Ryan Tannehill Analytics Profile Video.

[Digital]

200 years no, sample size is the number of observations included (QBs in this case) and again were talking about 1969 so QBs with only a 3 year career will be included. Actually 4 years can show normalization of all of those crucial factors depending on the sample size, (bigger sample more qbs with Tannehills situation) the author is reserving judgement on Tannehill for this year which will be his 6th season.

No, the point of discussion is the equality of the OL. We aren't talking about nearly enough time to normalize.

When the Author mistakenly equates all OL as equal, he makes a fatal flaw in analysis. He incorrectly assumes it is normalized. As I mentioned, given the career of offensive linemen and the frequency with which teams change their OL, 4 years is not nearly enough to expect a normalization. More importantly, we have lots of metrics to determine whether there actually HAS BEEN comparable performance, and the Author ignores them in favor of his biased generalization. The 200 years does not represent how many throws Tannehill makes ... obviously, as the contributing factor that is making such a dramatic difference is the Offensive Line. What we want to normalize is the OL performance.

Here's how it works. When you do an analysis where you assume that all other variables are held constant, you can do either of two things: 1) Bury your head in the sand and just assume it's all good even if you know that not to be true, or 2) Use a period of time where those data points normalize. The author chose the first option, which is to mis-represent the facts about the OL in front of Tannehill and act as if it had no effect on Tannehill's play. This is a mistake by the Author that fundamentally taints his attempted analysis.

It's just bad analytics. If you are not going to evaluate it--and there are a lot of available metrics to use to evaluate it--then you are obligated to mention that you didn't consider it and at least mention that it can have a significant impact on your analysis. But then again, it doesn't make the Author look professional, so he didn't do it.

Hey, like whatever you like man. But calling this analytics is kinda like calling the Browns a good NFL team ... you can love 'em all you like, but they are not a shining example of a good NFL team. That's my $0.02.

 

[Eesti]

The Miami Dolphins overachieved last year all while missing a Pro Bowl safety, 2 starting linebackers, starting DT, a starting CB, s starting RB, a starting QB and center...all while learning two brand new systems with a brand new coaching staff and rookie head coach ? Is this news? LOL

 

[J. David Wannyheimer]

The difference between touchdown percentage and interception percentage is the best predictor of team success in the NFL since 2004. Better than passer rating, better than yards per pass attempt, and better than (yes), even A/YA. At least it was the last time I checked, in 2015. Football is about scoring points and preventing the opposing team from scoring points. That much is pretty stupidly obvious.

And I'll tell you something about ALL of these statistics: outside of an extreme few fringe cases, quarterbacks perform best in these efficiency statistics when they are in balanced offenses working with a competent rushing attack.

When you can run the football, you pressure and fatigue the opposing pass rush. You create more favorable passing situations. You create a more dangerous play-action game. You keep the opposing team guessing, you wear them down, and you reduce the number of risks you have to take throwing the football.

That's why teams with strong running games usually have quarterbacks who throw a high percentage of touchdowns (they throw the ball less while moving downfield), they produce a higher yards per pass attempt (more favorable down and distance, more defenders stacked on the run), and they get sacked less frequently (less pass rush opportunities for the opposition).

As Ren so rightly points out in this thread. analytical observation is a tool. You observe quantifiable data points and see how they trend and if they match up with other data sets you have collected in the past. When you take analytics and use it to try to make proclamations like "Player X is elite," you're not just missing the boat, you're missing the entire goddamn sea.

 

[Jerrysanders]

The difference between touchdown percentage and interception percentage is the best predictor of team success in the NFL since 2004. Better than passer rating, better than yards per pass attempt, and better than (yes), even A/YA. At least it was the last time I checked, in 2015. Football is about scoring points and preventing the opposing team from scoring points. That much is pretty stupidly obvious.

And I'll tell you something about ALL of these statistics: outside of an extreme few fringe cases, quarterbacks perform best in these efficiency statistics when they are in balanced offenses working with a competent rushing attack.

When you can run the football, you pressure and fatigue the opposing pass rush. You create more favorable passing situations. You create a more dangerous play-action game. You keep the opposing team guessing, you wear them down, and you reduce the number of risks you have to take throwing the football.

That's why teams with strong running games usually have quarterbacks who throw a high percentage of touchdowns (they throw the ball less while moving downfield), they produce a higher yards per pass attempt (more favorable down and distance, more defenders stacked on the run), and they get sacked less frequently (less pass rush opportunities for the opposition).

As Ren so rightly points out in this thread. analytical observation is a tool. You observe quantifiable data points and see how they trend and if they match up with other data sets you have collected in the past. When you take analytics and use it to try to make proclamations like "Player X is elite," you're not just missing the boat, you're missing the entire goddamn sea.

TD/Int ratio for QBs is the no 1 predictor of wins for NFL teams, which is the reason the author uses that as his number 1 stat metric.

 

[Sherif]

Thanks for the post; The statistics were not interesting or meaningful but a good post for the offseason. I felt the guy had an agenda as indicated by digital. Line charts of Tannehill's progression would have been helpful to see a trend and would have been easy to produce. Clearly the data showed improvement in 14 and 16 with a dip in 15 but the overall line trend would have been upward. He damns T'hill with the Int/ TD and the ANY/A ratio but there should have been a deeper dive into each Int if you are going to ding him with a minus * 45 ratio per Int. Perception may be that T'hills YPA is low but statistics show otherwise.

http://www.espn.com/nfl/statistics/player/_/stat/passing/sort/yardsPerPassAttempt

 

[tazthenomad]

Fantasy driven analytics is different from win outcomes predicator Analytics. Kirk Cousins is a Elite in Fantasy Leagues but in these type of analytics he's just above average.

This is the basic fallacy in your use of this statistical method. It is a predictor ONLY if you are saying that the player is not changing.

The correct tools for showing changes from a past established by averages is a "run chart" which shows trends and the based performance deviating from means and using established rules like times above or below the average, position by standard deviation, etc.

 

[J. David Wannyheimer]

TD/Int ratio for QBs is the no 1 predictor of wins for NFL teams, which is the reason the author uses that as his number 1 stat metric.

Please cite your source. I compile my own data, and my sample was only for one specific ten year period. And my conclusion was based on the rate differential, not volume statistics.

 

[Gsmack_42]

Except in 2015 where Manning had more int than TD, and still won the Super Bowl.

Please cite your source. I compile my own data, and my sample was only for one specific ten year period. And my conclusion was based on the rate differential, not volume statistics.

 

[Sons Of Shula]

Martellus Bennett ✔@MartysaurusRex
- Not worried about fantasy football or stat predictions. No ****ing algorithm can predict the game of football. So sit back and chill

 

[finsfan43]

TD/Int ratio for QBs is the no 1 predictor of wins for NFL teams, which is the reason the author uses that as his number 1 stat metric.

So Brett Favre and EIi Manning should not have won Super Bowls. Here are Int\TD's. By your logic with that ratio of Int\TD's being so high they should not have won many games or Super Bowls?

Brett Favre Int\TD 1 Super bowl Win
336\508

Eli Manning 215\320 2 Superbowl Wins

Ryan Tannehill 66\106 Hopefully multiple - But he should never every win if the int\TD ratio is high..Oh wait, maybe just maybe there are other factors involved. Such as Oline play, COACHING (Play Calling), WR tipping passes. A D that sucks.

If a QB only throws 1 INT a game and the offense scores 28 points and say 3 are thrown by the QB so it is a 1\3 ratio and the D gives up 31, how is the QB INT\TD ratio even a valid point?


#QBWINZ?