1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Titans to start Ryan Tannehill

Discussion in 'Miami Dolphins Forum' started by bbqpitlover, Oct 16, 2019.

  1. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    @brgridiron
    Titans are playing their best football of the year. 5-1 record, Tannehill with 15 total TDs in last 6 games

    @JaysonBraddock
    Ryan Tannehill's last 6 games
    5 wins 1 loss
    72% comp
    243 pass yds per gm
    12 passing TDs
    3 INTs
    3 rushing TDs

    @jwyattsports
    @ryantannehill1 has passed for 1,458 yards, 12 touchdowns and three interceptions with a passer rating of 117.1 in six games since taking over the starting quarterback job in Week 7. @Titans

    @Michael_Nania
    Ryan Tannehill goes to 5-1 as a starter with a 100+ passer rating in five of those games and a 130+ rating in each of the last three. I wonder what he would accomplish if he got to play under the tutelage of QB guru Adam Gase!

    @JWyatt
    *RT passer rating in his first 5 starts is 114.9. It's the best in #Oilers/@Titans franchise history for a QB's first 5 starts with the franchise. The previous record holder was @Hasselbeck, who joined the #Titans in 2011 & produced a 95.9 rating in his 1st 5 starts.

    The #Titans have averaged 445.3 yards from scrimmage per game (1,336 yards / 3 games) since Week 9 -- best in NFL; League avg: 358.9.


    Per JWyatt @MySportsUpdate
    #Titans blow it open. Ryan Tannehill finds Kalif Raymond for a 40-yard TD. With a win, Tennessee will improve to 7-5.

    @2ndandVictory
    AND RYAN TANNEHILL PUTS THE ****ING DAGGER INTO THEIR HEARTS.

    @ckparrot
    Ryan Tannehill has completed 27 of 41 passes for 458 yards, 1 TD and 1 INT when NOT working from the shotgun. That's a 101.5 passer rating.

    @MikeMiracles
    The Tannessee TitanHills have put up at least 31 points in 3 straight games. This is only the second time this team has managed that since they arrived in Nashville in 1997.

    @LateRoundQB
    Since Ryan Tannehill became starter, here's a list of quarterbacks (minimum four games played) who have averaged more fantasy points per game:
    1. Lamar Jackson


    @JeffDarlington
    Ryan Tannehill told me he doesn’t feel any different — same QB he’s always been — but his arm during warmups looked as precise as I can remember. Maybe it’s the confidence of winning or more of a “f*** it” mentality... but he’s slinging it. Perfectly placed passes on every route.

    @JeffDarlington
    Probably the stat that will baffle Dolphins fans the most when it comes to Tannehill’s resurgence: The Titans were 19th in red zone efficiency when Ryan took over. In just five starts, he has moved them all the way to... first in the NFL.


    https://www.cbssports.com/nfl/news/...ng-his-way-into-a-franchise-tag-at-tennessee/

    Tannehill has been a top-five quarterback in virtually every key metric since replacing former No. 2 overall pick Marcus Mariota under center, leading Tennessee to a 4-1 record and putting them in position to possibly win the AFC South. Since Tannehill took over, the Titans are averaging 50 more passing yards per game and 44 more rushing yards, with the quarterback pushing the ball deep downfield, getting multiple receivers involved and making it more difficult to load the box against star running back Derrick Henry.
     
  2. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    What Brady has shown on the contrary is the ability to play very well with receivers with various traits, as he adjusts his game to fit their abilities. Note for example that the highest season passer rating of his career was in 2007, when he had the fleet-footed Randy Moss and Donte Stallworth, paired with Wes Welker, and their offensive game featured a great deal of downfield passing. Note the following article for example.

    https://patriotswire.usatoday.com/2...s-that-tom-brady-is-only-a-check-down-passer/

    What’s more likely happening in my opinion is that he is simply getting old in NFL terms, and the performance of quarterbacks in the NFL tends to decline suddenly rather than gradually. Note the following article for example.

    http://archive.advancedfootballanalytics.com/2011/08/how-quarterbacks-age.html?m=1
     
  3. AGuyNamedAlex

    AGuyNamedAlex Well-Known Member

    1,696
    1,475
    113
    Sep 12, 2015
    Honestly if you watched Brady last year it was pretty obvious the tank was close to empty.
     
    KeyFin likes this.
  4. resnor

    resnor Derp Sherpa

    13,278
    7,494
    113
    Nov 25, 2007
    New Hampshire
    Brady can "adjust" to throw the ball to the greatest receiver of all time? Huh. So could Culpepper.
     
    Mexphin and danmarino like this.
  5. Stringer Bell

    Stringer Bell Post Hard, Post Often Club Member

    43,939
    21,516
    113
    Mar 22, 2008
    Best models can predict with ~75% accuracy (thats based on performance within a single season, and based on much more than just a single stat).
     
  6. BevoPhin

    BevoPhin Well-Known Member

    1,436
    733
    113
    Mar 15, 2013

    Good for him. I want him to succeed. Just didn’t happen here.

    It’s like when you’re GF doesn’t cook, clean, or give head when you’re together. Then after your break up, what do you think she’s doing with the new dude? Cooking, cleaning, and giving head. Sooner or later, she’s going to stop with the new dude as well.
     
  7. cbrad

    cbrad . Club Member

    7,398
    8,683
    113
    Dec 21, 2014
    Interesting.. you have a link? I'd like to see their general approach.

    btw.. my first thought was this: 75% seems way too high because I would think you could beat Vegas if you could do that. So IF what you're saying is true my initial guess is that they're not trying to predict every game. But I'll wait to see which models you're referring to.
     
    Last edited: Dec 2, 2019
    Pauly likes this.
  8. KeyFin

    KeyFin Well-Known Member

    6,340
    8,302
    113
    Nov 1, 2009
    I was thinking the same thing- if it averages .75 accuracy over the course of a season (or even the close of a season) then it's worth betting on. I mean, put a grand on each game (16k) and you win 75% of those, that's 8k total profit A WEEK!
     
    resnor and Pauly like this.
  9. KeyFin

    KeyFin Well-Known Member

    6,340
    8,302
    113
    Nov 1, 2009
    Tannehill sits at a 113.9 passer rating on the year, which is 2nd overall only to Cousins. He passed Wilson and Jackson this past weekend. However, he's also the 19th most sacked QB this season (24) despite playing only 6 games...average that out for the year and he's still the most sacked QB in football (currently it's Murray with 41 sacks/12 games). Not that it matters with the numbers he's putting up....just added that to show his pocket presence and protection hasn't necessarily increased this season.

    For ESPN's Total QBR, Tannehill is 16th overall...one slot ahead of Brady. Fitzpatrick is a surprising 8th in Total QBR, by the way. For last week, Fitzpatrick was 3rd in the league while Tannehill was 22nd...QBR just doesn't like the guy I guess.

    Whether this is a fluke or a genuine resurgence, I'm genuinely happy to see him move on and flourish for however long it lasts. I'm looking forward to him playing his first playoff game this year as well...it's been great having another team to root for this season that's not named the Dolphins!
     
    Last edited: Dec 2, 2019
  10. cbrad

    cbrad . Club Member

    7,398
    8,683
    113
    Dec 21, 2014
    Have to take into account moneylines so you would win a lot less. But regardless.. IF it's really true you can win 75% of the games, then I'm going to look at archived odds and see if that would have meant you could beat Vegas. Intuitively I doubt this.
     
  11. Pauly

    Pauly Season Ticket Holder

    3,145
    3,065
    113
    Nov 29, 2007
    One of the (many) reasons ESPN’s QBR is garbage is the hate it has given to Tannehill over his career. According to it Tannehill belongs in the Blaine Gabbert/Nathan Peterman category of QBs which is plainly false. Tannehill has been* an average QB who takes too many sacks. His traditional NFL passer rating reflects this.
    According to QBR a team QBed by Tannehill should consistently be in the 4-12 range unless they have an exceptional defense. According to NFL passer rating a team QBed by Tannehill should be in the 8-8 range.

    * i.e. prior to 2019
     
    Tin Indian, resnor and KeyFin like this.
  12. KeyFin

    KeyFin Well-Known Member

    6,340
    8,302
    113
    Nov 1, 2009
    Well, QBR obviously takes sacks into account and ignores offensive line performance...that would explain a lot of the Tanne-hate from ESPN. If he's supposed to be a 4-win QB though and he's won 4 out of 5 total, I think it's pretty safe to say that the rating system itself sucks.

    And not tot take anything away from our boy Fitzpatrick, but he's not playing like a top-8 QB for the season. He's carried this team and balled out as much as possible under the circumstances, but the reality is that he's running for his life and having a blast doing it. I'd call him a player of the year in my book....but I wouldn't call him a top-8 QB. =)
     
    resnor likes this.
  13. Pauly

    Pauly Season Ticket Holder

    3,145
    3,065
    113
    Nov 29, 2007
    Having built predictive models and used predictive models this is bunkum.

    1) passer rating is not a stat, it’s a blend of passing attempts, completions, TDs and interceptions. So passer rating is not a “single stat”, it is 5 stats that have been combined into a single rating.
    2) as cbrad and keyfin say, show us your workings.
    3) predictive models have an optimum number of variables. With enough variables and assumptions you can hind-cast anything with astonishing accuracy. Many variables however are interactive or are measuring the same things just in different ways, which is why adding more variables to a predictive model often has the effect of making it less accurate. The most robust predictive models tend to be those with the fewest moving parts.
     
    cbrad likes this.
  14. Sceeto

    Sceeto Well-Known Member

    12,456
    5,069
    113
    Oct 13, 2008
    New York
    Probably a big part of Tanne's resurgence or whatever is that he is no longer being coached by that little bugeyed douche of a HC. Look at what that little scammer is doing with Darnold.
     
    resnor likes this.
  15. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    Apparently you didn’t read the article, which goes well beyond Randy Moss.
     
  16. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    That isn’t supported by any statistic.
     
  17. resnor

    resnor Derp Sherpa

    13,278
    7,494
    113
    Nov 25, 2007
    New Hampshire
    I don't need to read the article to comment on something you wrote.
     
  18. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    What I wrote regarding Randy Moss said “note for example,” meaning that there’s more to the story than just Randy Moss.
     
  19. Stringer Bell

    Stringer Bell Post Hard, Post Often Club Member

    43,939
    21,516
    113
    Mar 22, 2008
    Just to be clear - 75% referred to predicting the winner of the game, not who would win against the spread. Nobody is coming close to 75% accuracy against the point spread. The point spread predicts the winner of a game ~70% of the time. So if someone can beat Vegas 53% of the time (which is realistic), then you're close to the 75%.

    Here is an example of a model getting > 70%:
    http://cs229.stanford.edu/proj2016/report/WadsworthVera-PredictingPointSpreadinNFLGames-report.pdf

    Thats just what I found by a quick google search. Please don't roast me if theres something wrong there. Back in the day I know Brian Burke had a pretty solid model that was based on a logit regression. I can find more, or even show an example, if you all really are interested. From my experience, the key to an effective model is effectively incorporating opponent adjustments. I'm curious how people have started to incorporate the player tracking data thats becoming available.

    This shouldn't be that controversial. If passer rating can predict at 60%, you should be able to achieve higher results by removing the variance that exists within that stat (volatility of TD or INT %, for example).

    Yes, I agree - obviously overfitting is a concern, especially when many variables in football are corollary. You just have to be careful and understand the interplay.

    That should not dissuade anyone from leveraging all the available data, especially now that player movement tracking data is becoming available.
     
    Last edited: Dec 2, 2019
  20. resnor

    resnor Derp Sherpa

    13,278
    7,494
    113
    Nov 25, 2007
    New Hampshire
    Actually, that's you saying that Moss is an example of Brady's adjusting his game. I'm still not clear on how one has to adjust their game when throwing to the greatest receiver of all time.
     
  21. cbrad

    cbrad . Club Member

    7,398
    8,683
    113
    Dec 21, 2014
    Won't roast you, but the key part of their article/report is under the section "Features". There they say the INPUTS (what they call "features" because they're using machine learning and the individual components of a neural network are "features") to their models are:

    1. Each team's win streak
    2. Each team's NFL ranking from the previous season
    3. Each team's ESPN preseason "power ranking"
    4. The predicted Vegas spread

    They're using the Vegas spread to predict outcomes! And they explicitly say that the Vegas spread was the most impactful input variable (see the graph in that section). So.. no duh they can get 70-75%.


    Also, two other things from a scientific point of view:

    1. The reason Logistic Regression proved useful is precisely because the math behind transforming rankings to measures uses logistic regression. I've said many times before that ranks aren't measures because the difference between ranks 1 and 2 isn't a priori the same as the difference between rank N and N+1, while any measuring instrument such as a ruler doesn't have this flaw. Logistic regression is the basis behind math like Rasch analysis that fixes that problem.

    2. Most of their methods have TONS of parameters. These are machine learning methods, and the one that worked the best called Ada Boosting is machine learning ON machine learning lol. So there is a serious danger to overfitting with these models. In other words, just because one model worked on one test set does not in any way suggest it will work on another test set.

    So.. a purely analytical approach that uses commonly available stats with NO Vegas spreads (that's a huge advantage) or rankings by experts who are pretty good if it's just one game they have to predict? Not sure how good you can do there. Probably better than 60% but I'll start to be impressed if they can do 65%+ on any test set.
     
    Last edited: Dec 2, 2019
    The Guy likes this.
  22. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    The point I responded to was this one:

    So the point is that Brady can win with receivers with various traits, including deep speed a la Moss and Stallworth, because he has not only the ability to be accurate in the short area, but also to hit the longball.

    From the article I linked:
    So again the point is that Brady has adjusted his game to fit the characteristics of his receivers. When Randy Moss was on his team, he wasn't stuck functioning like Chad Pennington, leaving Moss underutilized. He passed the ball in such a way that Moss's unique skill set was utilized.
     
  23. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    Huh. Then maybe you should ask yourself why you couldn't be the man to get that from her to start with...
     
  24. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    He was mismanaged from the start by Philbin. With Gase, he was only healthy for less than a dozen games in three seasons. 2016 started off well until his knee got caved in. Last year, he was still recovering from the knee when he almost had his shoulder ripped off against the Bengals. He's healthy and having fun.

    And considering what TN is doing, Ross should have hired Dan Campbell or one of the Shannahan's. The offenses are very similar. Run heavy and lots of PA. They don't ask him to pass 35+ times a game and aren't demanding that he stay in the pocket, nor are they screen pass heavy, or calling lots of slow developing plays. When they do, they roll him out like on the 40 yard bomb TD yesterday.

    He's finally in a system that fits him.
     
    danmarino, xphinfanx, Pauly and 2 others like this.
  25. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    If it’s true that he’s finally in a system that fits him, then it’ll be interesting to see how he performs once the league accrues a significant amount of film of him playing in that system.

    This is again why it’ll be important to follow his performance for the next two to three years, to determine whether he’s experiencing an Andy Dalton 2015 season this year. Whatever put Andy Dalton at an elite level in 2015 hasn’t done that again in any other season of his career.
     
    Last edited: Dec 2, 2019
    Pauly likes this.
  26. resnor

    resnor Derp Sherpa

    13,278
    7,494
    113
    Nov 25, 2007
    New Hampshire
    Dude just stop. Reading that makes it sound like it's hard to throw the ball to a 6'6 4.4 speed receiver.

    Sorry...4.25
     
  27. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    The point is that you either have those kinds of receivers on the field, or you don’t. Not all quarterbacks would be able to fully utilize the skill set of such a receiver.
     
  28. Stringer Bell

    Stringer Bell Post Hard, Post Often Club Member

    43,939
    21,516
    113
    Mar 22, 2008
    Not sure why the Vegas line wouldn’t be an acceptable input to predicting outcomes? Maybe I missed something earlier in the thread.
     
  29. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    You're missing the obvious. Complimentary football. Something that's been rarely seen here the past decade.

    In Cinci, Hill and Bernard combined for ~2100 yards rushing and receiving and 14 TDs.

    Dalton only played in 13 games and was in his second year of Hue Jackson's offense/system in 2015. Hue was gone after that season.
     
    resnor likes this.
  30. cbrad

    cbrad . Club Member

    7,398
    8,683
    113
    Dec 21, 2014
    Vegas lines and ESPN power rankings aren't football stats.

    If you go back to the post of mine you initially responded to (post #791) my claim was that you can't do much better than 60% predicting wins in the NFL with a single stat, and you'll probably only do slightly better with a combination of stats. So we have to restrict ourselves here to models using only football stats as inputs. I definitely agree that collective human intelligence (Vegas lines) is better than just using stats.
     
  31. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    It’s also likely that the Vegas opening line is based on a regression model. The closing line, if it has changed, would partly represent collective human influence.
     
  32. cbrad

    cbrad . Club Member

    7,398
    8,683
    113
    Dec 21, 2014
    Well here's how one betting site describes it:
    https://www.oddsshark.com/sports-betting/who-sets-line
    Sounds about right. If my livelihood depended on setting good opening lines I wouldn't rely on fancy algorithms alone.
     
    Irishman and The Guy like this.
  33. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    You’re underscoring my point. If a quarterback is that dependent on some conglomeration of surrounding variables for his success at that level, then his success at that level becomes less likely to be replicated as those variables potentially disappear for one reason or another.

    Hell, if the lynchpin for Ryan Tannehill right now is let’s say Derrick Henry and Derrick Henry gets injured at practice tomorrow, you could very well see Ryan Tannehill fall off a cliff starting next Sunday.
     
    Last edited: Dec 2, 2019
  34. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    And vice versa, which is what the Titans had prior to the switch from Mariota. Henry was averaging 69.33 ypg. Since week 7, Henry is averaging 124.33 ypg.

    He and Tannehill are complimenting each other.

    That's hardly a startling revelation though. No one succeeds in a vacuum. It's why football is the ultimate team sport.
     
    Mcduffie81 likes this.
  35. Dol-Fan Dupree

    Dol-Fan Dupree Tank? Who is Tank? I am Guy Incognito. Club Member

    34,164
    24,445
    113
    Dec 11, 2007
    It is almost like because of Tannehill, the team is playing better.
     
  36. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    Right, but there are quarterbacks who function consistently at the elite level individually despite changes in their surroundings, and that can be demonstrated only over several years of play.

    The question still remains for Tannehill whether he can somehow accomplish that during the remainder of his career, or whether what he’s demonstrating currently is something much more fragile and prone to collapse.

    It’s probably the case that the majority of average quarterbacks in the league could play at a level significantly higher than their norm if they have the surroundings optimal for them, but if those surroundings are either unlikely to be assembled or unlikely to be sustained, then that’s hardly a feather in the average quarterback’s cap.
     
  37. cuchulainn

    cuchulainn Táin Bó Cúailnge Club Member

    11,040
    9,534
    113
    Sep 7, 2012
    Hattiesburg, MS
    Elite? Where did you get that from?

    Ryan Tannehill isn't elite and never will be. Any talk of such is a fallacy so someone can say they told you so.
     
    Mcduffie81 and resnor like this.
  38. Irishman

    Irishman Well-Known Member

    334
    356
    63
    Oct 16, 2017
    High Point, NC
    He seems to be playing like an elite QB. Hes rated as one of the top 5 by many networks and sports outlets during the time he has played this year, so this is not just my opinion.

    Now can I tell you "I told you so"?
     
    The_Dark_Knight likes this.
  39. Dol-Fan Dupree

    Dol-Fan Dupree Tank? Who is Tank? I am Guy Incognito. Club Member

    34,164
    24,445
    113
    Dec 11, 2007
    One thing I do agree with Guy is that Tannehill needs to do this multiple years to be considered elite.

    Though, IMO, this does prove the fact that Tannehill does improve the play of the people around him. Which he did in Miami.

    There is a contingent of people who like acting like Tannehill is a woman scientist in the 1800s and doing everything in their power to make sure they don't get credit.
     
    Hiruma78 and cuchulainn like this.
  40. The Guy

    The Guy Well-Known Member

    1,643
    841
    113
    Oct 1, 2018
    If he plays at the level he's playing at now, over several years, despite significant changes in his surroundings, then one could make the argument that his surroundings in Miami were so poor or sub-optimal for him that they functioned like a tremendous ball and chain on an otherwise elite quarterback.

    But again, it'll take a good bit of time for that argument to be convincing.
     

Share This Page