MVS Appreciation Thread

[AlexGreen#20]

5 minutes ago, craig said:

Thanks, Alex, for your thoughtful and detailed response.  Really helpful.  Expected points net must obviously calculate in all of the stuff I was wondering about. 

I agree, they complement each other.  Adams catch percentage of 77% is pretty remarkable, to go with his other good numbers.  It's been really good, obviously.  And I think the attention that Adams commands, plus the running game and underneath stuff, helps to situate MVS with some single coverage or access deep, depending on the defensive personnel and scheme.  

I wonder if MVS  would benefit from more targets?  Often a player gets into a better groove, and sync between player and QB improves, etc..  But at the same time, often with more targets a guy gets more defensive attention, and the QB lthrowing into tighter windows?  

Alex, expected points was based on league average; Packers offense being better, their expected points must be higher.  That must reduce the expected points differential, to some degree.  

I assume the catch% of 52.4% (less on long ones) is why they don't go deep to MVS more.  If you go deep on 1st, assuming it is NOT complete, then you're looking at 2nd-and-10.  Uncomfortably high risk that will turn into a turnover via punt. 

But yeah, seems things have complemented nicely, Adams and MVS in particular.  It may be that MLF is using MVS very appropriately?  When the defense give the deep stuff, take shot?  If it doesn't, don't force it?  Matchup stuff, which is why MVS might catch a bunch one week versus a particular defense (Detroit?) then go a couple of games with no targets.  It's not that MVS is inconsistent; it's that defensive game plans and personnel are inconsistent.   

 

The counterpoint to the second to last paragraph is that the EPN difference between 2nd and 10 and 2nd and 6 coming from a run isn't that significant. 

I don't think there's an issue with MLF's use of MVS. I suspect that it's a Rodgers thing. 

[Mazrimiv]

19 hours ago, AlexGreen#20 said:

It's based on Expected Points Net. League average chance of scoring based on down and distance and field position before and after each play. My statement is based on 1st and 10 from the opposing 23. Obviously if it's 3rd and 4 from the opposing 17, Davante is the more efficient target. 

In fairness this isn't a fair comparison, throwing deep is almost always more efficient than throwing short, but it isn't as though the 1st/TD% marks are significantly higher throwing to Adams than they are to MVS. An 8.5% difference there is huge, but alternatively so is the 1.9 yards per target. 

+++

Remove running in general because it's almost always less efficient.

Then with this team you basically look at Davante vs. MVS because we don't pass to anybody but Davante. 

MVS has a catch% of 52.4%. Adams has a catch percentage of 77.2%.

MVS has a 1st/TD% of 52.4%. Adams has a 1st/TD% of 61%.

MVS has a Yards/Target of 11.1. Adams has a Yards/Target of 9.2

MVS has a Yards/Completion of 20.9. adams has a Yards/Completion of 11.9

+++

Ultimately they compliment each other. MVS' ability to be hyper efficient in limited reps without being a diva is a tremendous advantage in this offense. Davante's ability to just eat targets at an absurdly high level and still maintain efficiency is a staple of the offense. 

 

So MVS is like DeSean Jackson with bad hands, a good attitude, and the ability to block in the running game?

[Sasquatch]

32 minutes ago, Mazrimiv said:

So MVS is like DeSean Jackson with bad hands, a good attitude, and the ability to block in the running game?

Edit: good hands with a yips (brain) issue.  And he draws penalties.

[incognito_man]

1 hour ago, Sasquatch said:

Edit: good hands with a yips (brain) issue.  And he draws penalties.

and caricatures.

true story: he worked as a caricature artist at Disney Land for a summer

[craig]

2 hours ago, AlexGreen#20 said:

"I assume the catch% of 52.4% (less on long ones) is why they don't go deep to MVS more.  If you go deep on 1st, assuming it is NOT complete, then you're looking at 2nd-and-10.  Uncomfortably high risk that will turn into a turnover via punt."  (Craig) 

The counterpoint to the second to last paragraph is that the EPN difference between 2nd and 10 and 2nd and 6 coming from a run isn't that significant.

You're saying data shows no significant difference between 2nd-and-6 versus 2nd-and-10?  That seems counterintuitive, and doesn't seem consistent with observational experience. 

Puzzling.  I'm plenty open-minded towards data and statistical evidence, and I love to find new info that's better or smarter or more nuanced that simplistic assumptions that I or others have accepted previously.  But usually when there is new or unexpected data, it's a prompt to think harder, and after doing so it ends up making good sense.  

But how can it be that 2nd-and-6 versus 2nd-and-10 makes no difference?  Alex, any suggestion as to how that would be true, or why; or how to rationalize that?  because 2nd-and-6 teams are going to run and end up 3rd-and-3 and passing anyway?  Because 2nd-and-10 teams are going to pass twice, and defense is going to give excessive cushion anyway?  

I admit I'm struggling with this one....  Any help or clarification or rationalization appreciated.  

Tangent 1:  If 2nd and 6 makes no difference from 2nd and 10, then is it consistent to fault Pettine for the recurring use of a light box?  Isn't light box pretty much inviting opponents to run on 1st down and to set up 2nd-and-6?   Packers run-defense is at 4.5 ypc for the season; a couple of big gashes but median is certainly ≤4.  If 2nd-and-6 is actually no different from 2nd-and-10, isn't the defensive approach with a light box essentially inviting opponents to run into 2nd-and-6, and wasting a down for no benefit?    *IF* we actually accept that 2nd-and-6 is no better offensively than 2nd-and-10, why do we then fault a defense for being structured to invite 2nd-and-6?  

Tangent Q2:  Is EPN for a particular drive, or for a full game?  I assume it's on a per-drive basis?  Partly the reason I'm asking is because *IF* it wasn't, and was instead on a full-game basis, that might shift some considerations.    

 

[AlexGreen#20]

17 minutes ago, craig said:

You're saying data shows no significant difference between 2nd-and-6 versus 2nd-and-10?  That seems counterintuitive, and doesn't seem consistent with observational experience. 

Puzzling.  I'm plenty open-minded towards data and statistical evidence, and I love to find new info that's better or smarter or more nuanced that simplistic assumptions that I or others have accepted previously.  But usually when there is new or unexpected data, it's a prompt to think harder, and after doing so it ends up making good sense.  

But how can it be that 2nd-and-6 versus 2nd-and-10 makes no difference?  Alex, any suggestion as to how that would be true, or why; or how to rationalize that?  because 2nd-and-6 teams are going to run and end up 3rd-and-3 and passing anyway?  Because 2nd-and-10 teams are going to pass twice, and defense is going to give excessive cushion anyway?  

I admit I'm struggling with this one....  Any help or clarification or rationalization appreciated.  

Tangent 1:  If 2nd and 6 makes no difference from 2nd and 10, then is it consistent to fault Pettine for the recurring use of a light box?  Isn't light box pretty much inviting opponents to run on 1st down and to set up 2nd-and-6?   Packers run-defense is at 4.5 ypc for the season; a couple of big gashes but median is certainly ≤4.  If 2nd-and-6 is actually no different from 2nd-and-10, isn't the defensive approach with a light box essentially inviting opponents to run into 2nd-and-6, and wasting a down for no benefit?    *IF* we actually accept that 2nd-and-6 is no better offensively than 2nd-and-10, why do we then fault a defense for being structured to invite 2nd-and-6?  

Tangent Q2:  Is EPN for a particular drive, or for a full game?  I assume it's on a per-drive basis?  Partly the reason I'm asking is because *IF* it wasn't, and was instead on a full-game basis, that might shift some considerations.    

 

It's not that there's no difference, it's that the difference is smaller than you would expect and the difference is easily made up for by the potential for a big gain. 

EPN is both. It's designed to report along one drive, but it minorly accounts for the counter drive that the opposing offense is likely to muster.

[Smidgeon]

1 hour ago, craig said:

You're saying data shows no significant difference between 2nd-and-6 versus 2nd-and-10?  That seems counterintuitive, and doesn't seem consistent with observational experience. 

Puzzling.  I'm plenty open-minded towards data and statistical evidence, and I love to find new info that's better or smarter or more nuanced that simplistic assumptions that I or others have accepted previously.  But usually when there is new or unexpected data, it's a prompt to think harder, and after doing so it ends up making good sense.  

But how can it be that 2nd-and-6 versus 2nd-and-10 makes no difference?  Alex, any suggestion as to how that would be true, or why; or how to rationalize that?  because 2nd-and-6 teams are going to run and end up 3rd-and-3 and passing anyway?  Because 2nd-and-10 teams are going to pass twice, and defense is going to give excessive cushion anyway?  

I admit I'm struggling with this one....  Any help or clarification or rationalization appreciated.  

Tangent 1:  If 2nd and 6 makes no difference from 2nd and 10, then is it consistent to fault Pettine for the recurring use of a light box?  Isn't light box pretty much inviting opponents to run on 1st down and to set up 2nd-and-6?   Packers run-defense is at 4.5 ypc for the season; a couple of big gashes but median is certainly ≤4.  If 2nd-and-6 is actually no different from 2nd-and-10, isn't the defensive approach with a light box essentially inviting opponents to run into 2nd-and-6, and wasting a down for no benefit?    *IF* we actually accept that 2nd-and-6 is no better offensively than 2nd-and-10, why do we then fault a defense for being structured to invite 2nd-and-6?  

Tangent Q2:  Is EPN for a particular drive, or for a full game?  I assume it's on a per-drive basis?  Partly the reason I'm asking is because *IF* it wasn't, and was instead on a full-game basis, that might shift some considerations.    

 

I like the questions you're asking.  Better formed than mine, but on the same topic.

[craig]

1 hour ago, AlexGreen#20 said:

It's not that there's no difference, it's that the difference is smaller than you would expect and the difference is easily made up for by the potential for a big gain. 

EPN is both. It's designed to report along one drive, but it minorly accounts for the counter drive that the opposing offense is likely to muster.

Thanks, Alex.  The latter is key inclusion, so I'm glad that's not overlooked.  I'd been thinking that if you throw an incomplete bomb from the 23, then 2-3 punt, they'll get the ball at the 35 and good field position.  If we try step-drive, even if we do punt and get no points, we might get some 1st downs first and punt them to the 12 or whatever.  So including that field-position factor for the counter-drive is helpful and appropriate, even if modest.  Thanks.  Keeps an honest analyst like myself from getting distracted.

I admit I'm still struggling to wrap my head around the 2nd-and-10 factor, though.  Especially for the Packers this season, where changes of possession (other than following a score) have been few. 

 I view punt as a turnover; a turnover with better field position, but a turnover nonetheless. 

More than half of our games Scott has punted ≤3 times, 7 games with ≤2 punts.  (And some of his punts have been in conservative clock-kill time, when we weren't really even prioritizing scoring.)  Turning the ball over without scoring is not common, and you can't afford to do it very much.  

So I'm struggling to understand how 2nd-and-10 versus 2nd-and-6 doesn't elevate the risk of turning it over via punt, and how that isn't a non-trivial risk?  

Edited January 5, 2021 by craig

[AlexGreen#20]

1 minute ago, craig said:

Thanks, Alex.  The latter is key inclusion, so I'm glad that's not overlooked.  I'd been thinking that if you throw an incomplete bomb from the 23, then 2-3 punt, they'll get the ball at the 35 and good field position.  If we try step-drive, even if we do punt and get more points, we might get some 1st downs first and punt them to the 12 or whatever.  So including that field-position factor for the counter-drive is helpful and appropriate, even if modest.  Thanks.  Keeps an honest analyst like myself from getting distracted.

I admit I'm still struggling to wrap my head around the 2nd-and-10 factor, though.  Especially for the Packers this season, where changes of possession (other than following a score) have been few. 

 I view punt as a turnover; a turnover with better field position, but a turnover nonetheless. 

More than half of our games Scott has punted ≤3 times, 7 games with ≤2 punts.  (And some of his punts have been in conservative clock-kill time, when we weren't really even prioritizing scoring.)  Turning the ball over without scoring is not common, and you can't afford to do it very much.  

So I'm struggling to understand how 2nd-and-10 versus 2nd-and-6 doesn't elevate the risk of turning it over via punt, and how that isn't a non-trivial risk?  

The difference in EPN between a turnover and a punt is massive. IIRC a turnover generally occurs on average at basically the LOS, +/- 5 yards. A punt generally nets you something like +45 yards.

Just going off of the Titans game (Bears numbers aren't up yet)

1st and 10 from the GB20 had an EPB of 0.28

1st and 10 from the TEN30 had an EPB of 3.58

+++

In the 3rd quarter at 10:22:

It was 1st and 10 from the Titans 30. At that point, the Titans had an Expected Points of 0.94 points. 

Tannehill threw an incompletion, so it shifted to 2nd and 10. 2nd and 10 from the Titans 30 has an Expected Points of 0.39 points

So that incompletion was worth -0.55 points for the Titans.

 

In the 3rd quarter at 3:31:

It was 1st and 10 from the Titans 24. At that point, the Titans had an Expected Points of 0.54 points. 

Henry had a 4 yard run, so it shifted to 2nd and 6. 2nd and 6 from the Titans 28 has an Expected Points of 0.54 points

So that 4 yard run on 1st down was worth 0.00 points. 

 

The next play,

Henry had a 3 yard run to make it 3rd and 3.

That 3 yard run was worth -0.31 points. 

+++

Drive success rates generally speaking aren't swung by short plays. You basically have to get 5 yards for a play to break even. Pettine's entire defensive philosophy revolves around this principle. Don't give up anything long, force them to go 15 play 75 yard drives. Eventually you'll make a play and put them in a spot where those 5 yard gains aren't going to get the 1st down marker and you'll get off the field. 

The downside of that philosophy is that occasionally you will run into a team that can really punish your risk averse scheme for long drives of long runs (49ers)

Or it will run into a team like the Bears who were very comfortable in their sequence going 8 yards in 3 plays and then going for it on 4th down consistently. In the case of the Bears it worked, only 16 given up. Got a 4th down spot. But god damn it was a miserable game to watch. 

[MrBobGray]

Good lord this thread has really turned around.  Actually too much good stuff to get all of my thoughts on it out now, so I'll have to come back to this later.

The one thing I'll get to here is I think there's an easy way to explain the value of a long play in general, but especially long TDs.  Chicago had five drives of 10+ plays, and they got points out of 3 of them.  Not even TDs, just points at all.  For the game, they went 4 out of 9 for scoring drives.  The 72 yard TD to MVS, as a play, added as much value to the Green Bay Packers offensive output as fully 14 plays did for Chicago.  Even further, although the pass itself was on 3rd down the nature of a big play TD like that is it would score on any down from any distance.  It just takes one, and as Chicago proved the rest of the game, even 14 isn't always enough.  It's so hard to be sure of scoring in the NFL, even for high scoring teams, that you really can't over value the ability to just guarantee a TD(in the sense that if the play works it's guaranteed, as opposed to a drive's worth where every play needs to succeed to a degree).

[craig]

15 hours ago, AlexGreen#20 said:

...+++It was 1st and 10 from the Titans 30. At that point, the Titans had an Expected Points of 0.94 points. 

Tannehill threw an incompletion, so it shifted to 2nd and 10. 2nd and 10 from the Titans 30 has an Expected Points of 0.39 points

So that incompletion was worth -0.55 points for the Titans.

In the 3rd quarter at 3:31:

It was 1st and 10 from the Titans 24. At that point, the Titans had an Expected Points of 0.54 points. 

Henry had a 4 yard run, so it shifted to 2nd and 6. 2nd and 6 from the Titans 28 has an Expected Points of 0.54 points

So that 4 yard run on 1st down was worth 0.00 points. ...

Thanks, Alex, this helps, and puts numbers to my argument.  

-0.55 (2nd-and-10), versus 0.00 (2nd-and-6).  That's different, and it's not insignificant.  Teams don't throw deep on 1st downs more often because despite the reward if completed, they understand the risk of 2nd-and-10 is more probable.  

A 4-yard gain isn't a good play, but it doesn't hurt you.  A 0-yard play does hurt.

Tangent notes:

1.  Obviously 45 yards of field position is a big deal, so the distinction between and punt and a true turnover is serious. Still, any change of possession is a big deal.  We only get so many possessions per game, and we score on a lot of them; to punt it away is a non-trivial bad thing.  

2.  Obviously **successful** long passes are invaluable, and chunk plays are huge.  (We get lots of them).  

3.  risk-reward is always at play.  Smart NFL team analytics departments would possibly have their teams throwing deep much, much more frequently if risk-reward was hugely advantageous to do so, and if doing so gave such teams a significant competitive advantage.  (Weak argument, I know....)

4.  Existing data for bomb-completion is based on existing usage.  I suspect teams do it situationally (MVS singled with ILB).  If teams attempted lots more 1st-down bombs and did so less situationally,  my guess is completion rates would diminish to some extent.  

 

Edited January 6, 2021 by craig

[Mr Bad Example]

21 hours ago, Packer_ESP said:

I can never make my mind up about it... I see the merits of what you're saying but capping it also makes it so CBs would rather blatantly interfere than give up a long catch. Both options have pros and cons and I can't think of a way of making it better without introducing subjective referee appreciations regarding how blatant it is.

Half the distance to the spot of the foul? That way instead of a 40-50 yd penalty, it's 20-25 yds which is still massive but not necessarily game-changing. 

[craig]

Sorry for extending this discussion about the risk of 2nd-and-10.  But I thought a little more about it.  Some stats:  https://www.pro-football-reference.com/teams/gnb/2020.htm

Data suggests that with three plays, the Packers had  ~77% chance of converting (1st down or a score).  With 2 plays, 50%. 

Basically ~77% of their first downs result in either a TD or a first down within the first three downs.  Only 50% score or convert within only 2 downs.  

My thinking here is that *if* you throw long on first down and *don't* complete it, there's ~50% chance you'll subsequently end up facing 4th down.  If you don't go deep, and do what the Packers normally do, it's ~77%.   So that's just part of the risk-reward assessment.  There are negative consequences to unsuccessfully taking the shot.  

The other view, MVS has a 52% catch rate overall.  Obviously it's somewhat lower on the long stuff.  But one could argue that it's almost 50/50 that you might convert *if* you throw to MVS on first; and that even if you don't complete it, you've still got 50% shot to convert thereafter.  So if the Packers do what they normally do, it's ~77%.  But if you've got an ~50-50 shot throwing to MVS on first, and then a 50-50 shot to convert thereafter even if you don't complete the MVS throw, it would look like you'd still have basically a 3/4 chance of converting before 4th down.  ~3/4 shot of converting with an MVS attempt versus ~77% doing what the Packers normally do, it's kinda easy to see why taking MVS shots seems to make analytics sense.

The data:   Packers had 358 1st downs; 180 3rd down attempts; 46 punts; 16 FG; 21 4th down attempts.    .  

• 180 3rd down attempts per 358 1st downs suggests that chance of converting a first down within the first 2 plays was ~50%.
• 46 punts + 16 FG + 21 4th down attempts, relative to 358 1st downs, is 23%.  That suggests that when they had a 1st down, there was a 23% chance that after downs 1-2-3 they hadn't scored or converted a first down yet, and a 77% chance that they had. 
• Note:  Those numbers show 50% conversion within the first two downs.  It's possible that conversion using downs 2+3 following 2nd-and-10 might be variably different.  QB and play-calling know exactly where the target line is, so maybe they'd convert a little better.  Or perhaps the defense has a clearer view of what they need to stop, and whether you'll run or pass, so maybe they actually can defend it a little better?  Beats me.  I'm just thinking that the conversion rate using downs 2+3 might not be exactly the same as the conversion rate using downs 1+2.  

 

 

[Arthur Penske]

Considering the stakes, could be MVS’s best game as a Packer. Hopefully he continues to develop into next year and get even better.

 

 

 

[vegas492]

I thought he played a really good game yesterday.  Made a great catch and run with those strong hands.

I've been down on MVS a lot.  But it does look like he's taking the next step.  Great game yesterday.  Looking forward to many more to come.