With the 2017 fantasy football season now several weeks in the rearview, I thought it would be a healthy and valuable exercise to take a look back and see where we were right and where we were wrong. Who did we overrate incessantly throughout the offseason only to have them bust in the second round? Who was the public way too low on in hindsight? Was it possible to predict breakout seasons from Carson Wentz and Alvin Kamara? Could we have predicted Amari Cooper’s lost season?

This is the first in a multi-part series that will focus on 2017 successes, failures and surprises in order to leverage that information to find ADP (Average Draft Position) values heading into 2018 drafts. Ideally, I’ll be able to revisit the final article(s) again – focusing on potential value picks – in July or August when ADP information is rounding into form. In the first installment, we’ll focus on quarterbacks, looking at players we were right about and those that vastly exceeded expectations, in addition to the passers we ended up overvaluing.

*All ADP information comes from fantasyfootballcalculator.com and is based on 12-team PPR drafts

You can find me on Twitter @eweiner_bball where I’ll likely be tweeting about fantasy football and Malcolm Brogdon throughout the offseason.

What We Got Right

As it turns out, we didn’t get a lot right about quarterback in 2017. Drew Brees, annually one of the safest floor/ceiling combination picks at the position, finished as the per-game QB17, beneath Josh McCown and Case Keenum. Rodgers and Brady, considered in an ADP class of their own, finished sixth and seventh, respectively. Based on ADP, the players the market was the most correct about were Kirk Cousins, Philip Rivers, Matthew Stafford and Dak Prescott. Cousins likely would have been a slight value if he hadn’t lost Jordan Reed and Chris Thompson for the season, but he finished as the QB8 after going ninth at the position. Rivers, Stafford and Dak all finished within three spots of their preseason rank. Considering Cousins, Rivers and Stafford all have reasonably large bodies of work, it makes sense that they’d be easier to predict given we have a large sample size of information to use.

For me, all of this only enhances the notion of paying down at quarterback. As you’ll see in the other quarterback section below, there was massive value to be had at the end of drafts and in free agency this year. While certain players like McCown, Keenum and Deshaun Watson probably serve as outliers, it doesn’t change anything about the late round quarterback strategy. It also serves as a lesson that we don’t know very much heading into the season. Crazy, I know. While it’s important to draft with conviction after doing your own research, it’s also important to remember we’re wrong about this stuff all the freaking time. With coaching and personnel changes, injuries, in-season adjustments and just plain ol’ variance, there’s a lot that can happen in an NFL season.

What We Got Wrong

2017 may go down as the Year of the Undrafted Quarterback. Or at the very least, the undervalued. Don’t believe me? By per-game scoring, the top four fantasy quarterbacks, in order, were Deshaun Watson, Russell Wilson, Carson Wentz and Alex Smith. On average, Russ was an early sixth-round pick, while Wentz went around the 11/12 turn. Smith and Watson weren’t even drafted in your typical 12-team league! If we average their ADPs together (giving Smith and Watson round 16 grades in a typical 15-round draft), their average ADP was the third pick of the 12th round. That’s incredible value! If we bring in Cam Newton, fantasy’s fifth-best signal caller this year, that average pick only drops to the middle of the 11th round. Even in an age where paying down at quarterback is increasingly more common, nobody could have predicted such a successful year from the pre-draft bottom of the barrel. Was any of it foreseeable?

Carson Wentz and Progression

Well, yes and no. Let’s look at Wentz to start. Wentz flashed as a rookie, then saw his efficiency and production tank when the Eagles lost right tackle Lane Johnson to a 10-game suspension. But Wentz was also operating with Jordan Matthews as his lead receiver and Ryan Mathews as the team’s starting running back. The former caught just 25 passes and one touchdown in Buffalo this year, while the latter didn’t play a single NFL snap. Philadelphia reloaded around their young quarterback, adding a true number one receiver in Alshon Jeffery, a vertical-stretching deep threat in Torrey Smith, and LeGarrette Blount, a bruising back who had just led the NFL in rushing touchdowns. The team further moved Nelson Agholor to his more natural position in the slot and drafted an exciting receiver named Mack Hollins out of North Carolina.

Wentz was entering his second year in Doug Pederson’s offense, armed with a top-ten offensive line and a vastly-improved weapons arsenal. He already had a nice rapport with Zach Ertz, a top-ten tight end in his own right, and had always flashed athleticism and an impressive ability to avoid the rush in the pocket. Put it all together, and you get one helluva second season. In hindsight though, it’s probably a good idea to bet on a talent like Wentz when he’s had a year to adjust to the NFL and has an elevated supporting cast. We should specifically aim to do this when the player may be undervalued relative to his market price. Some of you may point to Marcus Mariota, Jameis Winston and Derek Carr and try and tell me progress isn’t always linear, which of course it’s not. But all three of them were priced as QB1s and had projected progression built into their respective ADPS. Your aim for 2018 is finding the players with potential growth who don’t have said progress baked into their price point.

Jared Goff, who finished as the per-game QB10, provides a similar case study. When you turn Kenny Britt, Tavon Austin and Brian Quick into Sammy Watkins, Robert Woods, and Cooper Kupp, improve the offensive line at its most important position and give an offensive guru the keys to the offense, a second-year jump can turn into a giant leap. If we see San Francisco add a wide receiver or two through free agency or the draft, a certain handsome quarterback comes to mind as a similar target for 2018.

Russell Wilson, Alex Smith and Sample Size

Russ was someone I identified as an ADP value heading into the season and hopefully can act as a usable value-finding template moving forward. He had finished as the QB11 in 2016, but if you looked under the hood it was easy to explain his low finish. You can click the link for a longer write-up (please ignore the Eddie Lacy sentence), but he had dealt with multiple injuries throughout the year which led to career lows in rushing, passer rating, YPA and QBR, in addition to sporting a career-low 3.8 TD% that was primed for positive regression back to his career average of 5.6%. On top of all that, he had never had Paul Richardson, Tyler Lockett and Jimmy Graham all healthy at the same time.

Sure enough, Russ’ TD% shot up to 6.1% this year. I pointed out that Russ’ pass attempts had increased by an average of 38 attempts per season. While he only increased this number by 7 this year, it’s clear the Seahawks simply can’t and don’t run the ball like they used to. With a large sample size of past fantasy success and identifiable as a touchdown regression candidate, Russ paints the picture of players we’ll be targeting in 2018 drafts. You know who else has a proven fantasy track record and was below his career TD% this season? Drew Brees.

Alex Smith and Deshaun Watson? Not so much. Just like it’s safe to trust the large sample size of Russ’ work, 2017 stands as an outlier for Smith, another passer with a large performance sample size we can look to for answers. Prior to Smith’s QB4 finish this year, he had never even finished as a QB1. His best overall finish was QB13 in 2013, a year in which he sported a 4.5% TD%, higher than his career 4.0% average. In fact, Smith had only finished as a top-20 fantasy passer four times in his nine years as a starter.

Not only was this unpredictable, but it feels unsustainable. Smith’s 5.1% TD% was a full 1.1% above his career average and feels bound for regression, while this was the first time he’d ever surpassed 3502 yards in a season. The only real takeaway here is to try and target passers with elite playmakers at multiple positions like Travis Kelce and Tyreek Hill, in an offense directed by someone with a proven track record like Andy Reid. Again, these blueprints are meant to lead us to target players in future articles when we have more information, but whoever is quarterbacking for the Vikings and Giants next year will have a slew of playmakers to target.

Watson is tricky. While clear to most he was better than the sloth that is Tom Savage, Watson somehow still didn’t earn the starting job until Week 2 and hadn’t really impressed to that point in the preseason or his Week 1 second half appearance. He proceeded to go nuclear after winning the job, scoring eight more points than anyone else in fantasy from Week 2 through his final game of the season in Week 8. But like I said, he hadn’t flashed to that point and didn’t appear to have NFL-caliber arm strength, and he wasn’t even projected to be the team’s starter. It’s hard to point to his college production as a selling point considering just about every quarterback drafted in the first round of the NFL Draft has stellar numbers on their resume. The lessons here? Bet on young dual-threat talents with an elite receiver to target. Don’t laugh, but DeShone Kizer fits the bill and will be dirt cheap in 2018 if he’s able to win the starting job.

Matt Ryan and Regression to the Mean

Regression to the mean is a relatively simple concept and one that’s important to understand for fantasy football. Essentially, when looking at samples, it means that following an outlier the next expected samples should approach the baseline, or average. If an NBA player averages 12 points per game and scores 30 one night, probability says he is likely to regress towards his mean of 12 points the next night. Vice versa, if he averages 30 a night and only scores 12 in one game, rather than assume he’s in a shooting slump, it’s more likely he’ll regress towards his mean and score much closer to 30 points the next night. This concept is often most applicable when we look at a TD% that’s way above or below a player’s career average, as I’ve referenced several times above.

Matt Ryan was 2017’s most obvious regression candidate heading into the season and as such I didn’t draft him even once. While that’s easy for me to say in hindsight, the numbers back it up. Ryan’s 2016 MVP season was by all accounts an outlier, not just for Ryan’s career but for any quarterback. Ryan’s 10.1 Adjusted Yards gained per pass attempt (AY/A) – a metric that takes the predictive yards/attempt metric and then accounts for touchdowns, interceptions, sacks and throwing the ball away – ranks as the third-best single-season mark ever in the modern era. It was a full 2.4 AY/A higher than his previous career high and 2.7 AY/A above his career average.

But on top of that, Ryan threw for an astronomical career-high 7.1 TD% and a career-low 1.3 INT% in 2016, compared to his career averages of 4.6 TD% and 2.3 INT%. Sure enough, this year Ryan regressed to a 3.8 TD% and 2.3 INT%, more in line with his career norms and nowhere near his 2016 outlier. It might sound petty, but those percentage differences (and Kyle Shanahan leaving) mark the contrast between Ryan’s MVP season and his lowly QB22 finish in 2017. Ryan now has five top-ten finishes to his name, but also five finishes as the QB15 or lower.

It is worth pointing out that there are other variables in consideration that can affect projected regression. Joe Flacco’s YPA has now trended downward for four consecutive seasons. Since his YPA this season was below his career norm, you may be tempted to point to next year as an opportunity for him to regress closer to his mean. While that’s possible, it could also be the case that after 10 seasons Flacco is simply getting worse and unable to make throws that he used to. It’s still important to be analytical while we research these numbers for potential values. Two names I’m already intrigued by heading into 2018, in addition to the aforementioned Brees, are Eli Manning and Marcus Mariota.

We’ll need to see who becomes the head coach of the Giants and Titans (though Pat Shurmur looks to be inline for the New York gig), but it’s safe to assume both teams will look at finding coaches who aim to maximize the strengths of the team’s quarterback. Mike Mularkey was decently successful in Tennesse, but he ran a slow offense with little play action despite Mariota excelling in an up-tempo, play action-heavy offense in college. A year of improvement from rookies Corey Davis, Taywan Taylor and Jonnu Smith should help Mariota take a step forward, and we can assume that Mariota’s price will drop significantly coming off a down year with high expectations. Mariota had just a 2.9 TD% this year despite a career average of 4.6%, so he’s someone I’m projecting positive regression for in 2018.

Manning, meanwhile, is guaranteed to have a super late ADP next year if he’s able to retain the job in New York. He still has Odell Beckham who’s capable of changing a fantasy line in one play, while the emergence of Sterling Shepard and Evan Engram gives him several other young playmakers to target. Given what we saw Shurmur do with Sam Bradford and Case Keenum, New York presents a great opportunity for him to continue building off his success in Minnesota. Manning’s 2017 3.3 TD% was significantly below his career 4.6% average.

Statistics courtesy of pro-football-reference.com, yahoo.com, fantasyfootballcalculator.com, and fantasypros.com