see an offense undergo this much change from one year to the next, but Philadelphia isn’t far behind, with new receivers for Wentz in Alshon Jeffery and Torrey Smith as well as new running backs LeGarrette Blount and Jay Ajayi.

There really isn’t an easy answer to the question of how much of this improvement is due to the quarterback and how much is due to the talent around him. But we can answer some other questions about how a quarterback’s first and second year productivity corresponds to their career success. Above I listed the worst and best rookie seasons of all time, and it was a fairly mixed bag. Bert Jones and Alex Smith were the only passers with rookie seasons worse than Goff's, and while neither is an all time great, they both became competent starting quarterbacks. The best three rookie seasons belong to Dan Marino, Marc Bulger, and Pat Haden—a Hall of Famer and two middle of the road starters.

To try to sort through this, I went through my list and collected the career ANY/A+ for every quarterback who threw more than 1000 passes in his career. This dropped 33 quarterbacks from my list (sorry, John Skelton), but it still gives a pretty good sample size to work with.

For each of the metrics I listed I ran four different regressions and pulled out the R-squared value. For each metric I took the rookie and sophomore year value and regressed them both against the career numbers for the given metric and for ANY/A+ as a whole. (For those who aren’t familiar with R-squared, it’s a measure of the strength of a relationship. Essentially it’s a number between 0 and 1, with a value closer to 1 meaning the relationship is stronger.)

Below is a table of these R-squared values.

	vs Self		vs ANY/A+
	Rookie	Sophomore	Rookie	Sophomore
ANY/A+	.15	.23	.15	.23
Y/A+	.26	.25	.11	.22
TD%+	.09	.15	.11	.13
INT%+	.14	.20	.02	.03
SACK%+	.49	.43	.03	.02

There are a few things we can take from this table. The numbers are low, but for football stats these are pretty impressive R-squared values. On the whole it appears that a quarterback’s second season is more meaningful to his career success than his first, which isn’t surprising. It’s slightly odd that this same relationship doesn’t appear to hold true for Y/A+, and that INT%+ and SACK%+ have so little predictive power when it comes to total career ANY/A+. This suggests that the drastic turnaround experienced by Goff in these areas may not have much meaning over the full course of his career. The lack of consistency in TD%+ also doesn’t bode well for Wentz, most of whose success has been driven by a potentially unsustainable jump in touchdown rate.

There’s a lot of margin for error in these numbers, and ultimately all of this doesn’t mean much. The only real answers will come as we watch these players play out their career, when we’ll have a chance to see whether their early career jumps were more like Dan Marino’s, or more like Nick Foles’s.

But we have these tools, so we might as well use them. So I ran one final regression, this time using both rookie and sophomore year ANY/A+ regressed against the career value. This regression returned a coefficient of first year ANY/A+ of .13 and a coefficient of second year ANY/A+ of .24, with an R-squared of .27. And if you plug the first two years from Goff, Wentz, and Prescott into this formula, you get the following career results.

	Career ANY/A+
Carson Wentz	102
Jared Goff	100
Dak Prescott	106

So there you have it. I won't guarantee these results, but if they do happen, I will hold this article over everyone for the rest of my life.