The Michael McCorkle "Mac" Jones Thread

[Dogman]

Etc covers a lot of possibilities.
In the thread someone asks him about QBs who hold the ball creating pressure, that's the part he handwaves.

And no... Evading sacks is not by itself good. Especially because he's not REALLY showing that he's calculating actual sack avoidance, just differential between pressure and sacks, which is doesn't necessarily mean much because not all pressure is as likely to become a sack (early vs late, in pocket or out, etc.)

I'm not entirely sure he does though. A QB holding a ball instead of a 2.4 second release could be because of factors not in a QBs control. Play call, excellent coverage, longer routes. I think a QB holding the ball based on those and then avoiding a sack is valuable. Limited value given what happens after that, but valuable.

I think evading sacks is good as it, by itself, means a much less likelihood of a negative play (ignoring Macs results). I think that differentlial you mention is likely a positive attribute given that Mac has taken sacks when he steps up in the pocket AND avoided sacks when he steps out of the pocket. Again, ignore Macs results as we know he needs to avoid the sack and not do dumb shit like fumble on his own 10 or throw across the field for a pick 6. Heck, the Saints pick 6 was Mac stepping up in the pocket and throwing while getting hit. Damned is he does, damned if he doesn't.

 

[rodderick]

It is almost like-
Last year: Mac is showing too much emotion!
This year: Mac isn't showing enough emotion!

This is a bit silly. I think it's super obvious most of his poor footwork, poor process and bad decision making has come as a response to the possibility of being hit. So yes, if he were less frantic in the face of pressure both mechanically and in terms of decision making it would be a lot better, even if it resulted in more sacks. It's not a necessary tradeoff. QBs who have the skill to avoid sacks don't need to sacrifice playmaking, pocket presence and their process in going through their reads to do so. Mac has done it very clearly.

But yes, I'm holding him to the impossible standard of not completely breaking down over the possibility of being sacked, you guys got me. The kid just can't win.

 

[Dogman]

It's not a gotcha thing though.

We clearly agree his decision making and fundamentals need to be better after avoiding the sack.

 

[Zincman]

Last January, I said to @SMU_Sox that I didn't see in Mac a QB that fit BB's three prerequisites, namely tough, smart (rapid processing) and accurate. We also added that he was harder to evaluate coming out of Alabama since he was mostly throwing from a clean pocket to three future NFLers running free through secondaries. The regression in Mac is stark, as he handles pressure poorly, has lost his accuracy, seems strangely unathletic and looks like he doesn't see the field well at all. At bottom he is currently broken and fixing him may only bring him back to a level of unforgiving mediocrity. Wishcasting will not do it.

 

[SMU_Sox]

Last January, I said to @SMU_Sox that I didn't see in Mac a QB that fit BB's three prerequisites, namely tough, smart (rapid processing) and accurate. We also added that he was harder to evaluate coming out of Alabama since he was mostly throwing from a clean pocket to three future NFLers running free through secondaries. The regression in Mac is stark, as he handles pressure poorly, has lost his accuracy, seems strangely unathletic and looks like he doesn't see the field well at all. At bottom he is currently broken and fixing him may only bring him back to a level of unforgiving mediocrity. Wishcasting will not do it.

Unforgiving mediocrity is such a great combo of words.

Last year my top 2 were AR and Stroud. AR needs to stop getting hurt but both guys have played damn well so far.

We should watch some Maye, Williams, Nix (I don’t like Nix), and Penix soon.

Honestly I am taking a vacation next week and I want to do some light film reviews every night that week if I can. Maybe some QB or OT all-22. Want to watch Joe Alt more. Also want to look at more WRs. Heavy film review starts in November for the draft but would love to get it started next week for QBs light. All are welcome - just shoot me a DM.

Look forward to watching the 2024 class with you, Zinc!

 

[Cellar-Door]

I'm not entirely sure he does though. A QB holding a ball instead of a 2.4 second release could be because of factors not in a QBs control. Play call, excellent coverage, longer routes. I think a QB holding the ball based on those and then avoiding a sack is valuable. Limited value given what happens after that, but valuable.

I think evading sacks is good as it, by itself, means a much less likelihood of a negative play (ignoring Macs results). I think that differentlial you mention is likely a positive attribute given that Mac has taken sacks when he steps up in the pocket AND avoided sacks when he steps out of the pocket. Again, ignore Macs results as we know he needs to avoid the sack and not do dumb shit like fumble on his own 10 or throw across the field for a pick 6. Heck, the Saints pick 6 was Mac stepping up in the pocket and throwing while getting hit. Damned is he does, damned if he doesn't.

It's has nothing to do with Mac or any other QB, my point was the metric itself does not measure anythings valuable because it doesn't differentiate a nunber of very different things. It is sometime good to avoid sacks, it is Aldi sometimes bad to avoid sacks by sacrificing offense. The stat he created did not have enough quality to give any insight. Mac has faced a lot of pressure and not been sacked much, you don't need to the metric to find that, and the metric is poorly designed to such that it adds no context was the point.

 

[Jinhocho]

Anyone on the QB School Patreon? JT O'Sullivan just dropped a 1 hour long breakdown of Mac's performance against the Saints, so if there were some interesting nuggets anyone out there could share about what he saw, it'd be great (yes, I'll probably just pay five bucks and see for myself eventually).

I watched the first quarter of it. I took some notes:

Seems down on BoB
Points out Mac's footwork
Questions why not more runs when running well
Not much separation, but thats the norm with this team. Some guys open but nothing looks easy.
Footwork looked better but still a mess - drops back looking to the right then throws left. "get your feet to the first read".

Saints not respecting Pats receivers or QB so running into heavy numbers.

"What is happening with Mac Jones feet? What is he doing? Where is he going?"

Gesiki looks slow

Jones floats out of pocket when shouldnt - can hide it on film - everyone including o line sees him doing this and its a joke. Makes life harder for oline and everyone on team knows and sees it in film room.

On pick 6, LT and G did not pass off the rush correctly, Jones panicky no idea where he was going but shouldnt have thrown it.

2nd drive
why not run the ball more after screen?
Then they run outside and its a meh play.
Big time but risky throw on long completion to Bourne. Guy in flat was open.
Not a great throw but big time completion.
Slow set up at line - hurry up or some quick plays would help them.
Lack of RPO surprising
Poor TE blocking
Guys open on 3rd 10 but went to the flat with no chance of success instead of slant
Andrews gets beat. Pass pro impacting Jones

 

[FL4WL3SS]

This is too funny
Bleacher Report animated short called the struggling QBs support group that pokes fun at a lot of QBs including Mac.

The league hasn't done these guys any favors. By effectively making the running game useless and this a passing league, there will continue to be terrible QB play and teams that can't overcome it. 10-15 years ago, you could have an average QB, strong running game, and a great defense. You can't build teams like that anymore and this is turning into the NBA where the select few teams that are lucky to have superstars will win and everyone else are nobody's. The NFL is really starting to suck as a product.

 

[rodderick]

Apparently the Patriots have faced the 5th lowest amount of "perfectly covered" plays in the passing game, and while Mac has been one of the worst QBs in the league versus perfect coverage, he's also been the very worst when coverage isn't perfect.




Yet another data point to throw into the offense's struggles. Determining whether or not a play has been "perfectly covered" is obviously subjective. Might be a sign the OL has been disproportionately to blame and play calling/offensive scheming might actually be good given the circumstances.

 

[j44thor]

Unforgiving mediocrity is such a great combo of words.

Last year my top 2 were AR and Stroud. AR needs to stop getting hurt but both guys have played damn well so far.

We should watch some Maye, Williams, Nix (I don’t like Nix), and Penix soon.

Honestly I am taking a vacation next week and I want to do some light film reviews every night that week if I can. Maybe some QB or OT all-22. Want to watch Joe Alt more. Also want to look at more WRs. Heavy film review starts in November for the draft but would love to get it started next week for QBs light. All are welcome - just shoot me a DM.

Look forward to watching the 2024 class with you, Zinc!

A few other names that I hear a lot in this class are Ewers, JJ McCarthy and perhaps the one really catching helium is Jaxson Dart of Ole Miss.
Dart might be the most intriguing as he has a good blend of size/speed and isn't playing with all world talent. All 3 are on the younger side which I consider a plus as well.

 

[BaseballJones]

If the Pats are not seeing Mac as the QB of the future, then obviously they need to go after another one in the 2024 draft. Let's say they draft third. They can get either Harrison or Maye (just for this thought experiment).

Would it be better to grab Maye in the first and then a receiver in the second, OR to grab Harrison in the first and then take two shots at QB by going for, say, Penix in the second and maybe Dart in the third?

 

[Cellar-Door]

If the Pats are not seeing Mac as the QB of the future, then obviously they need to go after another one in the 2024 draft. Let's say they draft third. They can get either Harrison or Maye (just for this thought experiment).

Would it be better to grab Maye in the first and then a receiver in the second, OR to grab Harrison in the first and then take two shots at QB by going for, say, Penix in the second and maybe Dart in the third?

Who went #2 in this scenario?
I think the top 3 are going to be Williams, Maye, Harrison in some order.

 

[BaseballJones]

Who went #2 in this scenario?
I think the top 3 are going to be Williams, Maye, Harrison in some order.

I've seen some mock drafts where OT Fashanu of Penn State is there.

 

[SMU_Sox]

@Cellar-Door I think you just nailed the order. That's the chalk now anyway.

As for the question on what would I do? It isn't ideal to drop a QB into this wasteland but I would take the QB and then commit to investing draft and FA capital into making sure he was protected and had good pass catchers. Obviously you have to buy into the QB though.

 

[SMU_Sox]

I think MH Jr is a better WR than Fashanu an OT but that isn't a knock on Fashanu. MH Jr. is the best WR to come out in a long time.

 

[Arroyoyo]

I think MH Jr is a better WR than Fashanu an OT but that isn't a knock on Fashanu. MH Jr. is the best WR to come out in a long time.

What I love about MH Jr outside of his incredible route running and athleticism is his attitude. You can tell - and not just because of who his father is - he genuinely loves football and approaches it like a true professional even as a collegiate athlete. He appears to have impeccable character, he’s clearly very smart, and he pairs that with incredible athleticism.

I wish the Patriots were even mediocre at QB, because there hasn’t been a player I want to see on the Patriots more than MH Jr in a long, long time.

 

[j44thor]

The league hasn't done these guys any favors. By effectively making the running game useless and this a passing league, there will continue to be terrible QB play and teams that can't overcome it. 10-15 years ago, you could have an average QB, strong running game, and a great defense. You can't build teams like that anymore and this is turning into the NBA where the select few teams that are lucky to have superstars will win and everyone else are nobody's. The NFL is really starting to suck as a product.

NFL has always been a QB league, we just had the best for 2 decades so didn't notice as much. SF nearly won the SB with Jimmy G 4 seasons ago. If anything I think the QB pool has improved over the last 5yrs as QBs like Burrows, TLaw, Tua have come into the league not to mention Purdy and likely Stroud and Richardson's names included before long. The college passing game has become a lot more pro-ready which is improving QB play more than anything.

 

[jodyreeddudley78]

Apparently the Patriots have faced the 5th lowest amount of "perfectly covered" plays in the passing game, and while Mac has been one of the worst QBs in the league versus perfect coverage, he's also been the very worst when coverage isn't perfect.
View attachment 72386

View attachment 72385

Yet another data point to throw into the offense's struggles. Determining whether or not a play has been "perfectly covered" is obviously subjective. Might be a sign the OL has been disproportionately to blame and play calling/offensive scheming might actually be good given the circumstances.

I saw this same tweet and it got me thinking about how much of the fault actually *does* fall on the QBs for these teams. I fully realize that the attached stats come from a fantasy football site, and, let’s be honest, I suspect that there is a heavily subjective element attached to these numbers, but I went to this site ( https://www.fantasypros.com/nfl/advanced-stats-wr.php ) and did a quick and dirty addition of targets and “catchable targets” for WR and TE, then divided catchable targets by targets for a catchable target rate by team. I also totaled both categories for the league to get an average. SSS applies more in football than any other sport for obvious reasons, but it’s not good for Mac.

The league average for “catchable targets” for WR was 69.39%. NE (can’t sort by QB, but Mac has the vast majority of the attempts this year) was at 57.14%. There were only two teams worse than NE: Cleveland and GB.

The league average for “catchable targets” for TE was higher (probably due to the fact that most TE throws have a much lower aDOT) at 78.29%. NE is better here, but still below league average, coming in at 75%.

I then looked at the team teams that are below average in both categories, and there were 7 teams that had below league average results with bth WR and TE. Below are those teams and their total rates between WR and TE:

ATL 66.95%
Houston 68.42%
Indy 69.78%
LV 69.23%
NE 62.68%
NO 62.59%
Pitt 59.35%

Two of these teams are helmed by rookie QBs, with Indy dealing with injuries the AR. The rest are probably all teams that are looking to add a QB in the offseason. And if there is any validity to these numbers, NE has been a bottom 3 team and are likely a semi decent Carr game away from being a bottom two team.

Once again, I realize that fantasy football numbers in no way represent real football, but the results at the top end kind of jived with what you would expect through 5 games (Buff, KC, Detroit, and LAC all do well).

 

[TravisMight]

Hey Everyone,

I'm the author of this particular model and tweet.

A coworker of mine told me it got dropped into this forum and asked if I would hop in to answer any questions and talk about it if you all would like.

This model was birthed from some predictive model building work I was doing on the CFB/NFL Draft prospect side of things, but I was curious as to how it might apply to NFL QBs.

Mac Jones has certainly struggled in other areas this season, that much is unfortunately obvious. And actually, as a rookie, and for even much of last year he wasn't as good in this particular area of sack avoidance, or minimization. But! Adjusting for several key statistically meaningful measures it seems he has taken major steps this season in that regard.

View: https://twitter.com/FF_TravisM/status/1711871005843439983?ref_src=twsrc%5Etfw

 

[Jimbodandy]

I'm fine with drafting a top QB high, even moving up to get him. But if they don't do something different at tackle, may as well start the "When do we move on from Williams/Maye" thread next summer.

edit: thanks for dropping in, Travis.

 

[sezwho]

If the Pats are not seeing Mac as the QB of the future, then obviously they need to go after another one in the 2024 draft. Let's say they draft third. They can get either Harrison or Maye (just for this thought experiment).

Would it be better to grab Maye in the first and then a receiver in the second, OR to grab Harrison in the first and then take two shots at QB by going for, say, Penix in the second and maybe Dart in the third?

I'm thinking differently as I want my young gun to work with a Hoyer/clipboard carrier. Especially given current practice limitations, there are fewer reps and coaching opportunities for developmental QBs, and my focus is purely on what's good for #1.

Also, and I get both sides, I'd think long and hard before putting my new QB in front of this offense. His safety is real (again, three QBs in a row to x-rays in '22 including Hoyer's career ending concussion) and he will look like crap even when not cat-on-a-hot-tin-roofing it. I think the all-22s from JP and others show Mac sucks (yes) AND bad separation AND seemingly not great play designs as well.

If I couldn't stitch together something better functioning on offense by the opener (I'd buy a line and 1-2 WRs then maybe get to D) then I'd probably want the new guy to sit until we have the house in order.

Also, and this is just WR adjacent, but Rham just watched Myers lay it on in the line even in crap seasons...and get let go. Maybe keep a little in the tank?

edit - yes welcome Travis...I suspect you'll get some peer review : )

 

[8slim]

What I love about MH Jr outside of his incredible route running and athleticism is his attitude. You can tell - and not just because of who his father is - he genuinely loves football and approaches it like a true professional even as a collegiate athlete. He appears to have impeccable character, he’s clearly very smart, and he pairs that with incredible athleticism.

I wish the Patriots were even mediocre at QB, because there hasn’t been a player I want to see on the Patriots more than MH Jr in a long, long time.

This cracks me up because (a) Harrison Senior didn't love football at all (I had friends close to the program when he was at Syracuse), and (b) he has/had pretty lousy character (see: car wash shooting).

 

[8slim]

NFL has always been a QB league, we just had the best for 2 decades so didn't notice as much. SF nearly won the SB with Jimmy G 4 seasons ago. If anything I think the QB pool has improved over the last 5yrs as QBs like Burrows, TLaw, Tua have come into the league not to mention Purdy and likely Stroud and Richardson's names included before long. The college passing game has become a lot more pro-ready which is improving QB play more than anything.

Agreed. Plus, expectations about QB play have changed dramatically.

Once upon a time, interceptions weren't nearly the catastrophe that they are perceived to be today. Because the passing game was MUCH more vertical. So many of the "great" QBs from years ago had an absurd amount of picks compared to now. Jim Kelly regularly threw 18-19 INTs a season. Warren Moon threw over 20 multiple times. There isn't the tolerance for that today.

 

[Dogman]

Hey Everyone,

I'm the author of this particular model and tweet.

A coworker of mine told me it got dropped into this forum and asked if I would hop in to answer any questions and talk about it if you all would like.

This model was birthed from some predictive model building work I was doing on the CFB/NFL Draft prospect side of things, but I was curious as to how it might apply to NFL QBs.

Mac Jones has certainly struggled in other areas this season, that much is unfortunately obvious. And actually, as a rookie, and for even much of last year he wasn't as good in this particular area of sack avoidance, or minimization. But! Adjusting for several key statistically meaningful measures it seems he has taken major steps this season in that regard.

View: https://twitter.com/FF_TravisM/status/1711871005843439983?ref_src=twsrc%5Etfw

Hey Travis, thanks for being here. Not sure if you read the discussion regarding your tweet as that is on the previous page or two(depending on the number of posts per page you are set at). There are some questioning the methodology and if there are missing pieces to give us a bit more information on Jones. From the eye test, I certainly didn't expect Jones to be found near Mahomes on this list but, outside of Mac, the list does look pretty close to what we are seeing during the games. Namely, Dan Jones and Watson are pretty terrible at avoiding sacks.

In that tweet thread, you wrote a bit about how the numbers are derived. You mentioned pressure %, drop back rate, sack rates, and then scaling these to come to a positive or negative number. Can you talk a bit more about this?

 

[Arroyoyo]

This cracks me up because (a) Harrison Senior didn't love football at all (I had friends close to the program when he was at Syracuse), and (b) he has/had pretty lousy character (see: car wash shooting).

You realize I’m talking about Marvin Jr., right?

 

[8slim]

You realize I’m talking about Marvin Jr., right?

Yes. You said "You can tell - and not just because of who his father is", which I took to mean that he may have gotten his love for the game and strong character from his Dad.

 

[Auger34]

This cracks me up because (a) Harrison Senior didn't love football at all (I had friends close to the program when he was at Syracuse), and (b) he has/had pretty lousy character (see: car wash shooting).

I was about to bring up the shooting myself...

However, I do find it interesting he didn't love football. I guess because he was quiet and played with Peyton everyone assumed he was some sort of football rat obsessed with the game

 

[TravisMight]

Hey Travis, thanks for being here. Not sure if you read the discussion regarding your tweet as that is on the previous page or two(depending on the number of posts per page you are set at). There are some questioning the methodology and if there are missing pieces to give us a bit more information on Jones. From the eye test, I certainly didn't expect Jones to be found near Mahomes on this list but, outside of Mac, the list does look pretty close to what we are seeing during the games. Namely, Dan Jones and Watson are pretty terrible at avoiding sacks.

In that tweet thread, you wrote a bit about how the numbers are derived. You mentioned pressure %, drop back rate, sack rates, and then scaling these to come to a positive or negative number. Can you talk a bit more about this?

Sure!

I also included some various OL quality metrics derived from PFF, Sports Info Solutions, and SportRadar. Allowed pressure %, pass block win rate, pass block efficiency namely, to compare them. There's some collinearity between them all, so rather than double/triple count the signal there I standardized (via z-scores) and combined them.

When you account for all those variables together you can derive baseline sack probabilities as well as how each QB measures up when adjusting for all the variables working together in a multi-pronged model.

I too found it interesting that Mac had done so well this year by this measure, especially given how his overall performance has been poor. So I backdated this model a few seasons and found he wasn't as good the last two seasons, but overall still was above average since joining the league.

The model isn't perfect though. It pulls an r^2 of 0.52, which is strong, but essentially only accounts for just over 50% of the variance in sack probability. What does that mean? There's an assumed standard error, so the exact order (especially in the middle where several players are close) might be a bit different. But Mac is so far above league expectation that it's still highly likely he is the 2nd best by this measure even accounting for potential standard error adjustments.

 

[Arroyoyo]

Yes. You said "You can tell - and not just because of who his father is", which I took to mean that he may have gotten his love for the game and strong character from his Dad.

No, it was guarding against the “nepotism” retort of “of course he loves football, his dad was Marvin Harrison.”

I was saying “even though his dad is one of the greatest receivers in NFL history, he doesn’t appear to rest on the laurels of his family’s name.”

I wasn’t saying “because MH was a spectacular A+ human, so is MH Jr!”

 

[8slim]

No, it was guarding against the “nepotism” retort of “of course he loves football, his dad was Marvin Harrison.”

I was saying “even though his dad is one of the greatest receivers in NFL history, he doesn’t appear rest his laurels on his family name.”

I wasn’t saying “being MH was a spectacular A+ human, so is MH Jr!”

Gotcha. I wasn't trying to contradict you, just making the observation that it's funny that Junior may love the game and be a decent human. Since his Dad didn't and isn't.

 

[Arroyoyo]

Gotcha. I wasn't trying to contradict you, just making the observation that it's funny that Junior may love the game and be a decent human. Since his Dad didn't and isn't.

No worries, I remember the shooting too. I don’t think there’s a Patriots fan that’s forgotten

 

[Dogman]

@slamminsammya @rodderick @Cellar-Door

You are the most vocal in disagreement of the worth/validity regarding the sack avoidance tweet. @TravisMight is the author of the tweet and here if you wanted to discuss with him his methods, etc. Anything from you?

 

[rodderick]

@slamminsammya @rodderick @Cellar-Door

You are the most vocal in disagreement of the worth/validity regarding the sack avoidance tweet. @TravisMight is the author of the tweet and here if you wanted to discuss with him his methods, etc. Anything from you?

I never had disagreement with the worth/validity of the sack avoidance formula, I just think that Mac has done a whole lot of bad in order to avoid sacks and his reluctance to take hits has led to bad process, bad footwork and bad decisions. I 100% buy that he has been good at converting pressures to sacks, but the means through which he has achieved that have been detrimental to the offense.

 

[TravisMight]

I never had disagreement with the worth/validity of the sack avoidance formula, I just think that Mac has done a whole lot of bad in order to avoid sacks and his reluctance to take hits has led to bad process, bad footwork and bad decisions. I 100% buy that he has been good at converting pressures to sacks, but the means through which he has achieved that have been detrimental to the offense.

Totally agree with Mac’s means in avoiding sacks in some cases being much worse.

In Expected Points Added modeling, sacks are generally a net negative 1.5-2 points. Interceptions? More like 4-5 in the negative direction. And that is where Mac simply can’t keep struggling if he wants to be long for this league.

 

[Bergs]

NFL has always been a QB league, we just had the best for 2 decades so didn't notice as much. SF nearly won the SB with Jimmy G 4 seasons ago. If anything I think the QB pool has improved over the last 5yrs as QBs like Burrows, TLaw, Tua have come into the league not to mention Purdy and likely Stroud and Richardson's names included before long. The college passing game has become a lot more pro-ready which is improving QB play more than anything.

Burrow.

 

[slamminsammya]

Sure!

I also included some various OL quality metrics derived from PFF, Sports Info Solutions, and SportRadar. Allowed pressure %, pass block win rate, pass block efficiency namely, to compare them. There's some collinearity between them all, so rather than double/triple count the signal there I standardized (via z-scores) and combined them.

When you account for all those variables together you can derive baseline sack probabilities as well as how each QB measures up when adjusting for all the variables working together in a multi-pronged model.

I too found it interesting that Mac had done so well this year by this measure, especially given how his overall performance has been poor. So I backdated this model a few seasons and found he wasn't as good the last two seasons, but overall still was above average since joining the league.

The model isn't perfect though. It pulls an r^2 of 0.52, which is strong, but essentially only accounts for just over 50% of the variance in sack probability. What does that mean? There's an assumed standard error, so the exact order (especially in the middle where several players are close) might be a bit different. But Mac is so far above league expectation that it's still highly likely he is the 2nd best by this measure even accounting for potential standard error adjustments.

Strictly speaking the model is only predicting sack rate right? So why do you make the leap of attributing this statistic to QBs?

 

[Cellar-Door]

@slamminsammya @rodderick @Cellar-Door

You are the most vocal in disagreement of the worth/validity regarding the sack avoidance tweet. @TravisMight is the author of the tweet and here if you wanted to discuss with him his methods, etc. Anything from you?

I'm not seeing where the jump from pressures to sack probability is coming from is a big part of it. Pressures vary a great deal in terms of the likelihood they translate into actual sacks, so if the metric is supposed to indicate sack avoidance, it would need to take that into account. I also don't really see where it differentiates between pressures over time, (which might benefit Mac honestly who gets a lot of line created quick pressure, versus someone like Russ who creates pressure).

I think for the most part I don't know what the metric really tells us beyond, who takes the least sacks per pressure.

Not sure that really translates to anything particularly useful given all the variables involved in terms of when sacks are good/bad outcomes, and how the things that decrease sacks may also decrease positive plays.

 

[Dogman]

I'm not seeing where the jump from pressures to sack probability is coming from is a big part of it. Pressures vary a great deal in terms of the likelihood they translate into actual sacks, so if the metric is supposed to indicate sack avoidance, it would need to take that into account. I also don't really see where it differentiates between pressures over time, (which might benefit Mac honestly who gets a lot of line created quick pressure, versus someone like Russ who creates pressure).

I think for the most part I don't know what the metric really tells us beyond, who takes the least sacks per pressure.

Not sure that really translates to anything particularly useful given all the variables involved in terms of when sacks are good/bad outcomes, and how the things that decrease sacks may also decrease positive plays.

@TravisMight can provide context given your thoughts on this. Definitely appreciate another analytical mind around here as someone who puts in the legwork for a living/hobby.

 

[Slocumb If Ya Got Em]

Mac has his issues. But from Welker to Edelman to ???, who is the slot receiver with stones these days? It has been a foundation of the offense for years and it's gone.

 

[Jinhocho]

Mac has his issues. But from Welker to Edelman to ???, who is the slot receiver with stones these days? It has been a foundation of the offense for years and it's gone.

Do not forget Troy Brown. They also had Meyers the last few years.

 

[TravisMight]

Strictly speaking the model is only predicting sack rate right? So why do you make the leap of attributing this statistic to QBs?

One way to think of it is that the model is attempting find what portion of a team's predicted sack rate can be attributed to the QB given how they play in comparison to all other QBs in the league, adjusting for all relevant tested variables in combination together. It's not just sack rate alone.

 

[Slocumb If Ya Got Em]

Do not forget Troy Brown. They also had Meyers the last few years.

Agree, you make a good point.

 

[TravisMight]

I'm not seeing where the jump from pressures to sack probability is coming from is a big part of it. Pressures vary a great deal in terms of the likelihood they translate into actual sacks, so if the metric is supposed to indicate sack avoidance, it would need to take that into account. I also don't really see where it differentiates between pressures over time, (which might benefit Mac honestly who gets a lot of line created quick pressure, versus someone like Russ who creates pressure).

I think for the most part I don't know what the metric really tells us beyond, who takes the least sacks per pressure.

Not sure that really translates to anything particularly useful given all the variables involved in terms of when sacks are good/bad outcomes, and how the things that decrease sacks may also decrease positive plays.

It's much more complex than sacks per pressure. Pressure rate is included, but it's just one piece. It's helpful to find base level expected sack rate given pressure rate, a handful of standardized OL quality metrics standardized and combined to correct for collinearity, dropback volume, sack rate, etc. working all together after all have been verified as sticky independent variables that maintain stability when combined into a singular model. From there it's easy to find sack probability over expected for each individual QB.

I agree that if we had spatial tracking data we could utilize some k-means cluster analysis to group types of pressures into archetypes, which would help qualitatively grade pressures differently. That would almost certainly increase the model's accuracy given that pressures aren't all created equal.

But as is, an r^2 of 0.52 is pretty strong through 5 games. And actually previous seasons suggest that the model stabilizes and becomes more accurate later in the year (likely due to OL quality grades becoming more accurate on larger samples). Every other season's complete year statistical signal registered above 0.6 (explaining 60%+ of the variance in sack probability), which is pretty solid.

 

[slamminsammya]

One way to think of it is that the model is attempting find what portion of a team's predicted sack rate can be attributed to the QB given how they play in comparison to all other QBs in the league, adjusting for all relevant tested variables in combination together. It's not just sack rate alone.

you are predicting sacks - what part of the model is QB specific and not just team specific?

also is there any attempt to control for confounding variables related to play type eg playaction, down/distance, blitz/no blitz?

edit: it feels like measuring baseball runs allowed controlling for opponent quality, and then presenting the result as a measure of shortstop defense. am I misunderstanding?

 

[DeJesus Built My Hotrod]

@TravisMight are there any notable examples of QBs improving in this area dramatically year over year or at least over a career?

 

[slamminsammya]

@TravisMight are there any notable examples of QBs improving in this area dramatically year over year or at least over a career?

Piggybacking on this, have you tried a mixed effects model to see whether you can attribute the remainder of the variane to the QB specifically?

 

[Jimbodandy]

It's much more complex than sacks per pressure. Pressure rate is included, but it's just one piece. It's helpful to find base level expected sack rate given pressure rate, a handful of standardized OL quality metrics standardized and combined to correct for collinearity, dropback volume, sack rate, etc. working all together after all have been verified as sticky independent variables that maintain stability when combined into a singular model. From there it's easy to find sack probability over expected for each individual QB.

I agree that if we had spatial tracking data we could utilize some k-means cluster analysis to group types of pressures into archetypes, which would help qualitatively grade pressures differently. That would almost certainly increase the model's accuracy given that pressures aren't all created equal.

But as is, an r^2 of 0.52 is pretty strong through 5 games. And actually previous seasons suggest that the model stabilizes and becomes more accurate later in the year (likely due to OL quality grades becoming more accurate on larger samples). Every other season's complete year statistical signal registered above 0.6 (explaining 60%+ of the variance in sack probability), which is pretty solid.

I feel a little smarter having read this. Thanks for the details.

 

[TravisMight]

@TravisMight are there any notable examples of QBs improving in this area dramatically year over year or at least over a career?

Great question! Yes, actually. Year over year ~80% of QBs are within 2-3% of their previous season normally, more than half of QBs stay within 1% of previous year. However, with younger QBs we often do see improvements, and if we don't they typically wash out of the league. This seems to stabilize for the most part once players have been in the league a few years.

Justin Fields (so far) has made the most significant positive strides this year by this measure. Mac Jones is actually better by this measure than he was a year ago as well, but unfortunately has just shifted his struggles elsewhere (with turnovers and other bad decisions instead of sacks).

 

[TravisMight]

I feel a little smarter having read this. Thanks for the details.

Happy to talk this through. Glad you enjoyed it!

 

[TravisMight]

Piggybacking on this, have you tried a mixed effects model to see whether you can attribute the remainder of the variane to the QB specifically?

To this question, this particular model wasn't a logistic regression build. It was MLR that was cross-checked via logarithmic to see which best fit the distribution. Tbh it's not really a finished product either. This model was a curiosity-based byproduct of something else I was building for CFB/NFL Prospect analysis. I may continue to attempt to improve it using other methods, but that day is not today.

To your previous question, if it were just testing for team sack prediction then QB teammates would all test the same, and they don't. Why they don't can be attributed to a few things, one being they often have vastly different pressure percentages and blocking success in front of them even when they're on the same team. In recent years it has been proven that QBs actually own a significant portion of their pressure and can have a significant effect on the blocking grades in front of them due to style of play. When you adjust for those differences you begin to see how well QBs both deal with pressure and even perhaps bring it more upon themselves.

So no, in this case it isn't like your shortstop example. Is it perfect, and solving for every bit of sack attribution of the individual QB? Nah. But it does at least paint a significant portion of that picture given what we know. I'm sure either I, or someone else, will improve upon this measure in the future. And I know personally that multiple NFL teams build their own kind of metrics like this to help with personnel decision-making (and they probably do it better than me).