Jump to content
AlNFL19

Non-Rush Linebacker Statistical Projections

Recommended Posts

Posted (edited)

A little while ago, I rolled out a quarterback projection model here that projected a college quarterback's average Approximate Value (ProFootballReference's all-in-one stat) in Years 3-4 (player's "prime" and still on their first contract) based on their college stats. The idea's the same here, but it's a (slightly better-correlating) model to project the AV of non-rush linebackers drafted in the first three rounds.  The model projects not only an AV, but a percent chance that the player reaches the "Pro Bowl" and "Bust" thresholds. The "Pro Bowl" threshold is 10.0 AV, whereas the "Bust" threshold is below 5.0 AV. The model has been remarkably consistent with its projections by percentage during the sample it was built from (2006-2015 drafts), with most coming within 1%:

Probabilities of Hitting Thresholds (2006-2015)
Players Projected To Surpass 5.0 AV
Threshold Reality Projected Avg.
Pro Bowl 14.55% 15.91%
Hit 63.64% 63.93%
Bust 36.36% 36.07%
Players Projected To Fail To Surpass 5.0 AV
Threshold Reality Projected Avg.
Pro Bowl 4.76% 1.18%
Hit 28.57% 27.81%
Bust 71.43% 72.19%

The correlation between projected AV and real average AV is a fairly significant 0.50.

Here's this years, which was touted as a poor class minus the Devins. (Spoiler alert: its order ended up the same as their draft order).

1. DEVIN WHITE, LSU

D. White, LB, LSU
Statistic Figure
Projected AV 7.35
Bust Chance 1.38%
PB Chance 30.06%

The non-rush linebacker projection model is much "nicer" with its projections than the quarterback model, as a bust chance below 50% is uncommon for QBs, but White here hits 1%. White's projection is second since 2006, behind Ryan Shazier's 7.60. Bucs fans should be happy with that. White's projection is also close to that of Luke Kuechly and Aaron Curry.

2. DEVIN BUSH, MICHIGAN

D. Bush, LB, Michigan
Statistic Figure
Projected AV 7.03
Bust Chance 9.64%
PB Chance 26.70%

My model also loves the second of the Devins, just not quite as much. Still, Bush's projection ranks sixth since 2006, just behind Patrick Willis (not a bad name to be linked to). Bush's 26.70% Pro Bowl chance doesn't look impressive, but the model tends to give out low Pro Bowl projections (which has been warranted, seeing the above results comparing real Pro Bowl chances to projected ones). In fact, only Shazier and White have ever crossed a 30.0% chance. Still, the model thinks Bush will be a good NFL linebacker.

3. JAHLANI TAVAI, HAWAII

J. Tavai, Hawaii
Statistic Figure
Projected AV 5.35
Bust Chance 52.98%
PB Chance 9.02%

The model agrees with consensus: there's the Devins, and then there's a huge gap. Tavai, drafted 43rd overall, has a bust probability above 50%, rocketing up from Bush's by over 40%. Tavai's 5.35 projection equals Rey Maualuga's. Though Tavai crosses the 5.0 "bust" threshold, the history of linebackers with projections between 5.0 and 5.5 isn't much better. Tavai will probably bust, per the model, but he's got a decent chance to have a real NFL future ahead of him.

4. GERMAINE PRATT, NORTH CAROLINA STATE

G. Pratt, LB, N.C. State
Statistic Figure
Projected AV 5.23
Bust Chance 56.07%
PB Chance 7.75%

Pratt tested pretty well at the combine and racked up plenty of tackles, so his projection is above 5.23. Still, there's a good (56.07%) chance Pratt doesn't make it at the next level. 

5. SIONE TAKITAKI, BYU

S. Takitaki, LB, BYU
Statistic Figure
Projected AV 5.14
Bust Chance 58.39%
PB Chance 6.81%

Takitaki is the last linebacker in this class to have a projection above 5.0. Takitaki's 119 final-year tackles ranks second to White, and his 125" broad bests the list. Takitaki's 5.14 is pretty close to the projections of Demario Davis and Abdul Hodge.

6. CODY BARTON, UTAH

C. Barton, LB, Utah
Statistic Figure
Projected AV 4.81
Bust Chance 66.91%
PB Chance 3.33%

Barton is a bit of a longshot, per his 3.33% Pro Bowl chance. His middling draft position and relatively poor vertical are two areas that concern the model. Still, Barton could have a shot to make it: his projection is just 0.01 off of DeAndre Levy's 4.82. The only other LB to ever see a 4.81 projection was Miami's Sean Spence.

7. BOBBY OKEREKE, STANFORD

B. Okereke, LB, Stanford
Statistic Figure
Projected AV 4.77
Bust Chance 67.94%
PB Chance 2.91%

Despite being one of the more impressive athletes of the group, Okereke's projection is (just like his draft position) the worst on the list (Rounds 1-3 only). His 4.77 falls between Sean Spence and NaVorro Bowman, the most wild outlier in the data set (Bowman's projection suffered greatly from a seventh-percentile 29.5" vertical jump).

I might take a look at other positions later, but feel free to check out the QB one here (or don't  ¯\_(ツ)_/¯ ).

 

Edited by AlNFL19

Share this post


Link to post
Share on other sites
1 hour ago, AlNFL19 said:

A little while ago, I rolled out a quarterback projection model here that projected a college quarterback's average Approximate Value (ProFootballReference's all-in-one stat) in Years 3-4 (player's "prime" and still on their first contract) based on their college stats. The idea's the same here, but it's a (slightly better-correlating) model to project the AV of non-rush linebackers drafted in the first three rounds.  The model projects not only an AV, but a percent chance that the player reaches the "Pro Bowl" and "Bust" thresholds. The "Pro Bowl" threshold is 10.0 AV, whereas the "Bust" threshold is below 5.0 AV. The model has been remarkably consistent with its projections by percentage during the sample it was built from (2006-2015 drafts), with most coming within 1%:

Probabilities of Hitting Thresholds (2006-2015)
Players Projected To Surpass 5.0 AV
Threshold Reality Projected Avg.
Pro Bowl 14.55% 15.91%
Hit 63.64% 63.93%
Bust 36.36% 36.07%
Players Projected To Fail To Surpass 5.0 AV
Threshold Reality Projected Avg.
Pro Bowl 4.76% 1.18%
Hit 28.57% 27.81%
Bust 71.43% 72.19%

The correlation between projected AV and real average AV is a fairly significant 0.50.

Here's this years, which was touted as a poor class minus the Devins. (Spoiler alert: its order ended up the same as their draft order).

1. DEVIN WHITE, LSU

D. White, LB, LSU
Statistic Figure
Projected AV 7.35
Bust Chance 1.38%
PB Chance 30.06%

The non-rush linebacker projection model is much "nicer" with its projections than the quarterback model, as a bust chance below 50% is uncommon for QBs, but White here hits 1%. White's projection is second since 2006, behind Ryan Shazier's 7.60. Bucs fans should be happy with that. White's projection is also close to that of Luke Kuechly and Aaron Curry.

2. DEVIN BUSH, MICHIGAN

D. Bush, LB, Michigan
Statistic Figure
Projected AV 7.03
Bust Chance 9.64%
PB Chance 26.70%

My model also loves the second of the Devins, just not quite as much. Still, Bush's projection ranks sixth since 2006, just behind Patrick Willis (not a bad name to be linked to). Bush's 26.70% Pro Bowl chance doesn't look impressive, but the model tends to give out low Pro Bowl projections (which has been warranted, seeing the above results comparing real Pro Bowl chances to projected ones). In fact, only Shazier and White have ever crossed a 30.0% chance. Still, the model thinks Bush will be a good NFL linebacker.

3. JAHLANI TAVAI, HAWAII

J. Tavai, Hawaii
Statistic Figure
Projected AV 5.35
Bust Chance 52.98%
PB Chance 9.02%

The model agrees with consensus: there's the Devins, and then there's a huge gap. Tavai, drafted 43rd overall, has a bust probability above 50%, rocketing up from Bush's by over 40%. Tavai's 5.35 projection equals Rey Maualuga's. Though Tavai crosses the 5.0 "bust" threshold, the history of linebackers with projections between 5.0 and 5.5 isn't much better. Tavai will probably bust, per the model, but he's got a decent chance to have a real NFL future ahead of him.

4. GERMAINE PRATT, NORTH CAROLINA STATE

G. Pratt, LB, N.C. State
Statistic Figure
Projected AV 5.23
Bust Chance 56.07%
PB Chance 7.75%

Pratt tested pretty well at the combine and racked up plenty of tackles, so his projection is above 5.23. Still, there's a good (56.07%) chance Pratt doesn't make it at the next level. 

5. SIONE TAKITAKI, BYU

S. Takitaki, LB, BYU
Statistic Figure
Projected AV 5.14
Bust Chance 58.39%
PB Chance 6.81%

Takitaki is the last linebacker in this class to have a projection above 5.0. Takitaki's 119 final-year tackles ranks second to White, and his 125" broad bests the list. Takitaki's 5.14 is pretty close to the projections of Demario Davis and Abdul Hodge.

6. CODY BARTON, UTAH

C. Barton, LB, Utah
Statistic Figure
Projected AV 4.81
Bust Chance 66.91%
PB Chance 3.33%

Barton is a bit of a longshot, per his 3.33% Pro Bowl chance. His middling draft position and relatively poor vertical are two areas that concern the model. Still, Barton could have a shot to make it: his projection is just 0.01 off of DeAndre Levy's 4.82. The only other LB to ever see a 4.81 projection was Miami's Sean Spence.

7. BOBBY OKEREKE, STANFORD

B. Okereke, LB, Stanford
Statistic Figure
Projected AV 4.77
Bust Chance 67.94%
PB Chance 2.91%

Despite being one of the more impressive athletes of the group, Okereke's projection is (just like his draft position) the worst on the list (Rounds 1-3 only). His 4.77 falls between Sean Spence and NaVorro Bowman, the most wild outlier in the data set (Bowman's projection suffered greatly from a seventh-percentile 29.5" vertical jump).

I might take a look at other positions later, but feel free to check out the QB one here (or don't  ¯\_(ツ)_/¯ ).

 

Fantastic job, once again. I'm curious as to the projections of last year's four stud off-ball linebackers - Roquan Smith, Tremaine Edmunds, Leighton Vander Each, and Darius Leonard.

Share this post


Link to post
Share on other sites
34 minutes ago, HoboRocket said:

Fantastic job, once again. I'm curious as to the projections of last year's four stud off-ball linebackers - Roquan Smith, Tremaine Edmunds, Leighton Vander Each, and Darius Leonard.

Here's the 2018 LB class. Leonard, who would probably be the most interesting projection, is left off because he played in the FCS and didn't do the vertical jump or broad jump, meaning most of his projection is incomplete or incompatible.

2018 Non-Rush Linebacker Class Projections
Player Projected AV Bust Chance PB Chance
R. Smith 7.09 8.09% 27.33%
L. Vander Esch 6.68 18.67% 23.01%
T. Edmunds 6.63 19.96% 22.49%
R. Evans 6.14 32.60% 17.33%
M. Jefferson 5.30 54.27% 8.49%
J. Baker 5.28 54.78% 8.28%
F. Warner 5.26 55.30% 8.07%
O. Burks 4.92 64.07% 4.49%
D. O'Daniel 4.48 75.42% <1.00%

Share this post


Link to post
Share on other sites

Man that's a shame that Oren Burks isn't projected higher. He was my favorite sleeper LB. And I completely forgot about Rashaan Evans; he was buried behind Wesley Woodyard and Jayon Brown as a rookie, much to the disappointment of my fantasy football team.

Share this post


Link to post
Share on other sites
28 minutes ago, HoboRocket said:

Man that's a shame that Oren Burks isn't projected higher. He was my favorite sleeper LB. And I completely forgot about Rashaan Evans; he was buried behind Wesley Woodyard and Jayon Brown as a rookie, much to the disappointment of my fantasy football team.

There's hope. He's an athletic linebacker with some upside who compares pretty closely in projection to Christian Kirksey. I liked Burns too last year, in what looked like a great class.

Share this post


Link to post
Share on other sites

Fantastic stuff. Keep it up. I'm looking forward to analysis of other positions.

I value the numerical stuff exponentially above subjectivity, which owns an absolutely miserable record given how similar the college and pro game are. Same number of players on the field, same basic rules, etc. College games publicly available to watch and scrutinize, instead of cloaked and mysterious. In any other profession the same parallel failure rate would never be tolerated. I've always been astonished that subjectivity remains king and the analytical stuff demoted, but it's a good ole boy network so traditions and conventional wisdom die hard. 

It is simply too difficult for a subjective type to stand far above the norm. Even when it's there I'm never fully impressed with it because it is so fragile and undependable. If the guy gets distracted or ill or simply too old, then you've got nothing. Meanwhile a plug and play formula is gold.

We are still in the infancy of this numbers stuff. I'm so impressed that analytics is spreading. By far my favorite guy on YouTube is Jim Cobern of CommonManFootball. He's got the proper bust rate instead of raving about everybody. Granted, this forum is excellent at that also. That's why I prefer it above other subjective oriented venues. The 16 boom and 16 fail thread is proper. Meanwhile the network versions can't contemplate anything but overflowing praise. Just too bad Todd McShay wasn't on that ESPN first round panel, or the Daniel Jones pick would have received the condemnation it warranted.

Sorry for the wander. I realize some of the above won't be appreciated here. Not concerned. Threads like this never earn the appropriate number of responses on a subjective favoring forum. I have a feeling that's why the poster Kent Hullamania didn't appear this year with his traits study thread. I thought that was one of the most interesting draft-related things I had ever seen. He was taking the subjective scouting reports but interpreting them as a whole better than the people who were producing them. 

Share this post


Link to post
Share on other sites

Interesting. So both athletic testing and production go into your projection?

Share this post


Link to post
Share on other sites
8 hours ago, 11sanchez11 said:

Interesting. So both athletic testing and production go into your projection?

Yes. 

Share this post


Link to post
Share on other sites

Curious why Barton is so low? Tested a lot better than some of the guys above him. I like a combination of RAS and sparq when considering athleticism and generally look at:

-quickness

-speed

-explosion; and

-bench is just a checkbox doesn’t matter as much above a certain point. More guys that I’ve done more than might draw flags

Share this post


Link to post
Share on other sites
3 hours ago, lark25 said:

Curious why Barton is so low? Tested a lot better than some of the guys above him. I like a combination of RAS and sparq when considering athleticism and generally look at:

-quickness

-speed

-explosion; and

-bench is just a checkbox doesn’t matter as much above a certain point. More guys that I’ve done more than might draw flags

There's a couple of reasons.

  • The model takes into account draft position (or, for pre-draft projections, expected draft position) because it's relatively significant. Barton's was obviously relatively low.
  • Barton had average college production that alone would project as 5.23 AV, in the bucket of "good but not great"
  • Barton tested much better than some of these other guys in events that have a significantly lower correlation to NFL production than others (bench press, 20 yard shuttle, etc.) and didn't do as well in higher-correlated events like the vertical jump

Share this post


Link to post
Share on other sites
On 5/3/2019 at 4:00 PM, HoboRocket said:

Man that's a shame that Oren Burks isn't projected higher. He was my favorite sleeper LB. And I completely forgot about Rashaan Evans; he was buried behind Wesley Woodyard and Jayon Brown as a rookie, much to the disappointment of my fantasy football team.

He was injured through the first half of the season.  He played a lot in the last six games, and looked very good.  He should be the player he was projected to be in 2019.  Titans will have a crazy good set of ILBs, at least.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×