Dolphins’ QB Josh Rosen solid in first start with Miami
148 replies
"KeyFin wrote:
Again though, you're talking about margins across every team- others are talking about specific teams in specific years that ran the ball effectively. Philly was a recent example with their high-scoring offense and it earned them a Super Bowl.
What I'm saying is, you can't solve a specific problem with generalized data. We're not trying to measure the league-wide effectiveness of running the ball- we're trying to measure how good running teams fare when they choose to pass. Do their QB's face the same levels of pressure? Do their completion percentages or average pass distance vary from league averages? Do they tend to score more or less points than the average team? That's where the actual data is that would answer this question.
For instance, last season the Ravens rushed for 205.8 yards per game to lead the league. The Ravens also tied with he Chiefs for most points per game even though they were 7th overall in passing (Chiefs were 1st overall passing, 20th rushing). That's two completely different offenses with the same end result, so I'm not sure how you can twist this to say running isn't effective and/or doesn't benefit passing. Mahommes was elite last season and the only reason they hung with the Ravens to begin with.
Who was the worst rushing last season? The Bengals, Skins and Dolphins- three lousy teams. Who was the best? The Ravens, 49ers, Vikings, Bills and Cowboys....three of five won their division.
Umm.. first of all everyone up till now was talking about a general effect, not an effect that only applies to a specific team. So it's fine if you want to change the question but let's not act like I wasn't answering the question asked.
Secondly, you CANNOT solve a specific problem without looking at generalized data because of confounds. There are too many interaction effects you can't tease apart if you JUST look at one team. The only way to tease apart the effect of X on Y when X and Y interact but aren't deterministically related is to look at what happens across multiple teams.
Third, for no team should you use yards per game as a measure of rushing effectiveness. That should be clear by now since yards is confounded with rushing attempts, and the primary difference in rushing attempts between winning and losing teams is in the 4th quarter when the winning team kills the clock.
Fourth, where are you getting the stat that the Ravens, 49ers, Vikings, Bills and Cowboys were the "best rushing teams" in 2018?
https://www.pro-football-reference.com/years/2018/index.htm
Like.. what stat are you using? What you said is not true based on ANY rushing stat as you can see.
Fifth, I never said rushing isn't effective. I said that passing efficiency is about 5.5 times more effective than rushing efficiency w.r.t. winning, and that otherwise rushing attempts are not improving unpredictability in play calling beyond a minimum level that it looks like no team is consistently below. So the "twisting" isn't being done by me here.
Finally, there is literally no stat in the NFL that suggests that a particular effect (e.g., rushing attempts on passing Y/A) suddenly appears for ONLY the top 3-5 teams but does not occur for any of the others. It's actually amazing how many NFL stats show linear relationships. So the CORRECT way of analyzing this is to NOT say "what happens only for the very best teams". The effect of almost everything you analyze is seen across the entire spectrum. It's like saying "let's look at only the top 3-5 teams in TD% and assume the effect of TD% only applies to those teams". You can look at stat after stat and show the relationship holds across either all or most of the entire range.
So the way I analyzed it is the way you want to approach this, not only because you can't tease apart relationships among stats as easily when you isolate individual teams but also because you can show the relationship exists across the entire range.
resnorForum Veteran"cbrad wrote:
It was as "ideal" as you needed it to be in the first half. OL was fantastic. One key WR drop occurred (that might have turned into a TD), but otherwise an overthrown ball was caught by another receiver bailing Rosen out. Stuff like this happens no matter what team you're on. Yeah, that 1st half was good enough to call "ideal" for a QB. Even the defense kept the score low enough to win.
Maybe that's the only time Rosen has played where you could say that, but I think the description is accurate enough.
There nothing "ideal" about playing in Miami right now. It's crazy to even suggest that. Plus, you're ignoring how a person changes playstyle when they're constantly getting let down by their teammates.
But if you're going to refer to the first half as "ideal", then we just need to agree to disagree.
"resnor wrote:
There nothing "ideal" about playing in Miami right now. It's crazy to even suggest that. Plus, you're ignoring how a person changes playstyle when they're constantly getting let down by their teammates.
But if you're going to refer to the first half as "ideal", then we just need to agree to disagree.
The specific word used is for me irrelevant. Point is, Rosen wasn't facing any pressure worth mentioning due to the OL and there was only one WR drop worth mentioning. He played well in that situation as most QB's would. What he hasn't demonstrated, that any franchise QB should demonstrate, is that he can play well when there's adversity or when there's more pressure on the QB to lead the team to a win.
That's all that matters here. The specific word used to describe the conditions in the 1st half is totally irrelevant, at least to me.
KeyFinForum Veteran"cbrad wrote:
Umm.. first of all everyone up till now was talking about a general effect, not an effect that only applies to a specific team. So it's fine if you want to change the question but let's not act like I wasn't answering the question asked.
Secondly, you CANNOT solve a specific problem without looking at generalized data because of confounds. There are too many interaction effects you can't tease apart if you JUST look at one team. The only way to tease apart the effect of X on Y when X and Y interact but aren't deterministically related is to look at what happens across multiple teams.
Third, for no team should you use yards per game as a measure of rushing effectiveness. That should be clear by now since yards is confounded with rushing attempts, and the primary difference in rushing attempts between winning and losing teams is in the 4th quarter when the winning team kills the clock.
Fourth, where are you getting the stat that the Ravens, 49ers, Vikings, Bills and Cowboys were the "best rushing teams" in 2018?
https://www.pro-football-reference.com/years/2018/index.htmLike.. what stat are you using? What you said is not true based on ANY rushing stat as you can see.
Fifth, I never said rushing isn't effective. I said that passing efficiency is about 5.5 times more effective than rushing efficiency w.r.t. winning, and that otherwise rushing attempts are not improving unpredictability in play calling beyond a minimum level that it looks like no team is consistently below. So the "twisting" isn't being done by me here.
Finally, there is literally no stat in the NFL that suggests that a particular effect (e.g., rushing attempts on passing Y/A) suddenly appears for ONLY the top 3-5 teams but does not occur for any of the others. It's actually amazing how many NFL stats show linear relationships. So the CORRECT way of analyzing this is to NOT say "what happens only for the very best teams". The effect of almost everything you analyze is seen across the entire spectrum. It's like saying "let's look at only the top 3-5 teams in TD% and assume the effect of TD% only applies to those teams". You can look at stat after stat and show the relationship holds across either all or most of the entire range.
So the way I analyzed it is the way you want to approach this, not only because you can't tease apart relationships among stats as easily when you isolate individual teams but also because you can show the relationship exists across the entire range.
I goofed that up- I quoted 2019 rushing stats so far from ESPN. The tab must have flopped on me. I'll update that in a minute...busy for the moment at work.
Winning football games at any level comes down to winning in the trenches.
"cbrad wrote:
That's correct, though for future reference I think it's important to separate two periods: pre-1978 and post-1978. The rule changes in 1978 dramatically changed the nature of the game, and what you said applies much more to post-1978 than to pre-1978.]
Thanks, the cbrad statistics primer has been a nice upside to this horrible season. :-)
Re the thread discussion about evaluating Rosen, certainly posters mentioning the OL and the drops have a point, but the fact is he's got to be evaluated this year, because the team has to decide who to draft in the off-season. And obviously they're not just going to crunch numbers; they're going to look at film and evaluate him qualitatively. Unless he's just terrific, they're going to pick a QB first. I'm sure Rosen understands that better than anyone.
Here is a link to a l-o-o-o-n-g article by Bill Barnwell from April 2018 about how bad the NFL is at evaluating quarterbacks; there's also a link to a Part I that compares GM evaluations with how prospects actually turned out in the NFL, going back to 2005. Anyway, the article linked below also includes a discussion of 2018's prospects, the way they looked before they were drafted -- Josh Allen, Sam Darnold, Lamar Jackson, Baker Mayfield, and Rosen.
https://www.espn.com/nfl/story/_/id/23039883/history-tells-us-nfl-terrible-evaluating-quarterbacks-means-2018-draft-prospects
Quick takeaways: Mayfield looked like by far the best passer; Jackson was on a different planet from the rest as a runner; and Rosen was last in QBR. And no, I'm not going to defend QBR.
"keypusher wrote:
Thanks, the cbrad statistics primer has been a nice upside to this horrible season. :-)
Re the thread discussion about evaluating Rosen, certainly posters mentioning the OL and the drops have a point, but the fact is he's got to be evaluated this year, because the team has to decide who to draft in the off-season. And obviously they're not just going to crunch numbers; they're going to look at film and evaluate him qualitatively. Unless he's just terrific, they're going to pick a QB first. I'm sure Rosen understands that better than anyone.
Here is a link to a l-o-o-o-n-g article by Bill Barnwell from April 2018 about how bad the NFL is at evaluating quarterbacks; there's also a link to a Part I that compares GM evaluations with how prospects actually turned out in the NFL, going back to 2005. Anyway, the article linked below also includes a discussion of 2018's prospects, the way they looked before they were drafted -- Josh Allen, Sam Darnold, Lamar Jackson, Baker Mayfield, and Rosen.
Quick takeaways: Mayfield looked like by far the best passer; Jackson was on a different planet from the rest as a runner; and Rosen was last in QBR. And no, I'm not going to defend QBR.
Nice article. I also found the article they linked to that showed no team is consistently good at drafting to be really interesting:
https://fivethirtyeight.com/features/no-team-can-beat-the-draft/
These articles are really damning for the GM position lol.
"cbrad wrote:
Nice article. I also found the article they linked to that showed no team is consistently good at drafting to be really interesting:
https://fivethirtyeight.com/features/no-team-can-beat-the-draft/These articles are really damning for the GM position lol.
So football GMs can join stock pickers and political pundits as people who do no better than monkeys throwing darts at a board...of course that's not really fair. The draft works well overall in sorting talent, even though no single GM does consistently better than average. I assume that's also true for stock pickers -- probably not for pundits.
In Thinking, Fast and Slow, Daniel Kahneman contrasts groups that make largely invalid predictions (stock pickers, pundits) and valid ones (chess masters, firefighters, physicians) -- the difference is that members of the latter groups (i) operate in an environment that is sufficiently regular to be predictable (ii) have the opportunity to learn these regularities through prolonged practice. A GM's environment is more regular than a political pundit's, but not regular enough, it seems.
"keypusher wrote:
So football GMs can join stock pickers and political pundits as people who do no better than monkeys throwing darts at a board...of course that's not really fair. The draft works well overall in sorting talent, even though no single GM does consistently better than average. I assume that's also true for stock pickers -- probably not for pundits.
In Thinking, Fast and Slow, Daniel Kahneman contrasts groups that make largely invalid predictions (stock pickers, pundits) and valid ones (chess masters, firefighters, physicians) -- the difference is that members of the latter groups (i) operate in an environment that is sufficiently regular to be predictable (ii) have the opportunity to learn these regularities through prolonged practice. A GM's environment is more regular than a political pundit's, but not regular enough, it seems.
Yeah.. the cognitive and behavioral biases Kahneman and Tversky discovered are part of the experimental foundation of modern psychology, but I wouldn't give much credence to their theories. No one really knows why humans have the types of biases they and others have discovered. The number of different explanations offered is as great (or greater) than the number of different biases!
For example, regarding the theory about "predictability", note how predictable 2-point vs. 3-point shooting percentages in basketball are. Yet coaches never seemed to learn that it was more efficient to shoot 3 pointers than 2 pointers outside the restricted area. Of course today there's an analytics revolution in the NBA that rivals MLB and the average number of 3 pointers taken has steadily increased over the last decade. But the point is that once you define "predictability" you'll see tons of cases where people show mathematically suboptimal behavior even in highly predictable situations.
btw.. I actually have some indirect connection to Amos Tversky who died before he could receive the Nobel Prize (only Kahneman received it): his son Tal Tversky was my office mate while I was in graduate school lol. Guy now works at Apple.
"cbrad wrote:
Yeah.. the cognitive and behavioral biases Kahneman and Tversky discovered are part of the experimental foundation of modern psychology, but I wouldn't give much credence to their theories. No one really knows why humans have the types of biases they and others have discovered. The number of different explanations offered is as great (or greater) than the number of different biases!
For example, regarding the theory about "predictability", note how predictable 2-point vs. 3-point shooting percentages in basketball are. Yet coaches never seemed to learn that it was more efficient to shoot 3 pointers than 2 pointers outside the restricted area. Of course today there's an analytics revolution in the NBA that rivals MLB and the average number of 3 pointers taken has steadily increased over the last decade. But the point is that once you define "predictability" you'll see tons of cases where people show mathematically suboptimal behavior even in highly predictable situations.
btw.. I actually have some indirect connection to Amos Tversky who died before he could receive the Nobel Prize (only Kahneman received it): his son Tal Tversky was my office mate while I was in graduate school lol. Guy now works at Apple.
OK, hopefully this will be my last post on Daniel Kahneman! Thinking, Fast and Slow is a pop-science book with next to no math, so I don't know how rigorously he defines predictable situations -- probably not very. Similar to your observation about 2 and 3 point shooting, it took a while for NFL coaches to fully exploit the 1978 rules changes, and if the sabermetricians are right there has been a lot of suboptimal strategy in baseball over the years -- teams using up precious outs for paltry gains like "moving the runner over", using low-average speedsters to lead off, not using closers when they are most valuable, etc. Not sure how he would class all those. Anyway, Kahneman is very pleasant company, and he had a wonderful collaboration over decades with Amos Tversky -- he makes it very clear in the book that his Nobel would have been shared with Tversky, if only Tversky had lived long enough.
IrishmanForum Veteran"The Guy wrote:
Right, but the company he's in there can be informative. Again you have a virtual who's who of young QBs in the league in the 2018 list, Roethlisberger notwithstanding.
When youth and inexperience is such a strong common thread among those QBs, it suggests that youth and inexperience, and not the functioning of their offensive lines, is responsible for their performance in that regard. What are the odds that that list of nine QBs in 2018 is also a list of nine of the worst offensive lines in the league? Slim.
So, it behooves you to look further into the list than just Rosen. Doing so may be informative and may yield a different conclusion. It's entirely possible that Rosen, and not his offensive lines, is the active ingredient in his performance in that regard. His offensive lines could be among the worst in the league, and he could be entirely responsible for his performance in that regard nonetheless.
Just because a QB has a relatively poor offensive line doesn't mean it explains every negative facet of his performance. Some of that could be on him exclusively, and replacing his line with the best one in the league could result in nothing significantly different in that specific regard.
Again, an extreme analogy just to illustrate the point: throw a middle school QB into the NFL with the worst offensive line in the league, and then replace his offensive line with the best one in the league. Will that alone make him play at an NFL level, or will he continue to display inadequacies that have nothing to do with his offensive line?
"Right, but the company he's in there can be informative."
Your basic assumption about "....the company he's in..." being informative is wrong; it is subjective, not informative! You are not even clear as to which company your referring to. Is it the rest of his offense, which is literally "....the company he's in..." when he is playing or is it to a group QB's, which is an assumption If you make the assumption and say its a group of QB's, which group of QB's would that be? A virtual, but unspecified group?
That would be an assumption based on an assumption. Even if you make both assumptions, you would still need to know how what kind of effect those QB's offenses have on the QB's stats. I am of course assuming you are going to base your observations based on some kind of stats. I guess this makes any potential analysis and assumption based on an assumption based on an assumption.
Give me a break!
You can be a bit more specific, even when you try to show some examples of specific cases, it turns out they are based on incomplete assumptions. An example is:
" Again you have a virtual who's who of young QBs in the league in the 2018 list, Roethlisberger notwithstanding.
When youth and inexperience is such a strong common thread among those QBs, it suggests that youth and inexperience, and not the functioning of their offensive lines, is responsible for their performance in that regard. What are the odds that that list of nine QBs in 2018 is also a list of nine of the worst offensive lines in the league?"
Where is the this virtual list and exactly how does it address the situation you are trying to explain. Is it based on age, experience, both or something else you have yet to dredge up? How young is youth and much experience is experience? Was it good experience or bad experience, and how do you account for different variations of experience such as coaching, teammates, opposition?
The you go on with comparing this list to the worst offensive lines without stating who they are and how they were selected.
The thought process in your position is at times subjective, presumptive and almost always undefined, just generally suggested.
"cbrad wrote:
Yeah.. the cognitive and behavioral biases Kahneman and Tversky discovered are part of the experimental foundation of modern psychology, but I wouldn't give much credence to their theories. No one really knows why humans have the types of biases they and others have discovered. The number of different explanations offered is as great (or greater) than the number of different biases!
For example, regarding the theory about "predictability", note how predictable 2-point vs. 3-point shooting percentages in basketball are. Yet coaches never seemed to learn that it was more efficient to shoot 3 pointers than 2 pointers outside the restricted area. Of course today there's an analytics revolution in the NBA that rivals MLB and the average number of 3 pointers taken has steadily increased over the last decade. But the point is that once you define "predictability" you'll see tons of cases where people show mathematically suboptimal behavior even in highly predictable situations.
btw.. I actually have some indirect connection to Amos Tversky who died before he could receive the Nobel Prize (only Kahneman received it): his son Tal Tversky was my office mate while I was in graduate school lol. Guy now works at Apple.
You said it best when you say Rosen MAY be the reason for some of his struggles/performance.
That is why I keep advocating the use of advanced metrics as well as confirming them via tape, because it may or may not be him.
When I say tape I dont mean just watching the game, but true analysis to see where, when, what and indeed if Rosen can do things better.
The GuyForum Veteran"Irishman wrote:
"Right, but the company he's in there can be informative."
Your basic assumption about "....the company he's in..." being informative is wrong; it is subjective, not informative! You are not even clear as to which company your referring to. Is it the rest of his offense, which is literally "....the company he's in..." when he is playing or is it to a group QB's, which is an assumption If you make the assumption and say its a group of QB's, which group of QB's would that be? A virtual, but unspecified group?
That would be an assumption based on an assumption. Even if you make both assumptions, you would still need to know how what kind of effect those QB's offenses have on the QB's stats. I am of course assuming you are going to base your observations based on some kind of stats. I guess this makes any potential analysis and assumption based on an assumption based on an assumption.
Give me a break!
You can be a bit more specific, even when you try to show some examples of specific cases, it turns out they are based on incomplete assumptions. An example is:
" Again you have a virtual who's who of young QBs in the league in the 2018 list, Roethlisberger notwithstanding.
When youth and inexperience is such a strong common thread among those QBs, it suggests that youth and inexperience, and not the functioning of their offensive lines, is responsible for their performance in that regard. What are the odds that that list of nine QBs in 2018 is also a list of nine of the worst offensive lines in the league?"Where is the this virtual list and exactly how does it address the situation you are trying to explain. Is it based on age, experience, both or something else you have yet to dredge up? How young is youth and much experience is experience? Was it good experience or bad experience, and how do you account for different variations of experience such as coaching, teammates, opposition?
The you go on with comparing this list to the worst offensive lines without stating who they are and how they were selected.
The thought process in your position is at times subjective, presumptive and almost always undefined, just generally suggested.
The discussion on the message board as a whole is largely speculative in nature. There’s very little about the team we can know definitively here.
The GuyForum Veteran"The_Dark_Knight wrote:
Then go be a Seattle Seahawks fan. You always have a “yea, but” reply to everything instead of discussing the merits of a point someone makes.
It’s obnoxious and getting flat out annoying
If there is something about my behavior you or anyone else would like to discuss, please address it to me backchannel.
We are here to discuss the Miami Dolphins and pro football in general, not the behavior of individual posters. Thanks.
IrishmanForum Veteran"The Guy wrote:
The discussion on the message board as a whole is largely speculative in nature. There’s very little about the team we can know definitively here.
That is true, and because of that I get a little sensitive about being as particular as possible to offset the speculative part of many posts. At the very least, we should be able to qualify the limits of our speculations.
moosegutsForum Veteran"mlb1399 wrote:
Rosen is certainly not good but how many td drops, drops in general plus bad OL play has impacted that. I’m surprised more people aren’t understanding of that. On top of that, we all knew he was a project that needed developing. But he’s got no help around him.
According to http://hosted.stats.com/fb/tmleaders.asp?type=Receiving&range=NFL&rank=232 the Jaguars are 8th in the NFL in dropped passes with 5 while the Dolphins are 2nd with 8.
Also according to PFF the Jaguars have the 29th worst OL while the Dolphins have the 31st worst OL https://www.turfshowtimes.com/2019/10/1/20893932/pff-nfl-offensive-line-rankings-la-rams-dead-last-struggles-andrew-whitworth-rob-havenstein
Despite this 6th round rookie QB Gardner Minshew is completing 69.1% of his passes has 7 TD's and 1 int and a 106.9 QB rating. That's with Fournett only having 1 great game (the Jags were 23rd in rushing yards the 1st 3 games) and WR corps that's not exactly a who's who of NFL stardom. Now maybe Minshew is a fluke and will come back down to earth but he was able to play well despite similar circumstances to Rosen.
mooseguts out
The GuyForum Veteran"mooseguts wrote:
According to http://hosted.stats.com/fb/tmleaders.asp?type=Receiving&range=NFL&rank=232 the Jaguars are 8th in the NFL in dropped passes with 5 while the Dolphins are 2nd with 8.
Also according to PFF the Jaguars have the 29th worst OL while the Dolphins have the 31st worst OL https://www.turfshowtimes.com/2019/10/1/20893932/pff-nfl-offensive-line-rankings-la-rams-dead-last-struggles-andrew-whitworth-rob-havenstein
Despite this 6th round rookie QB Gardner Minshew is completing 69.1% of his passes has 7 TD's and 1 int and a 106.9 QB rating. That's with Fournett only having 1 great game (the Jags were 23rd in rushing yards the 1st 3 games) and WR corps that's not exactly a who's who of NFL stardom. Now maybe Minshew is a fluke and will come back down to earth but he was able to play well despite similar circumstances to Rosen.
None of those PFF offensive line stats has ever correlated strongly with passer rating or other stats commonly attributed to QBs.
In other words, the relationship between offensive line play and quarterback play that's thought by many to exist has never been supported by PFF's offensive line statistics.
Now, that could mean one or both of two things: 1) the relationship between offensive line play and quarterback play is far weaker than many people believe, and/or 2) PFF's offensive line statistics are invalid. There is certainly no gold standard measure of offensive line play against which to measure the validity of PFF's offensive line statistics.
KeyFinForum Veteran"mooseguts wrote:
According to http://hosted.stats.com/fb/tmleaders.asp?type=Receiving&range=NFL&rank=232 the Jaguars are 8th in the NFL in dropped passes with 5 while the Dolphins are 2nd with 8.
Also according to PFF the Jaguars have the 29th worst OL while the Dolphins have the 31st worst OL https://www.turfshowtimes.com/2019/10/1/20893932/pff-nfl-offensive-line-rankings-la-rams-dead-last-struggles-andrew-whitworth-rob-havenstein
Despite this 6th round rookie QB Gardner Minshew is completing 69.1% of his passes has 7 TD's and 1 int and a 106.9 QB rating. That's with Fournett only having 1 great game (the Jags were 23rd in rushing yards the 1st 3 games) and WR corps that's not exactly a who's who of NFL stardom. Now maybe Minshew is a fluke and will come back down to earth but he was able to play well despite similar circumstances to Rosen.
It's tough to use statistics in an argument four games in- especially since we've played three playoff teams and Jax has played three duds. Add to that our offensive line is changing weekly and Rosen's started only two games...and three of those 8 drops you're referencing were perfect TD passes. I don't think that's enough to prove anything.
Winning football games at any level comes down to winning in the trenches.
The GuyForum Veteran"KeyFin wrote:
It's tough to use statistics in an argument four games in- especially since we've played three playoff teams and Jax has played three duds. Add to that our offensive line is changing weekly and Rosen's started only two games...and three of those 8 drops you're referencing were perfect TD passes. I don't think that's enough to prove anything.
And that's true, but again, none of PFF's offensive line statistics has ever correlated strongly with statistics commonly attributed to quarterbacks, even over several seasons.
KeyFinForum Veteran"The Guy wrote:
And that's true, but again, none of PFF's offensive line statistics has ever correlated strongly with statistics commonly attributed to quarterbacks, even over several seasons.
I like the idea of PFF and it's great for general reference, but I have almost no faith in any statistic that doesn't show the validity of their formulas.
Winning football games at any level comes down to winning in the trenches.
The GuyForum Veteran"KeyFin wrote:
I like the idea of PFF and it's great for general reference, but I have almost no faith in any statistic that doesn't show the validity of their formulas.
The best of their offensive line statistics in my opinion is percentage of pass dropbacks in which there was pressure on the QB. Obviously there are confounds in that, and there is also no determination of whether there is agreement among multiple of PFF's observers about how pressure is defined or whether there was pressure.
However, compare that to the typical vantage point of someone here who posits that the offensive line is at fault for the Dolphins' quarterback's performance. Here, there is no systematic comparison between the Dolphins' offensive line and the other offensive lines in the league, and there remains the same possibility for bias or error in the observations of the Dolphins' offensive line's performance, good or bad.
So at least in the case of PFF, you have observers who are looking at every NFL team's offensive line in a systematic manner, and hopefully agreeing about how pressure is defined and how often it occurs. That at least would provide somewhat of a foundation for determining the relationship between offensive line and QB play throughout the league, whereas in the case of the individual observer of only the Dolphins' offensive line, that foundation can't possibly exist.
"The Guy wrote:
None of those PFF offensive line stats has ever correlated strongly with passer rating or other stats commonly attributed to QBs.
In other words, the relationship between offensive line play and quarterback play that's thought by many to exist has never been supported by PFF's offensive line statistics.
Now, that could mean one or both of two things: 1) the relationship between offensive line play and quarterback play is far weaker than many people believe, and/or 2) PFF's offensive line statistics are invalid. There is certainly no gold standard measure of offensive line play against which to measure the validity of PFF's offensive line statistics.
Depends on what your expectations are. The correlation between points scored and PFF's OL rankings in 2018 is 0.4234 which is not strong but isn't weak either, while the correlation between passer rating and PFF's OL rankings in 2018 is 0.3754.
https://www.footballoutsiders.com/stats/ol/2018
So 17.9% of variation in points scored can be explained by PFF's OL rankings in 2018 and 14.1% of variance in passer rating can be explained by PFF's OL rankings. I thought both would be around 20% so this isn't surprising me much. But for people who were expecting OL to be far more influential these numbers might be lower than expected.
Your last paragraph is spot on though. We don't have a good OL stat.
The GuyForum Veteran"cbrad wrote:
Depends on what your expectations are. The correlation between points scored and PFF's OL rankings in 2018 is 0.4234 which is not strong but isn't weak either, while the correlation between passer rating and PFF's OL rankings in 2018 is 0.3754.
https://www.footballoutsiders.com/stats/ol/2018So 17.9% of variation in points scored can be explained by PFF's OL rankings in 2018 and 14.1% of variance in passer rating can be explained by PFF's OL rankings. I thought both would be around 20% so this isn't surprising me much. But for people who were expecting OL to be far more influential these numbers might be lower than expected.
Your last paragraph is spot on though. We don't have a good OL stat.
The part I made yellow above is what I'd be curious about, in terms of people's expectations. I don't get the sense that many people who tout the relationship between the Dolphins' offensive line and its QB play would also believe that over 85% of the variance in passer rating throughout the league is unexplained by an offensive line measure.
Like I said, however, the offensive line measure could be invalid. But then again so can people's perceptions of the Dolphins' offensive line's performance, in comparison to other offensive lines in the league.
"The Guy wrote:
The part I made yellow above is what I'd be curious about, in terms of people's expectations. I don't get the sense that many people who tout the relationship between the Dolphins' offensive line and its QB play would also believe that over 85% of the variance in passer rating throughout the league is unexplained by an offensive line measure.
Like I said, however, the offensive line measure could be invalid. But then again so can people's perceptions of the Dolphins' offensive line's performance, in comparison to other offensive lines in the league.
I'm not surprised it's that percentage. That's actually a fairly decent number, and as with an statistic that is basically the average of all teams if I'm not mistaken which means team to team there will be greater or smaller variance.
That is also just the offensive line. There are also other factors that play in and while having one or two be an issue is normal, when they are all issues that 14% variance varies and increases further in regards to overall QB impact.
Vertical LimitForum VeteranEh, throw all them statistics in the trash. I just need my eyes. the Jaguars are a far more complete team than us. Dj Chark is better than any receiver we currently have. The only weakness on their offensive line is Jawaan Taylor. Cam Robinson is a LT anyone would take long term. Norwell is a legit guard, Linder is a very good center. And Minshew was a badass in college, the only issue is he played in a system that is hard to evaluate the talent. In Washington State every quarterback they throw in, delivers big numbers. Even Luke Falk had big numbers there and he is trash on the Jets.
With that said i have also used my eyes on Josh Rosen and some hard truths are coming... im giving him 3 more games and i think by then i will already know if this guy is a franchise quarterback. Some of the boneheaded things he did last Sunday in the second half has already made me think a certain way.
So, basically I just realized the QB being the main person to control his own fate is a logical fallacy.
It's been said numerous times that passer rating and passer rating allowed are the most important aspects of winning.
Assume that to be true.
If a QB exists in a vacuum, how can the defense impact him? If winning their individual assignments has no real impact, how can you possibly build a pass defense?
Now take the next leap in logic. By winning their individual matchups, offensive players are helping their QB thus impacting his play on the field and overall rating.