The Stats Room: Week 1 QB Projections

The Stats Room: Week 1 QB Projections

This article is part of our The Stats Room series.

With my basic quarterback projections available for this season, I will move onto the hardest part of football projections, playing time. I'll start with quarterbacks because their opportunities are the most consistent, if they're playing. Their opportunity estimates could be reasonably estimated, but the final values heavily regressed.

When estimating the opportunities, I found three measurable options:

1. Using the Vegas gambling lines on total points scored and team spread.
2. Last season's pass attempts per game.
3. This season's pass attempts per game.

Besides the individual measures, I figured some combination of values would work best as the season progress. With it being the season's first week, in-season data can't be used so I will ignore it. Next week, I will compare my results to the standard RotoWire projections and then add the in-season data to Week 2's projections.

Also, I have no method to create a projection for Browns rookie quarterback DeShone Kizer. I will just have to ignore him for now.

To start with, I looked at the how the total points and spread help to estimate the number of pass attempts. The first part of the theory is simple, the more the points a team is projected to score, the more pass attempts and touchdowns for the team. Also, if the point spread is close in these high-scoring games, teams will keep passing to win. If a team is heavily favored, it is more likely to run the ball more to run out the clock.

With my basic quarterback projections available for this season, I will move onto the hardest part of football projections, playing time. I'll start with quarterbacks because their opportunities are the most consistent, if they're playing. Their opportunity estimates could be reasonably estimated, but the final values heavily regressed.

When estimating the opportunities, I found three measurable options:

1. Using the Vegas gambling lines on total points scored and team spread.
2. Last season's pass attempts per game.
3. This season's pass attempts per game.

Besides the individual measures, I figured some combination of values would work best as the season progress. With it being the season's first week, in-season data can't be used so I will ignore it. Next week, I will compare my results to the standard RotoWire projections and then add the in-season data to Week 2's projections.

Also, I have no method to create a projection for Browns rookie quarterback DeShone Kizer. I will just have to ignore him for now.

To start with, I looked at the how the total points and spread help to estimate the number of pass attempts. The first part of the theory is simple, the more the points a team is projected to score, the more pass attempts and touchdowns for the team. Also, if the point spread is close in these high-scoring games, teams will keep passing to win. If a team is heavily favored, it is more likely to run the ball more to run out the clock. The ideas panned out but will have a heavy amount of regression.

Using all the Vegas lines going back to 2000, the projected number of plays per game ranged from 45 (2000-2001 Browns show up quite a bit) to 68 (2000-2001 Rams). But most of the estimates hover around 55. The number of pass attempts per game range from 48 to 61 percent.

So, using just the Vegas lines, the projected number of pass attempts is 27 to 39 per game with the average number about 35. For the first week, Derek Carr and Marcus Mariota have the highest projected pass attempts at 37.4 because it's supposed to be the highest scoring game at 52.5 total points and over/under line at even. Tyrod Taylor takes the bottom spot, with the Jets and Bills predicted to score only 42 points and the Bills expected to win by six.

Using my previously created rate projections, here are the quarterbacks ranked using this standard scoring system:

Passing Yard: 0.04
Passing TD: 4
Turnover: -2
Rushing Yards: 0.1
Rushing TD: 6

PLAYERATTCOMPCOMP%PASS YDSTDINTRUSHESRUSH YDSRUSH TDFUMTOTAL PTS
Aaron Rodgers 26.5 36.3 72.9 269.4 1.9 0.5 3.6 19.7 0.16 0.23 19.6
Russell Wilson 23.5 35.9 65.5 278.3 1.7 0.6 6.0 31.3 0.26 0.23 21.1
Drew Brees 22.3 35.2 63.3 270.9 1.8 0.8 1.7 3.1 0.07 0.22 16.6
Cam Newton 26.3 35.8 73.3 257.8 1.6 0.7 7.0 30.8 0.30 0.23 19.6
Kirk Cousins 25.0 35.7 70.0 286.1 1.7 0.7 2.2 5.9 0.09 0.22 17.7
Tom Brady 28.5 35.9 79.5 274.8 1.8 0.7 2.4 4.9 0.10 0.23 17.3
Matt Ryan 28.3 36.1 78.2 306.2 1.8 0.7 2.4 7.4 0.10 0.23 19.2
Tyrod Taylor 24.0 33.7 71.1 246.4 1.6 0.6 6.6 35.0 0.29 0.21 19.8
Matthew Stafford 26.0 36.3 71.7 272.0 1.7 0.6 2.7 9.9 0.12 0.23 17.7
Joe Flacco 22.3 34.8 64.0 244.3 1.6 0.7 1.8 4.3 0.08 0.22 15.2
Andy Dalton 21.3 34.6 61.4 264.2 1.6 0.6 3.7 11.2 0.16 0.22 17.4
Blake Bortles 18.8 33.2 56.4 222.6 1.4 0.8 3.2 18.8 0.14 0.21 15.4
Alex Smith 21.5 34.7 61.9 258.3 1.6 0.6 4.2 19.2 0.18 0.22 18.2
Carson Palmer 23.5 35.9 65.5 259.8 1.7 0.8 1.7 4.2 0.07 0.23 15.8
Sam Bradford 25.8 35.7 72.1 264.7 1.7 0.6 2.0 5.1 0.08 0.23 16.8
Derek Carr 26.3 37.4 70.2 267.8 1.7 0.7 2.5 7.5 0.11 0.24 17.3
Ben Roethlisberger 28.5 34.8 81.8 269.4 1.7 0.8 1.8 3.2 0.08 0.22 16.4
Philip Rivers 20.3 34.0 59.5 262.8 1.6 0.8 1.6 4.1 0.07 0.21 15.8
Carson Wentz 22.5 35.3 63.7 240.8 1.6 0.7 2.9 9.6 0.12 0.22 15.6
Eli Manning 22.3 35.2 63.3 250.4 1.7 0.8 1.5 3.3 0.06 0.22 15.4
Jameis Winston 22.8 35.4 64.2 250.7 1.6 0.9 3.4 11.9 0.15 0.22 16.2
Marcus Mariota 26.3 37.4 70.2 277.4 1.7 0.8 4.0 23.5 0.17 0.24 19.4
Dak Prescott 28.3 36.1 78.2 276.9 1.7 0.7 3.9 17.2 0.17 0.23 19.0
Trevor Siemian 23.8 34.6 68.6 248.5 1.5 0.7 2.5 6.8 0.11 0.22 15.6
Brian Hoyer 22.3 35.2 63.3 256.6 1.6 0.7 2.1 4.5 0.09 0.22 15.9
Josh McCown 18.0 32.8 54.8 235.3 1.4 0.8 2.4 10.2 0.10 0.21 14.7
Jay Cutler 24.8 35.8 69.2 276.3 1.7 0.8 2.7 12.1 0.11 0.23 17.6
Jared Goff 22.5 35.3 63.7 237.1 1.5 0.8 2.4 8.1 0.10 0.22 14.9
Tom Savage 23.3 33.9 68.5 250.7 1.6 0.6 2.9 8.3 0.13 0.21 16.2
Scott Tolzien 25.5 35.8 71.3 268.2 1.7 0.7 3.3 9.2 0.14 0.23 17.4
Mike Glennon 22.3 35.2 63.3 263.1 1.7 0.7 2.5 10.2 0.11 0.22 17.0

That list is a little non-standard, especially with Tom Brady coming in at No. 17. As can be seen, a ton of regression with not much deliberation from the top and bottom quarterbacks. Even with the disjointed ranking, I had one major takeaway — the rushing yards some quarterbacks accumulate can make a significant difference in their value.

Overall, I was a little disappointed with the results. I expected better.

Now it's time to predict using the previous season's pass attempts per game. The concept behind these projections is that while teams can change coaches and players, pass happy quarterbacks stay pass happy.

For this projection, I used the player's previous season pass attempts per game and then regressed the value. Most of the results pass the eye test, especially for the 2016 regulars.

PLAYERATTCOMPCOMP%PASS YDSTDINTRUSHESRUSH YDSRUSH TDFUMTOTAL PTS
Aaron Rodgers 37.6 24.6 65.4 278.6 1.9 0.6 3.8 20.3 0.16 0.24 20.3
Russell Wilson 33.6 22.2 65.9 261.0 1.6 0.6 5.6 29.4 0.24 0.21 19.8
Drew Brees 40.1 27.3 68.1 308.8 2.0 0.9 1.9 3.5 0.08 0.25 18.9
Cam Newton 33.8 20.0 59.0 243.3 1.5 0.7 6.6 29.1 0.28 0.21 18.5
Kirk Cousins 36.8 24.9 67.5 295.2 1.8 0.7 2.2 6.1 0.10 0.23 18.3
Tom Brady 37.1 23.9 64.3 284.6 1.8 0.7 2.5 5.1 0.11 0.23 17.9
Matt Ryan 33.3 22.7 68.3 282.1 1.7 0.6 2.2 6.8 0.09 0.21 17.7
Tyrod Taylor 30.1 19.2 63.8 219.8 1.4 0.5 5.9 31.2 0.25 0.19 17.7
Matthew Stafford 36.1 23.7 65.8 270.6 1.7 0.6 2.7 9.9 0.12 0.23 17.6
Joe Flacco 40.1 26.3 65.5 281.6 1.8 0.8 2.1 5.0 0.09 0.25 17.5
Andy Dalton 34.8 22.9 65.8 265.4 1.6 0.6 3.8 11.3 0.16 0.22 17.5
Blake Bortles 37.9 22.8 60.1 253.5 1.6 0.9 3.6 21.5 0.16 0.24 17.5
Alex Smith 32.8 21.9 66.6 244.1 1.5 0.5 3.9 18.1 0.17 0.21 17.2
Carson Palmer 38.4 24.0 62.4 278.4 1.8 0.9 1.9 4.5 0.08 0.24 17.0
Sam Bradford 36.0 24.8 69.0 266.8 1.7 0.6 2.0 5.1 0.09 0.23 17.0
Derek Carr 36.4 22.9 63.0 260.3 1.7 0.6 2.4 7.3 0.10 0.23 16.8
Ben Roethlisberger 35.1 23.2 66.1 271.3 1.7 0.8 1.8 3.2 0.08 0.22 16.5
Philip Rivers 35.5 22.7 63.9 273.9 1.7 0.9 1.6 4.3 0.07 0.22 16.4
Carson Wentz 36.9 23.2 62.9 252.0 1.6 0.8 3.0 10.0 0.13 0.23 16.4
Eli Manning 36.8 23.5 63.8 262.1 1.7 0.8 1.6 3.5 0.07 0.23 16.1
Jameis Winston 35.1 21.2 60.3 248.3 1.6 0.9 3.4 11.8 0.14 0.22 16.1
Marcus Mariota 30.8 19.1 61.9 228.4 1.4 0.7 3.3 19.3 0.14 0.19 16.0
Dak Prescott 30.2 19.8 65.5 231.6 1.5 0.5 3.3 14.4 0.14 0.19 15.9
Trevor Siemian 34.4 21.1 61.3 247.0 1.5 0.7 2.5 6.7 0.11 0.22 15.5
Brian Hoyer 33.3 21.2 63.5 243.3 1.5 0.6 2.0 4.3 0.08 0.21 15.0
Josh McCown 33.1 20.1 60.6 237.2 1.5 0.8 2.4 10.2 0.10 0.21 14.9
Jay Cutler 28.8 18.6 64.6 222.2 1.4 0.6 2.1 9.7 0.09 0.18 14.2
Jared Goff 30.2 18.0 59.5 202.9 1.3 0.7 2.1 6.9 0.09 0.19 12.8
Tom Savage 26.4 16.8 63.8 195.0 1.2 0.5 2.3 6.4 0.10 0.17 12.6
Scott Tolzien 21.9 14.1 64.4 164.0 1.0 0.4 2.0 5.6 0.09 0.14 10.7
Mike Glennon 16.1 10.3 64.2 120.4 0.8 0.3 1.2 4.7 0.05 0.10 7.8

While some of the top values seemed regressed, the order seems right. Now, there is a huge issue with the bottom four or five guys. The Bears' Mike Glennon threw only 11 passes in two games last season. Even with some heavy regression, he is projected to only throw 16 passes in the first game. This value is way too low and I was forced to use the attempts projected in the Vegas line table.

Finally, I wrapped up by combining the values from the Vegas line and previous season average attempts. And I got hogwash. Simply, the Vegas line adds zero value at this level.

One issue on why the Vegas line may be better over a full season compared to the first week is that the odds makers in Vegas are trying to get a read on the teams also. They're less sure on how the dynamics of each game will work out.

I would gladly go with the previous season pass attempt projections except where it is clearly off. In those instances, I will use the Vegas line.

And to see how my projections hold up, I will compare them to the weekly projections created here at RotoWire. Here are the final point totals for comparison.

PLAYERVEGAS LINEPREVIOUS
PASS ATT
FINAL PROJECTIONROTOWIRE PROJECTIONDIFF
Aaron Rodgers 19.6 20.3 20.3 25.8 -5.5
Alex Smith 18.2 17.2 17.2 14.4 2.8
Andy Dalton 17.4 17.5 17.5 19.5 -2.0
Ben Roethlisberger 16.4 16.5 16.5 21.8 -5.2
Blake Bortles 15.4 17.5 17.5 16.4 1.1
Brian Hoyer 15.9 15.0 15.0 13.2 1.8
Cam Newton 19.6 18.5 18.5 22.8 -4.3
Carson Palmer 15.8 17.0 17.0 17.5 -0.5
Carson Wentz 15.6 16.4 16.4 17.3 -0.9
Dak Prescott 19.0 15.9 15.9 20.0 -4.2
Derek Carr 17.3 16.8 16.8 21.5 -4.7
Drew Brees 16.6 18.9 18.9 24.1 -5.2
Eli Manning 15.4 16.1 16.1 17.9 -1.8
Jameis Winston 16.2 16.1 16.1 20.2 -4.2
Jared Goff 14.9 12.8 12.8 14.1 -1.3
Jay Cutler 17.6 14.2 14.2 17.3 -3.1
Joe Flacco 15.2 17.5 17.5 16.6 0.9
Josh McCown 14.7 14.9 14.9 5.8 9.0
Kirk Cousins 17.7 18.3 18.3 22.4 -4.1
Marcus Mariota 19.4 16.0 16.0 21.2 -5.3
Matt Ryan 19.2 17.7 17.7 22.9 -5.2
Matthew Stafford 17.7 17.6 17.6 20.3 -2.7
Mike Glennon 17.0 7.8 17.0 14.7 2.4
Philip Rivers 15.8 16.4 16.4 18.4 -1.9
Russell Wilson 21.1 19.8 19.8 20.1 -0.3
Sam Bradford 16.8 17.0 17.0 16.5 0.5
Scott Tolzien 17.4 10.7 10.7 8.8 1.8
Tom Brady 17.3 17.9 17.9 26.2 -8.3
Tom Savage 16.2 12.6 12.6 9.0 3.6
Trevor Siemian 15.6 15.5 15.5 10.3 5.2
Tyrod Taylor 19.8 17.7 17.7 19.3 -1.6
Total 531.8 503.7 513.0 556.4
Average 17.2 16.2 16.5 17.9

Overall, my values are lower than our standard projections, but the values are a little on the high side consider recent average Week 1 quarterback production:

2012: 12.9
2013: 16.2
2014: 14.3
2015: 13.1
2016: 15.1

Now, the actual values may be a bit lower than the expected values if a quarterback or two got injured and did finish the game. Our standard projections have the weekly quarterback point total at 17.9 with my combo one at 16.5.

Now to determine which system is best, I am going to sum the squares of the expected minus the actual values after removing any quarterback who got injured or didn't play. Whichever system has the lowest total value will be the Week 1 winner.

Comparing the lists, many values are close but we disagree on several players. I have Josh McCown generating nine more points (14.9 vs. 5.8) and the standard RotoWire projections have Tom Brady (17.9 vs. 26.2) with eight more.

I'm done with quarterbacks for this week. Next week, I will add Week 1 pass attempts into the equations. Also, I wanted to start working on wide receiver targets, rookie projections and finally get to the pesky tight ends. Truthfully, I am ready for the season to begin to watch some real games instead of the practice games played this past month.

Want to Read More?
Subscribe to RotoWire to see the full article.

We reserve some of our best content for our paid subscribers. Plus, if you choose to subscribe you can discuss this article with the author and the rest of the RotoWire community.

Get Instant Access To This Article Get Access To This Article
RotoWire Community
Join Our Subscriber-Only NFL Chat
Chat with our writers and other RotoWire NFL fans for all the pre-game info and in-game banter.
Join The Discussion
ABOUT THE AUTHOR
Jeff Zimmerman
Jeff is a former RotoWire contributor. He wrote analytics-focused baseball and football articles for RotoWire. He is a three-time FSWA award winner, including the Football Writer of the Year and Best Football Print Article awards in 2016. The 2017 Tout Wars Mixed Auction champion and 2016 Tout Wars Head-to-Head champ, Zimmerman also contributes to FanGraphs.com, BaseballHQ and Baseball America.
Sunday Night Football DFS Breakdown: Los Angeles Rams vs. Philadelphia Eagles
Sunday Night Football DFS Breakdown: Los Angeles Rams vs. Philadelphia Eagles
49ers at Packers: NFL Betting Odds, Picks and Predictions
49ers at Packers: NFL Betting Odds, Picks and Predictions
NFL Odds Picks and Predictions: Chicago Bears vs. Minnesota Vikings
NFL Odds Picks and Predictions: Chicago Bears vs. Minnesota Vikings
New York Giants vs. Tampa Bay Buccaneers NFL Betting Odds, Expert Picks & Props
New York Giants vs. Tampa Bay Buccaneers NFL Betting Odds, Expert Picks & Props