September 29, 2021

The Only Football

Get The Latest NFL News And More!

Eckert: QB Stats Study 3.0 Using Advanced EPA,…

4 min read
Eckert: QB Stats Study 3.0 Using Advanced EPA,...


Today I am continuing my “Quarterback Study Series”, this being the 3rd article of the study. Here are the links to the other studies, I would recommend checking these out if you dive deep into this post (for full context of this study):

Quarterback Study 1.0:

Quarterback Study 2.0:

In the first two studies, we looked at quarterback success in the 2020 regular season, then the last 5 years based on two stat points: EPA/Play = (expected points added) expected points after play – expected points before, and CPOE= (completion percentage over expectation) probability for a complete pass compared to similar situations.

Today, I wanted to dive into playoff performance.  We will be looking at the last 5 seasons (2016-2020), and I figured that since I got some good feedback on the data points I used in the first study overall, I kept them the same so we could get a larger sample size including playoffs. As we learn about the players, I would urge you to remember this study was not what the quarterback can do, but what did they do with the attributes that every quarterback requires (ex. not every quarterback is a rushing threat, but everyone is expected to add points for your team to win, and complete passes).  With that in mind, let’s see what 2016-2020 playoffs looked like:

I hope you are excited to see how we can continue to layer this information to get more value out of it, perhaps learn something about how we’ve previously thought about players, and hopefully get an opinion on if we can use this model somehow to predict future success.  For now, let’s rank them by stats similarly to last time, but with the 5 year playoff sample size, and adding some new value: We’ll use playoff dropback leaders to give us value from a winning/usage/durability/career longevity standpoint, EPA and CPOE ranks for quality of that usage, and Study 1.0 Rk for recency value.

Playoff Dropback Leaders (2016-2020) Playoff EPA Rk Playoff CPOE Rk Study 1.0 Rk (2020) Study 2.0 Rk (’16-’20) Study 3.0 Rk (All Studies)
1. Tom Brady 6 10 4 7 2
2. Patrick Mahomes 4 9 6 5 3
3. Aaron Rodgers 3 2 1 3 1
4. Drew Brees 11 7 1 12
5. Ben Roethlisberger 9 5 17 13 9
6. Russell Wilson 8 4 7 6 4
7. Jared Goff 14 15 15 13
8. Josh Allen 12 11 2 16 10
9. Matt Ryan 2 1 9 2 5
10. Nick Foles 1 3 31
11. Lamar Jackson 15 14 18 14
12. Deshaun Watson 13 12 3 4 6
13. Philip Rivers 7 13 14 8 7
14. Dak Prescott 5 8 9 11
15. Ryan Tannehill 10 6 8 15 8

Another great takeaway from the studies was quantifying the balance of several different layers to get a more overall picture.  For example, many people would have some of the younger “great potential” quarterbacks listed higher, but this ranking puts into context that carrying this strong play over multiple seasons/playoffs consistently will show what younger players need to achieve before I would rank them higher (with their smaller sample size or quality of play).  It also did not totally discount what they have done more recently (for example Patrick Mahomes, less years than some veterans, but already more consistent success than other quarterbacks and already 2nd in playoff dropbacks).  In regard to the Steelers, this should shed some additional light on my optimism that the Steelers and Big Ben can improve compared to the 2020 season.  Will he improve to his peak performance of the last 5 years? Most likely no, but they also do not necessarily need him to with a stronger offensive supporting cast than 2020.

Now for another fun ending outlook, combining what we’ve learned from all three studies. Without any personal opinion, I am going to rank them based on the information from Study 1, (2020 EPA/CPOE) Study 2, (2016-2020 EPA/CPOE), and Study 3 (2016-2020 Playoff EPA/CPOE), then by dropbacks (descending) to keep the usage/durability/career longevity value (playoff dropbacks rated stronger than regular season for winning value):

Top 5 in Study 1, 2, and 3

  1. Aaron Rodgers

Top 10 in Study 1, 2, and 3

  1. Tom Brady
  2. Patrick Mahomes
  3. Russell Wilson
  4. Matt Ryan
  5. Deshaun Watson

Top 15 in Study 1, 2, and 3

  1. Philip Rivers
  2. Ryan Tannehill

Top 20 in Study 1, 2, and 3

  1. Ben Roethlisberger
  2. Josh Allen

Top 15 in 2 out of 3 studies

  1. Drew Brees
  2. Jared Goff
  3. Dak Prescott

Top 20 in 2 out of 3 studies

  1. Lamar Jackson

Study 1.0 Rankings (2020 regular season, for QBS who didn’t make the Study 3.0 Rankings)

  1. Kirk Cousins
  2. Derek Carr
  3. Kyler Murray
  4. Teddy Bridgewater
  5. Justin Herbert
  6. Matthew Stafford
  7. Baker Mayfield

Study 2.0 Rankings (‘16-‘20 regular season, for QBS who didn’t make the Study 3.0 Rankings)

  1. Jimmy Garoppolo
  2. Jameis Winston
  3. Andrew Luck
  4. Ryan Fitzpatrick
  5. Alex Smith
  6. Marcus Mariota
  7. Carson Wentz
  8. Andy Dalton
  9. Tyrod Taylor
  10. Carson Palmer

Only appeared in Study 3.0 Rankings

  1. Nick Foles

I really feel good about this final list and the removal of my personal opinion along with the data points used to put these lists together.  This helped confirm my tape study mostly, but also helped me learn where I might have had personal biases on certain quarterbacks.  As we stand here today, with all the things I’ve learned from the studies, I am comfortable releasing the final list above as a “QB Performance Rankings by 2016-2020 Regular Season + Playoffs”.

This will most likely be the final study on the quarterbacks for now, until the 2021 season begins so we can filter that info in, or possibly add a predictions model prior to the 2021 season beginning.  Please share in the comments your thoughts of the overall study, favorite “quarterback results player comps”, my final “QB Performance Rankings by 2016-2020 Regular Season + Playoffs”, and other topics/player positions you would like me to study in the future!

http://feedproxy.google.com/~r/SteelersDepotBlog/~3/5cnkqI1aVDQ/