How well did fantasy draft strategy to avoid RB and WR busts work?

With the fantasy season in the books for the vast majority of leagues, it’s time to look dispassionately at the major theme of the summer draft season here which was avoiding players on teams that had bad quarterbacks in the early rounds of drafts. 

We noted that this is generally done in fantasy anyway. But the question we were positing was whether it should be a more uniform rule, like zeroRB once was. So what’s the verdict on zeroBadQB for 2017?

 The first test is the most basic — how did the fantasy players do on the five worst passing teams this year? We’re measuring based on yards per pass play, which is the team version of quarterback YPA but instead of passing yards divided by pass attempts, it’s pass yards minus sack yards divided by passing plays (attempts plus sacks).

Through Week 16, the bottom five passing teams were the Browns, Ravens, Broncos, Packers and Colts. If you look at their average ranking in fantasy scoring by the major positions (quarterback, running back, wide receiver and tight end), the average place is 22nd. That includes 26th at quarterback, 19th at running back, 23rd at WR and 20th at TE. 

[Ten things to watch for next fantasy season]

But the Packers are a tricky team because they were essentially two different teams (with and without Aaron Rodgers). So let’s substitute the next worse passing team in yards per pass play for them. The results do not change much. Adding the Giants and subtracting Green Bay the average position is 23rd and the breakdown becomes QB 29, RB 20, WR 27 and TE 15. 

The first thing we notice is that having bottom five QB play is least impactful on tight ends. But that isn’t a big deal with zeroBadQB because it’s designed to avoid players with top 60 picks, top 84 (seven rounds) at the most. And it’s very unlikely that a TE on a team with a bad QB is going to be drafted that highly anyway.

So, yes, we definitely do generally want to avoid fantasy players on teams with bad quarterbacks. But the next hurdle is the biggest — can we predict who those teams will be? 

My approach was to just pick the teams that you think have a 50% chance or greater of being bottom five teams. This of course included all the bad passing from last year that did not seem to seriously upgrade the position (or, in the case of the Bears, were playing a rookie QB). But I said your mileage may vary. Making this call is just part of the drafting process, as personal as your own rankings of players, for example.

My teams were the Rams, Texans, Broncos, Jets, Browns, Bears, 49ers, Jaguars and Eagles. Their average ranking in yards per play is 19th. This is a miss, as I hoped it would be low to mid-20s. The biggest misses were the Rams, Eagles and the Jaguars and for the life of me I can’t figure out some common thread there that would prevent a similar mistake in the future.  

Jared Goff like Carson Wentz was very unusual in getting so much better in their second year. Goff, of course, had a coaching change. When I had Stats, LLC research this question for The Wall Street Journal in 2012, they found just gradual improvement for quarterbacks by year through age 30 with the largest jump coming from age 26 to 27. Bortles, still somehow just 25, had a coaching change. But bad QB teams are very likely to have changes and I tried to address this by casting a wide net that accounted for this uncertainty to some extent — if I thought you were 50% likely to be bottom five or 95%, it didn’t matter; you were still on the list.  

[Advanced analytics hits and misses in 2017]

Bottom line: We do want to avoid all the players on the teams with the worst quarterback play as a rule but need to find a better way to predict who these teams will be.

Also humbling was that the No. 1 overall running back (Todd Gurley) and No. 1 overall WR (DeAndre Hopkins) in standard leagues were on the zeroBadQB list. But none of the other top 60 players were difference makers relative to ADP. Maybe the argument is to carve out an exception for the Texans along with the Rams because they had new/young quarterbacks. But if you limit your list to just returning bottom five(ish) quarterbacks who are not young and who are on the same team, you have a pretty useless model. 

But I don’t even think it’s debatable that the quality of the quarterback is the driving force in offensive performance and thus in fantasy scoring. Look at what’s happened to the 49ers offense since Jimmy Garoppolo has been the starter. They are ninth in QB scoring, 14th in RB points, ninth at WR and seventh at TE. Previously they were 22nd, 19th, 24th and 26th, respectively. And it was completely predictable that with Garoppolo the entire 49ers offense would become much better. That’s the inverse of zeroBadQB, but still its essence.  

And look no further than the Eagles offense without Wentz on Sunday against a bad Raiders defense for more zeroBadQB validation, conceptually.  

Perhaps rather than a red light, green light with teams based on expected quarterback play, the best approach will add a yellow one. So it no longer becomes binary. You’ll have the quarterbacks you are very confident in putting in the red/do not draft that team’s players with premium picks. Green will mean go with the players on teams where we know the QBs are good. And yellow would just be a caution where maybe you merely do not draft above ADP. Had I employed this approach this summer, the Rams and Eagles would have been yellow. But I still would have missed on the Jaguars (again no great loss here plus the benefit of zero Allen Robinson shares). I can’t fairly say what I would have done with the Texans and Bears, but rookie QBs generally are brutal. 

So this is better but not as crisp. We want a strategy to be yes or no, not “maybe.” Alas, football has so many moving parts and thus less predictive certainty. The concept of “true skill level” with any player is far more elusive than in any other major sport. And, sadly, this proved to be the case even at quarterback in 2017. 

The challenge for next year is to find a better predictive tool and there is hope in PACR, which is basically a quarterback’s success in converting air yards from scrimmage into actual passing yards. But I’m finding that it’s far more predictive in-season than from season-to-season. We’ll keep digging this offseason and revisit this and much more heading into 2018.

What to read next