Positive Regression Candidates by SQB%

Positive Regression Candidates by SQB%

A few weeks ago, in an effort to find negative regression candidates at quarterback, I looked which players overperformed in 2016 according to Startable Quarterback Percentage (SQB%).  For the sake of completeness, I’m back to examine the other side of the SQB% coin and look for positive regression candidates by comparing 2016 numbers to historical data from 2012 to 2015.

Before we get to the key players due for bounce-backs, let’s circle back to some methodology from the negative regression article linked above:

  • Bust weeks will again be ignored.  With that said, the idea of how to properly qualify a bad fantasy week recently led to an interesting discussion with Joshua Lake on our podcast.  Shameless self-promotion?  Absolutely, but it gave me plenty of food for thought on how to refine my process of identifying bust weeks in the future.
  • Sample size and quality of samples are very important when comparing a single season to multiple past seasons.  To help identify playing time discrepancies between past seasons and 2016, I have included games played numbers in both tables.  Games played doesn’t always tell the whole story, though, so I’ll attempt to point out corner cases where I can.  Shameless copy-and-paste?  Guilty as charged.
  • Here’s the list of quarterbacks left on the cutting room floor for lack of relevant sample size:  Tony Romo, Mike Glennon, Mark Sanchez, Geno Smith, Matt McGloin, Matt Cassel, Chase Daniel, Nick Foles, EJ Manuel, Derek Anderson, Shaun Hill, Drew Stanton, Charlie Whitehurst, Chad Henne, Matt Schaub, Scott Tolzien, Ryan Mallett, and Kellen Clemens.  Shameless excuse to segue to Tony Romo?  I couldn’t resist.

As Aside on Tony Romo

The data for Tony Romo from 2016 is essentially garbage.  He played one meaningless series.  Regardless, the data we have on Romo from 2012 to 2015 can still provide a good baseline of what to expect in 2017.  Houston as his landing spot is a safe bet, and Romo figures to come in on-beat and perform up to his previous standards.  He will be an injury risk for the rest of his career, but expectations of low-end QB1 or high-end QB2 value are reasonable.  Enough about Romo, though.  Let’s dive into some quantifiable regression hopefuls.

Top-20 SQB% Positive Regression

Player# Yrs2012-5
GP Avg.
2016
GP
GP Diff.2012-5
Top-20% Avg.
2016
Top-20%
Top-20%
Diff.
Jay Cutler413.858.7565.50%20.00%45.50%
Josh McCown395455.60%20.00%35.60%
Ryan Fitzpatrick413.814-0.2560.00%35.70%24.30%
Russell Wilson41616078.10%56.30%21.90%
Jameis Winston11616081.30%62.50%18.80%
Drew Brees415.816-0.2585.70%75.00%10.70%
Eli Manning41616059.40%50.00%9.40%
Ryan Tannehill41613370.30%61.50%8.80%
Robert Griffin III312.357.3367.60%60.00%7.60%
Matthew Stafford41616071.90%68.80%3.20%
Ben Roethlisberger414.3140.2573.70%71.40%2.30%
Cam Newton415.5150.582.30%80.00%2.30%

The Off-Year Elite

Injuries spelled doom for Russell Wilson in 2016.  Thomas Rawls and Tyler Lockett couldn’t play full seasons, and depletion along Seattle’s offensive line led to Wilson himself getting hurt.  While he healed over the course of the season, his O-line did not.  Wilson’s fantasy production suffered as a result, but his football ability hasn’t disappeared.  Most drafters rightfully expect positive regression, evidenced by Wilson’s MFL10 ADP of QB5.  He should be a fixture in the second and third rounds of 2QB drafts for 2017.  If the Seahawks can address their blocking issues up front, Wilson has upside to finish as the overall QB1.

Drew Brees actually overperformed in 2016 top-10 SQB% compared to his four-year average, posting ten QB1 finishes versus an average of 8.75 between 2012 and 2015.  On the top-20% side, however, he fell off his baseline of 85.7% and finished in the top-20 only 75% of the time.  This marks the fourth season in a row his top-20% has degraded.  Considering Brees’ age, it might be safe to assume this is confirmation of decline as his hall of fame career winds down, rather than a single suppressed year of startability.  We know the age cliff claims every quarterback eventually.  Will Brees continue to slide slowly toward the abyss or noticeably stumble in 2017?  I’d bet on the former if I was forced to choose, but there are so many other quarterbacks we can pick in fantasy at lower costs.  Brees may not be worth the early-round gambit in upcoming drafts.

Bad, But Not 2016 Bad (a.k.a. V-E-T-S, VETS! VETS! VETS!)

positive regression sqb% jay cutler 2017Jay Cutler is still a free agent as of this writing, but the Jets were sniffing around.  It would be a very J-E-T-S move to sign an aging veteran who might not be very good anymore, but top-20 SQB% indicates Cutler should bounce back to some extent in 2017, assuming he can stay healthy.  In the small five-game sample of 2016, he only cracked the top-20 once (20%).  But before last season, he posted a top-20 finish in just under two-thirds of his games — on par with Alex Smith and Blake Bortles, and better than Eli Manning and Colin Kaepernick.  He’s a backup fantasy quarterback in 2QB and Superflex leagues, but he is draftable if he lands a starting gig.

Of course, after I wrote the above paragraph about New York kicking the tires on Cutler, the Jets went out and signed Josh McCown.  McCown offers streamability as a starter, but whether by injury or poor play, he hasn’t stayed on the field.  His average of nine games played per season is worst among all the regression candidates in this article, and he’s entering his age 38 season.  SQB% predicts better per-game numbers in the upcoming season compared to last, but I expect they’ll come in yet another small sample of games if he even starts at all.

Ryan Fitzpatrick rounds out our tour of once and future Jets.  He of the enormous beard was a colossal disappointment after posting career-best numbers two seasons ago.  Even if SQB% shows a better track record for Fitzpatrick, he’ll have a difficult time landing a starting gig after crapping the bed with such gusto in 2016.  

Top-10 SQB% Positive Regression

Player# Yrs2012-5
GP Avg.
2016
GP
GP Diff.2012-5
Top-10% Avg.
2016
Top-10%
Top-10%
Diff.
Robert Griffin III312.357.3345.90%20.00%25.90%
Ryan Fitzpatrick413.814-0.2532.70%7.10%25.60%
Jay Cutler413.858.7523.60%0.00%23.60%
Eli Manning41616031.30%12.50%18.80%
Philip Rivers41616035.90%18.80%17.20%
Josh McCown395437.00%20.00%17.00%
Cam Newton415.5150.546.80%33.30%13.40%
Russell Wilson41616048.40%37.50%10.90%
Blake Bortles21516-133.30%25.00%8.30%
Andy Dalton41516-138.30%31.30%7.10%
Sam Bradford312.315-2.6727.00%20.00%7.00%
Marcus Mariota11115-445.50%40.00%5.50%
Tyrod Taylor37.315-7.6731.80%26.70%5.10%
Joe Flacco414.516-1.527.60%25.00%2.60%

Only You Can Prevent Recency Bias

Cam Newton’s top-20% last season was on par with the previous four years, but his top-10% stumbled, falling 13.4% off his four-year average.  Looking at 2012-2015, Newton’s top-10 SQB% of 46.8% ranks behind only five active quarterbacks.  They just so happen to be the players Cam has fallen behind in ADP after going first among quarterbacks last season:  Aaron Rodgers, Drew Brees, Andrew Luck, Tom Brady, and Russell Wilson.  

It’s fair to say Newton is now properly valued, but many fake footballers seem to be dismissing his QB1 upside.  In our recent run of head-to-head polls, Cam lost to the five guys ahead of him in ADP plus Matt Ryan and Derek Carr.  This is recency bias at its worst.  Matt Ryan and Derek Carr both just had the best fantasy seasons of their careers, posting 21.72 and 17.88 points per game respectively.  Those performances can’t compete with Cam’s seasonal ceiling from 2015, when he racked up 24.32 points per game.  Furthermore, Newton has the best track record of the three, averaging 6.8 top-10 finishes and 12.6 top-20 finishes per season, compared to 6.2/12.4 for Ryan and 5.0/10.3 for Carr.  Newton should best both in 2017, and considering the misgivings I laid out for Drew Brees above, I could see Cam reclaiming a spot among fantasy’s top-5 passers.

Middle Class Munsons

Roy Munson was the 1979 Iowa state bowling champion who lost his hand in a horrific gambling-related incident.  His name is synonymous with being up the creek without a paddle, so it’s only fitting I kick off this section with a guy named Rivers.  For two seasons in a row, Philip Rivers has seen multiple paddles go overboard into the currents of injured reserve.  He’s performed with admirable consistency despite losing so many offensive weapons in that span.  In 2016, his QB2 floor was as solid as ever, delivering top-20 finishes 75% of the time.  On the other hand, he lacked explosive fantasy weeks, notching only three top-10 weeks.  That’s a top-10 SQB% of only 18.8% compared to an average of 35.9% from 2012 to 2015.  

Who knows how Rivers will be affected after aging another year, moving to a new arena in Los Angeles, and losing Danny Woodhead in free agency.  I don’t put much stock into any of it.  Woodhead didn’t play much in 2016, and the rest of his receiving corps figures to return in good health, including the emergent Tyrell Williams.  If the Chargers can hold things together, I’m on board with Joshua Lake’s annual assertion that Rivers is undervalued.

I’m much more pessimistic about Eli Manning.  Josh and I have discussed Eli’s struggles with consistency at length on our podcast.  Predicting his weekly outputs in-season is a crap shoot, and those hot and cold swings translate to Eli’s seasonal finishes as well.  At the top of this article, I said I wouldn’t talk about bust rates, but I lied.  Manning’s bust percentage of 26.6% was the worst among quarterbacks who played at least 40 games between from 2012 to 2015.  He followed that up with four more worthless weeks last season.  

Eli’s 25% bust rate in 2016 was actually somewhat predictable, considering his previous history.  His disappointing top-10% is what really hurt fantasy owners last season.  After posting seven top-10 finishes in both 2014 and 2015 (43.8%), Manning accomplished the feat only twice in 2016 (12.5%).  It’s fair to expect some positive regression back toward his four-year average of 31.3%, but 2016 wasn’t the first time his fantasy value cratered.  He was even worse back in 2013, when he posted one top-10 finish (6.3%), five top-20 finishes (31.3%), and six bust weeks (37.5%) on a measly 10.65 fantasy points per game.  The law of averages tells us Manning should bounce back in 2017, and the addition of Brandon Marshall should help, but don’t count on a full return to 2014/2015 form.

Our final Munson is Andy Dalton, who has flashed brilliance when he’s had a full arsenal of weapons at his disposal.  Injuries wreak havoc on all teams, but Bengals games with Dalton, A.J. Green, Tyler Eifert, and Giovani Bernard all on the field have been few and far between in recent years.  With a healthy supporting cast, we might dream of Dalton recapturing the magic he found in 2015, when he finished 7 of his 12 games as a top-10 quarterback and 10 of 12 in the top-20 (58.3% and 83.3% respectively).  The chances are slim, though.  Every other season of work on Dalton’s resume points to a baseline of 30-35% top-10 weeks and 60-70% top-20 weeks.

positive regression sqb% andy dalton 2017

Some have begun to draw parallels between Andy Dalton entering 2017 and perception of Matt Ryan as he entered 2016, including yours truly.  SQB% doesn’t hold Dalton in as high of regard as Matt Ryan, though.  Since 2012, Ryan has maintained a safer floor of top-20%.  His mark of 73.0% from 2012 to 2015 ranks ninth among active players who played at least two seasons in that sample.  If we include his awesome performance from last season for the bigger span of 2012-2016, Matt Ryan leapfrogs Russell Wilson, Tom Brady, Carson Palmer, and Ben Roethlisberger to rank fifth with a five-year top-20% mark of 78.5%.  

For comparison, Dalton’s top-20% in that half-decade sample is 68.4%, which ranks 15th.  He’s right between Ryan Tannehill and Joe Flacco by that measure — not exactly elite company.  Nevertheless, Dalton should improve on his 2016 numbers next season.  We can’t count on a jump into the top-5 ala Matt Ryan in 2016, but ascending into the top-10 or -12 isn’t out of the question for the Red Rifle.

Statistical Mirages

Two other quarterbacks appear poised for positive regression based on previous SQB% numbers, but we can discount both fairly easily.  Blake Bortles’ previous two-year average of 33.3% top-10 weeks is inflated by his abnormal 2015 campaign.  Most knew that season was an outlier, even as it was happening.  He predictably suffered major regression, dropping from ten top-10 weeks in 2015 to only four in 2016 (62.5% down to 25.0%).  It’s safe to assume this most recent season is more indicative of his actual baseline — 25-35% top-10 weeks and 60-70% top-20 weeks.  He straddles the borderline between low-end QB2 and solid QB3 in two-passer fantasy.

Robert Griffin III’s 2012 season is another average-skewing statistical artifact.  He’s become notably worse every season since, with his top-10% dropping from 66.7% in 2012 to 38.5% in 2013, then to 22.2% in 2014 and 20.0% in 2016.  His career is on the rocks, and he’s much too far removed from his 2012 self to expect any real regression towards that rookie peak of fantasy success.  It’s a tragic story to this point, and I would love to see Griffin land a gig somewhere new and prove me wrong.

Closing the Session on Regression (and More Shameless Self-Promotion!)

That does it for the positive regression candidates according to SQB%.  I’ll be back later this week to discuss recent NFL happenings and delve further into the topic of recency bias with Joshua Lake on the 2QB Experience podcast.  If this article got your gears turning, give us a shot in audio form and subscribe to the podcast on iTunes, Stitcher, or Podbean.  Whether you like what you hear or not, please rate/review the show and let us know what you think.  Otherwise, thanks for reading and may all your regression be positive.

Greg Smith

Greg Smith is an engineer, co-founder of TwoQBs.com, and enthusiast for the strategy and design of variance-based games.  When he started playing fantasy football in 2001, his home league's small number of teams necessitated starting two quarterbacks.  That necessity has since grown into obsession, making Greg one of the preeminent champions of 2QB and Superflex formats.

Latest posts by Greg Smith (see all)



1 thought on “Positive Regression Candidates by SQB%”

Leave a Reply

Your email address will not be published. Required fields are marked *