Where does Tannehill rank among quarterbacks today?

http://www.coldhardfootballfacts.com/content/yearly-league-wide-passer-rating-passer-rating/6205/

EDIT: Sorry.. just noticed that doesn't go up to 2014.. it does give you an idea of where the average is though.. low 80's.

EDIT AGAIN: OK.. league-wide passer rating for 2014 was 88.9!! So it went way up from 80.9 in 2007 in that link above.
http://proplayerinsiders.com/nfl-pl...view-record-breaking-2014-nfl-regular-season/

 

I don't take it away from Flacco, but I also don't believe it trumps his inconsistency. I also don't see his postseason as impressive as you do. I saw his stats inflated by long passes that were great plays by WRs rather than QB skill. So yes, I have Tannehill above Flacco. Flacco has had the benefit of far better supporting cast, coaching and organization and more than double the NFL experience and produces about as well as Tannehill. I'll take the younger, still ascending player easily.

 

How do you know its not an aberration? Probability doesn't mean it's not an aberration. What if you flip 100 times to get that sequence? 1000x?

Not to mention, it seems very flawed to try and compare a human to the probability of coin flips. And, Flaccos averages tells us that, at least thus far, four games on a row well over 100 IS an aberration.

 

Just talking about post-season here because the entire question was about aberrations in post-season performance.

 

If the probability of flipping 3 heads in a row is 1/8 and you get that only 1 out of 100 times then yeah simple statistical tests will show it's too unlikely to support the hypothesis the coin is fair.

Humans aren't coins but the basic principle behind hypothesis testing applies to both. Basically you'd have to use Flacco's stats to estimate a distribution of passer ratings in the postseason from him and then ask how unlikely the 2012 run is. There are methods of doing that. I'm just saying that the 2012 run is probably not going be looked at as too unlikely to be from a "normal" Flacco in the postseason.

 

As Raphael pointed out, the answer is sample size.

Group A
Joe Flacco's career 112 regular season starts = 84.8 QBr
Joe Flacco's career post-season 15 starts = 88.6 QBr
Joe Flacco's 2012 post-season 4 starts = 117.2 QBr

Group B
Joe Flacco's best season (2010) 16 starts = 93.6 QBr
Joe Flacco's worst season (2013) 16 starts = 73.1 Qbr
Joe Flacco's best and worst seasons ('10, '13) combined 32 starts= 82.2 QBr
Joe Flacco's career minus rookie season and worst season 80 starts = 88.3QBr

The dramatic separation between the largest sample size and the smallest sample size in group A, suggest one statistical grouping isn't accurate. Which one is most trustworthy? It has to be the larger sample size. The other two sample sizes are so small, any statistical variance can manipulate the numbers in either a positive or negative direction. But I want to be fair... If Joe Flacco's 4 game sample size in the 2012 is too small to be considered accurate based on what we know about his entire career, and therefore might be aberration, let's remove two more aberrations from Flacco's career.


In Group B, I show Flacco's best season which 93.1 QBr. Almost 10 points higher than his regular season average for his career. I also show Joe Flacco's worst season, 73.1 QBr. Almost 10 points lower than his regular season average for his career. Adding the stats together for both seasons, Flacco comes out to an 82.2 Qbr. In line with his career average of 84.8.

Those 112 games of Flacco's career (84.8 QBr) consist of 16 rookie starts and Joe Flacco's worst season (16 starts, 73.1 QB). Rookie season's are almost universally weighted less than any other season. All rookies get better, even if they never turn out to be good in general. And Joe Flacco's 2013 was also an aberration from his usual statistical output. Excluding both, Joe Flacco would have an 88.3 QB rating over 80 starts. Very much in line with what his playoff numbers suggest he is (88.4).

All the numbers add up, DJ. Flacco is best described as a mid-upper 80's QB. Even when you manipulate the numbers to try and get the most accurate picture of Joe Flacco as a QB, you find the picture was pretty accurate from the beginning. Raphael's assessment of Flacco is pretty accurate too, IMO. Joe Flacco's greatness is being washed out by inconsistency due to poor play. And Flacco's career year by year numbers show a similar pattern, supporting that claim. Flacco is up and down. One year he's rated in or near the 90's, the next, lower 80's. As a mid-upper 80's rated QB, there is more good than bad though certainly. You could do a lot worse than a Joe Flacco. Don't misunderstand and think I'm suggesting he's bad. I'm not. At all. I'm saying, he's inconsistent, with more moments of greatness than bad play.


I will say this though, DJ... There is an intangible to Flacco that I do think works in his favor. He personifies his nickname, Joe Cool. The guy is a robot. Seriously, he might be the worlds first true A.I. Flacco's mood and personality don't ever appear to change regardless of his play, positive or negative. He's got the same expressionless, robotic face no matter what. And in high pressure moments or playoff games where it's win or go home, I certainly think that's an advantage working for him in comparison to defenders who might be feeling the pressure.

 

playing four games in postseason play plus his other games in postseason that he played at a high level to win a championship/afc championship games, is a big enough sample size to not deserve to be categorized as an aberration.

 

I believe the ranking was meant to be a current ranking, Im debating the logic that has ryan ahead of Flacco, if you would take ryan over flacco moving forward to run your franchise for lots of reasons and variables then thats a different thread right..

 

There is pretty much nothing you can point to that actually says that Tannehill is not as good as Flacco right now.

 

lol..ok, you use stats to make your football team...good to know.

 

Context always matters, especially w/stats. But, what is your specific issue with the stats being used in comparison between Tannehill and Flacco? From what I can tell and correct me if I'm wrong, Baltimore made the playoffs and Miami didn't. And Flacco played well in the playoffs in 2014, while Ryan watched them on TV. If that's the basis of your argument, it's flawed. Miami's defense is the reason we missed the playoffs. Whether it was giving up games in the 4th quarter when we had the lead (GB, Detroit, Denver) or being gashed for yards and points during December (28, 41, 35, 37 points allowed) while Ryan had a 95 QB rating in the same month.

I don't believe we had any common opponents in 2014, but The AFC North and AFC East did play in 2013... For comparison's sake.

Flacco v. AFC North : 78.1 QBr
Tannehill v. AFC North: 88.6 Qbr

Flacco v. AFC East: 63.4 QBr
Tannehill v. AFC East: 73.5 QBr

And for the record, I agree that you, Flacco > Tannehill at this very second. As in, prior to week 1 of the 2015 season. When I ranked the QB's, I gave Flacco the edge because of his last two post-season performances (2014, 2012). It was for me, a tiebreaker of sorts. But I don't think playoff numbers prove Flacco is the definitive superior QB when Tannehill hasn't had the chance to get there. Not to mention, the experience gap between the two. And I certainly won't hold Tannehill not making the playoffs against him these last two years, given the collapse of defense and offensive line in 2014 and offensive line in 2013.

 

I don't know. A current ranking is about who would you take for this coming season. I would take Tannehill over Flacco.

 

I don't make decisions based on stats, I make them based on performance levels plus situations..at this point I've seen better individual performance levels in more extreme situations from Flacco then I've seen from Ryan..Doesn't mean Ryan cant get there..

 

I would think that a current ranking was based on what they've done.

 

All they've done is considered and a current ranking would be who you'd want to lead your team today.

 

Again, DJ, EVEN WITH an out of this world performance in the 2012 playoffs skewing the numbers upward, Flacco's averages in the post season match pretty similarly to his regular season averages.

 

lol..THERES NO SKEWING OF THE NUMBERS!!....thats four freakin games and championship....peyton manning has won one Super Bowl..

he actually did those things, I'm in a debate with someone who thinks performing in four playoff games at a high level is skewing his numbers, do you know how ridiculous that sounds..

skewing numbers could mean padding stats late in games when you've already been blown out and the defense is not competing like it was earlier and has changed its scheme making it so much easier to throw against...in game dynamics change stats.

I mean really, stop with the 4postseason skewing games thing....some qbs retire before they ever win four postseason games.

 

Why are you getting so emotional? Yes, it skews the numbers up. I think you're being intentionally obtuse here. His playoff averages, even with one fantastic postseason, are right in line with what we've already agreed are pretty pedestrian regular season numbers. Yes, he did play fantastic in 2012 for four straight games. That is not his norm. Yes he plays better on the postseason, but not like that. Now, you might have a point, if this season he makes the playoffs, and again goes on a four game tear, then we can start saying that 2012 wasn't so abnormal. Until that happens, 2012 remains an anomaly for Flacco. Kinda like Kitna playing out of his mind for one season. No one was proclaiming Kitna to be awesome after that one season.

 

The problem seems to me, DJ, that you've sort painted yourself into a corner here. You claimed that Tannehill should hope to be as good Flacco, without perhaps ever looking at the stats, and now you're trying to make a case outside of stats that Flacco is better.

 

OK.. I had some time to actually do the math to essentially prove Flacco's passer ratings for the 4 games in the 2012 playoffs are NOT aberrations, statistically speaking. I'm making this a bit of a lengthy post to actually explain the concepts and the math behind the calculation, if anyone is interested. If not, then the takeaway is that something called the "Central Limit Theorem" + basic hypothesis testing proves that Flacco's 2012 playoff run was not significantly different than would be expected from what he's done in the playoffs so far.

First of all, here's a link to the data:
http://www.pro-football-reference.com/players/F/FlacJo00/gamelog/post/

Now, let's do this calculation for two separate cases: 1) take all of Flacco's playoff history as the best estimate of Flacco's playoff ability, and 2) use only playoff history from 2010 onwards as the estimate (why bother? because it at least looks like Flacco was a different QB in the playoffs from 2010 onwards). The results are the same either way, but I will revisit this second assumption later on.

OK, some basic stats from the link above: For case 1: mean = 84.8 and standard deviation = 37.2. For case 2: mean = 104.7 and standard deviation is 19.3.

We need those stats because they are the best estimates of the mean and standard deviation of the "population distribution" = distribution of Flacco's playoff performance IF he and the Ravens played essentially an infinite number of playoff games in similar conditions as the ones he played in (during the respective years for the two hypotheses).

Once we have the mean and standard deviations of the population distribution, we invoke what's called the Central Limit Theorem (call it CLT for short), which is arguably the most powerful result in statistics. The CLT basically tells you what the distribution of sample means (means of samples taken from the population distribution) look like. For example, suppose we repeatedly took samples with sample size 4 (in our case it will be 4 because there were 4 playoff games in 2012) and calculated the mean rating each time. The CLT tells you what the distribution of those means looks like without you actually doing the experiment.

Specifically, the CLT says that the mean of the "sampling distribution" (distribution of means of samples taken) is the same as the population mean. The standard deviation is different however. The standard deviation of the sampling distribution is: (std of population) / (square root of sample size).

In our case, this means that for case 1: the sampling distribution has mean 84.8 with standard deviation (37.2)/sqrt(4) = 18.6. For case 2: the sampling distribution has mean 104.7 with standard deviation (19.3)/sqrt(4) = 9.65.

Furthermore, the CLT tells you that this sampling distribution has the shape of a Gaussian distribution (also called a "normal" distribution, or many might know it as a Bell curve). This is important because it allows you to use statistical tests designed for Gaussian distributions. In fact, we have simple rules of thumb that can circumvent the math now: we know that anything beyond approximately 2 standard deviations from the mean is "too unlikely to come from the same population distribution" (actually it's a tad less than 2 std away but we can ignore that here).

So what is 2 std away? In case 1, it's 84.8 + 2*18.6 = 122, while in case 2 it's 104.7 + 2*9.65 = 124.

And what is the average passer rating of Flacco during the 2012 playoffs? It's 118. 118 is well less than 122 and also less than 124, so like my intuition suggested, statistically speaking, having 4 playoff games as Flacco had during 2012 is NOT statistically significantly different than would be expected by Flacco under either case 1 or case 2. Oh, and yes having such a small sample size (Flacco's entire playoff record) to estimate the population distribution is always a problem, but that's all the data we have so that's all you can do.

Hope that let's that one point rest now (at least based on stats).


On to another issue (that I said I would revisit): You can show by a separate calculation (this I am NOT going to write all down) that case 2 is also statistically speaking justifiable. That is, there are tests that show Flacco's playoff ratings in his first two years would be considered an aberration if you assumed the data came from Flacco of 2010 onwards (in the playoffs only), and vice versa. So, you can actually also make the case that Flacco is a different QB in the playoffs from 2010 onwards.

 

Maybe I'm dumb...but it seems that you're using the 2012 season to prove that the 2012 season wasn't an aberration.

Also, maybe using the term "aberration" means something different to you...I was simply using it to mean, 2012 was much different than his other playoff years. No amount of numbers crunching will change that. To this point, Flacco has never had such a postseason.

 

I'm not really, just annoyed that you keep saying that 4 postseason games skews things.

 

i never looked at the stats until you all started using them to make your whole argument..

 

Its because anyone can have a 4 game stretch of good play. It does not matter that it is in the playoffs.

 

If you're going to consider Tanny's upward trend the past 3 years, you should argue Flaccos 2012 performance is skewing his playoff numbers. Flacco has been good his past two playoffs.

It's like saying Ryan's 2014 numbers is skewing his rating.

But hey, at least were off the argument the Baltimore D was excellent in the 2012 playoffs. That is a case of just looking at stats not realizing they gave up over 30 in 2 out 4 games ...

 

You can't a priori assume the conclusion. That is, you can't a priori assume that 2012 is an aberration and leave it out to estimate Flacco's playoff ability, which you have to do first.

Now, there is another approach, which is to ask whether two samples come from the same distribution, and allows you to slice the data up. The best approach is to use what's called the Kolmogorov-Smirnov test (say for two samples), but what that does is look at what's called the cumulative distribution functions for the samples in question, and I guarantee you the sample sizes in question here are way too small to do anything reliable with that.

The approach I used is the best I think you can do with statistics here. And whatever your use of the word aberration is, statistically speaking Flacco's 2012 playoff ratings are not an aberration (all that means is that according a standard way of interpreting stats, it's not an aberration).

 

He's only had 1 game that anyone would consider "bad" since 2010, against Pittsburgh. Other than his first two years, he's played damn well in the playoffs. I'm not sure where anyone gets the idea it's only 2012.

So out of his last 10 playoff games he's had only one game that was bad. It's not an aberration it is a pretty solid trend.

 

I actually already admitted that it's possible that Flacco is showing a trend in the playoffs that goes against his averages...

 

He's never strung together four games in one postseason like he did in 2012, that's all anyone is saying. No one is saying he plays badly in the postseason.

 

Sure. But you're arguing that Flacco is better based solely on your opinion of individual impact plays, which can't be corroborated like looking at stats.

 

it matters when it matches up with what you expected to see.

 

Id like to know the list in the modern era that has won 4 playoff games and posted a higher number in those four..I have no idea actually how many as I write this...just posin the question, not asking you to do it.

 

when Res called it that I thought the same thing, obviously not in the way you interpret it, your a smart dude, but it feels like I was right in saying it wasn't...So in my football world is wasn't and even in the statistical world relative to football it isn't?

 

thats what I've been trying to say, its more than enough sample size.

 

what does that even mean, sounds like you've been painted into a corner and now your trying to get out.

you think every qb who gets into the playoffs has to string together 4 strait games of above average play each time they get there to be recognized as a very good qb ?

 

you or anyone doesn't have to agree with me, I didn't even look at their stats to come to my conclusion, its just my opinion and how i evaluate..I use stats differently then someone like yourself..

I will admit that the stats that have been presented are encouraging in a way, it makes me think more, but by no means will I let it dictate to me what i see with my eyes..and that simply is Flacco should be rated higher than ryan because I Have seen Flacco do things under pressure that I haven't seen ryan do..

Theres different ways to win games..some qbs do things different..,Rafs for example is making the point that ryans consistency over time is a bit better at the short passing game then flacco yet doesn't have the timely playmaking ability that flacco has...well, I will take the latter and sacrifice a little bit of that short game consistency.

 

Yeah, statistically speaking a 118 average rating for any 4-game stretch in the playoffs is not significantly different from what would be expected given Flacco's overall playoff record. That of course means Flacco's 2012 playoff run wasn't an "aberration" in that sense.

So the conclusions from the stats jibe with yours, but keep in mind that the arguments (yours vs. stats) are different. So, while I think you can (and should) use the stats argument to support your conclusion, keep in mind that this doesn't imply the specific reason you gave (about impact plays) is the reason for result. It might be, and it might not be.

Now, if you can define "impact plays" in a way that: 1) you're comfortable with, and 2) is recorded on the stat sheet, I can help with the analysis if we can get the data.

 

I think Tannehill will have a top 7 finish in 2015 statistically speaking. I think he will become the defacto challenger to Andrew Luck in the best of the under 30 class. Tannehill is having an almost classic career curve up to this point. The rule of thumb is always give a player three years to see whether he belongs and he has improved every year and year three was his statement year "I belong". What generally happens in year four of those kind of players is they go from proving to everybody and themselves that they belong to the game slowing down and the confidence rising of knowing that you belong and their peak years begin. I expect another arch up in the trajectory and it will be pretty close to Tannehill's plateau and we will see if we have the next Aaron Rodgers or the next Tony Romo but I expect him to settle into the top ten permanently this year with a shot at top five in a year or two when the near 40 guys retire

 

Who cares?

You're talking a 4 game stretch out of a 100 or so games. That's not a trend. That's not the norm.

 

I cannot agree with a stat that claims that a 118 rating is not significantly different from an 84. I don't care about standard deviation...there is so much different between an 84 rating and a 118 rating. They aren't even close. To argue that they are in any way similar is ridiculous.