Why Josh Allen will demonstrate the flaw in the analytics approach to football

[Trogdor]

8 hours ago, BB@Shooter said:

Staying positive as usual. I think it is OK if they work it in. It is not like Gregg wasn't trying to throw schemes at him the way it was. Gregg tried to put Allen on his ass , Allen decided he would not stand around and wait.

I completely agree that Greggo is an ass. He did put him down once with the DB shifting from the MLB to outside the DE. I was definitely not an Allen fan during the draft, but he has impressed me a lot so far. I just don't think over hyping it at this point is going to solve anything, there are still a lot of unknowns. The poster I replied to was trying to compare Foles and Allen when one played the starters and the other didn't. Those first 6 games are brutal.

[SlimShady'sSpaceForce]

Adjusted Completion %.  

 

Why do we see this?

 

because the stats are flawed, meaning they don’t take in all possibilities when advertising completion percentages.  

 

 

[artmalibu]

There is no replacement for old fashion scouting.... 

 

Game film trumps spreadsheets in this game...

[Just Joshin']

The team that finds a way to successfully blend analytics and traditional scouting will have an advantage.  There must be a way to combine the strengths of both methods in order to make better decisions.

[Matt_In_NH]

18 hours ago, HappyDays said:

 

The key word is "predicted." I work in finances and statistics is a hobby of mine. The biggest misconception people have about statistics is that they are making black and white predictions. Really they are assigning probabilities. So for example QBASE (the Football Outsiders analytics stat) gave Josh Allen a 62.7% chance of busting. That's a better than a 1 in 3 chance he makes it. A 1 in 3 chance is NOT that unlikely. If he succeeds that doesn't mean QBASE or whatever other analytics tool is complete bogus. If a model gives 10 QBs a 66% chance of busting and they ALL bust, that means the model is broken. The problem isn't analytics as a tool, the problem is how people interpret them.

 

Great post!   What we see on these forums is that a lot people see things as black and white about pretty much everything.

[Batman1876]

12 hours ago, SoTier said:

 

While Allen has grown a lot since the end of the college football seasons, there's no guarantee that he'll continue to improve.  At some point, all QBs hit "the wall" where they fail to improve significantly if at all.  Allen's impressive improvement might be that he's had more room to improve than most first round QB prospects, such as Mayfield.  Moreover, improvement because of mastering fundamentals comes easier than improvements in more complex skills like reading defenses or making the right decisions required because the pro game is much faster and more complex than the collegiate game.  It's not unusual to see a first year starter -- think Mark Sanchez, Colin Kaepernick, or EJ Manuel --  look pretty good in that first year starting and sometimes even early into his second season -- but then simply stop improving any more.  The problem with that is that what's "acceptable" or "good" for a first year starter isn't "acceptable" or "good" for a guy who's been starting for 4 or 5 seasons.

 

All QBs "hit a wall" at some point.  The poorer ones hit it sooner.  The great ones take much longer to do so, but with many QBs their performances aren't all that clear-cut.   They seem to be much more dependent upon their teammates or coaching or various tangibles/intangibles.  Think guys like Jay Cutler, Ryan Tannehill, Blake Bortles or Jameis Winston.   Analytics can't predict when a QB is going to stop improving.

Of course an injury could happen, Luck is a prime example of how injury can derail all development.  RG3 shows another failing that gets QB's, work ethic.  RG3 was concerned with the "RG3" brand more than with being a great player.  His natural ability made him 2nd overall, his inability to work on a team and work hard to improve made him a bust. EJ Sanchez and Kaep were the same player first season to now. I watched EJ play in preseason and he looked exactly the same as he did with us in 2013, same faults.  They were also the same faults identified pre-draft. We know that Allen is a team player, his team is already behind him, and we know he can work hard to get ahead.  Allen is the quintessential guy for demonstrating what you get with "old fashioned scouting" Arm strength, athleticism, team player, ability to learn & grow and desire to learn &grow. None of those traits show up on paper.

[Big Turk]

The problem with all the QBs is there are so many nuances that arr different from college to the NFL there is really no way of knowing how well any individual QB will pick those up and/or fix any flaws through better coaching.

 

Some are much better at fixing these flaws tham others and it appears Allen is a maestro when it comes to this

[Hapless Bills Fan]

19 hours ago, Batman1876 said:

I like analytics and was against drafting Josh Allen.  The analytics were clear his poor statistical performance against week competition predicted a poor success rate.  His inaccuracy was sure to doom him at the next level.  I bought into it and wanted analytics darling Baker Mayfield.

 

But the problem with these types of statistical analysis is that it is hard to translate them to individual case by case scenarios. When you get to the individual player there are various idiosyncrasies of their story experience and their career that may produce a result that is statistically improbable on the whole but entirely likely on the individual case.  For instance if a player put together a great college carrier but they struggle with addiction, injury or laziness they individually are far more likely to fail. The prime example of this would be Ryan Leaf, analytics would say he was as likely to succeed as Joe Flacco was but Leaf's individual laziness, bad attitude and addiction issues meant that he was always going to fail.

 

In his limited preseason game time Allen is demonstrating the opposite of the example above, instead of showing how analytics overvalue some players he is showing how analytics can undervalue one.

 

The strength of analytics is a more-or-less equivalent data set.  Analytics are valuable for assessing how different play calls and player performance stats impact winning at different points in a game.  And of course, there's the bloomin' onion of fantasy football, which has catalyzed a proliferation

The reason analytics aren't the Holy Grail of college player drafting and especially college QB performance in the NFL, is because college football is not a more-or-less equivalent data set.  Similar statistics can mean very different things depending upon the offensive scheme, the quality of the team around the QB, the quality of the defenses they face, and so forth.  So when looking at 3 QB from an air-raid offense or 3 QB playing at a similar level of competition, it's helpful to look at analytics to sort them out.

 

Where analytics are weaker and less predictive, is when the player under consideration falls outside the parameters of the primary data set.  That may be true of Josh Allen, coming from a smaller conference and a pro-style offense and having received apparently minimal coaching on his path to the NFL.  

The other part, is, as you say, analytics can't measure heart and work ethic, or other personal factors that may influence a player's success.

 

[billspro]

20 hours ago, Batman1876 said:

I like analytics and was against drafting Josh Allen.  The analytics were clear his poor statistical performance against week competition predicted a poor success rate.  His inaccuracy was sure to doom him at the next level.  I bought into it and wanted analytics darling Baker Mayfield.

 

But the problem with these types of statistical analysis is that it is hard to translate them to individual case by case scenarios. When you get to the individual player there are various idiosyncrasies of their story experience and their career that may produce a result that is statistically improbable on the whole but entirely likely on the individual case.  For instance if a player put together a great college carrier but they struggle with addiction, injury or laziness they individually are far more likely to fail. The prime example of this would be Ryan Leaf, analytics would say he was as likely to succeed as Joe Flacco was but Leaf's individual laziness, bad attitude and addiction issues meant that he was always going to fail.

 

In his limited preseason game time Allen is demonstrating the opposite of the example above, instead of showing how analytics overvalue some players he is showing how analytics can undervalue one. The issue at hand is coaching, Josh was under-coached throughout his time in football. He attended few QB camps went to a Small time High school, then a small time college. The overwhelming number of QB prospects, especially top prospects went to those QB camps, focused on football in highschool and went to a program far bigger than Wyoming. In other words the concept of having already received top flight coaching is inadvertently baked into the analytics. A player who has not received good coaching naturally has more to learn than a [player who has. 

 

Josh Allen has made huge strides (pun intended) when it comes to his footwork, which has improved his accuracy which was the prime concern about his game. Despite the fact that we are in preseason it is easy to see how his footwork is far better than before the off season. If this improvement continues Josh will be the biggest example yet of how the individual circumstances of a player can be missed by  statistical analysis.

 

So true, he had the hardest tape to analyze too because he had no talent around him. The talent he had to work with was much weaker than his competition. He was one of the highest pressured QBs in the NCAA because of his oline and always had to throw into tight windows because of his WRs. Allen had the most wow throws in tape, great athletic ability, and he did show a good grasp of a pro style offence. Working with Palmer was amazing for him because it was the first time he had real coaching. You could see he improved a ton at the senior bowl. I remember watching the game thinking maybe everyone is wrong about Allen (myself included, I was very worried about the completion percentage). 

 

All I can say now is thank you Wyoming for hiding such an elite QB prospect. If he had any oline at all he would have went first overall. So far in the preseason he has shown elite traits (pocket awareness, arm strength, accuracy, command of the huddle, quick release, and quick decision making).

 

Right now it is looking like Allen vs Darnold will be the future of the AFCE. I’m glad we have Allen because I think he is going to end up being better.

[JinxedBill1]

Still early, way too early but the signs are there.  Looking forward to Sunday

[cba fan]

19 hours ago, Buffalo Timmy said:

As posted previously in post there is a huge misconceptions about analytics- all the numbers say is that this guy has the best chance of success based on his numbers. It is not stating who is better because numbers only cover so much of a person. Also reading the numbers is not as easy as many think or the baseball A's would never be in the playoff hunt.

You are compleltly correct.

 

IMO. I have always believed baseball analytics are much more accurate as baseball is such a one on one sport with much better data to dissect and less of a team sport as most of the action involves one player doing something without much teamwork involved.

 

Football has such a poetry of motion on every play it has way too many variables to be as accurate with any analytics as baseball can be. The A's I think have proven this for years.

17 hours ago, The Wiz said:

The part about this that is funny is how many people are changing their minds about Allen but his stats are basically the same.  Yes there are throw aways and drops but they aren't pointing at the stats anymore and going by the eye test instead and liking what they see.

I like what I see. My eyes have never lied to me.

Edited August 25, 2018 by cba fan

[Orlando Buffalo]

10 minutes ago, cba fan said:

You are compleltly correct.

 

IMO. I have always believed baseball analytics are much more accurate as baseball is such a one on one sport with much better data to dissect and less of a team sport as most of the action involves one player doing something without much teamwork involved.

 

Football has such a poetry of motion on every play it has way too many variables to be as accurate with any analytics as baseball can be. The A's I think have proven this for years.

I like what I see. My eyes have never lied to me.

I think this is the first time i have agreed fully with someone on here- i am now certain we must be wrong.

[ddaryl]

4 hours ago, artmalibu said:

There is no replacement for old fashion scouting.... 

 

Game film trumps spreadsheets in this game...

 

Its a combination.. Neither trumps the other IMO... together they form a larger picture. The combines is mostly pure analytics, but the combine has turned lower rated prospects into 1st rd busts but has also found some hidden gems just the same

53 minutes ago, cba fan said:

.

 

IMO. I have always believed baseball analytics are much more accurate as baseball is such a one on one sport with much better data to dissect and less of a team sport as most of the action involves one player doing something without much teamwork involved.

 

Football has such a poetry of motion on every play it has way too many variables to be as accurate with any analytics as baseball can be. The A's I think have proven this for years.

I like what I see. My eyes have never lied to me.

 

 

I would agree the baseball analytics are definitely cleaner for interpretation based upon exactly what you are saying.. Its a game mostly of individual perfromace where the data can be dissected in a more cut and dry way.

 

but its still very useful in football its just has to be scrutinized and combined with scouting and game film much more

Edited August 25, 2018 by ddaryl

[Sky Diver]

QBASE  gave Allen a 62% chance of being a bust.

 

I would do a Pareto of the negative attributes that led to the assessment. I’d watch film keying in on the most significant negative attributes and look for root causes. For those unrelated to Allen (not running the right route, dropping the ball, etc), I would assess how fixable the problems were.

 

Whatever issues Allen had, the Bills certainly thought that they were correctable since he was a #7 pick.

 

I assume that they did the type of analysis I described. Clearly Buffalo didn’t rely on analytics only.

Edited August 25, 2018 by Sky Diver

[starrymessenger]

If by analytics we mean stats then you have to disregard them in evaluating Alan. His college stat sheet is bad, except maybe for rushing totals (still much less than Cam's). So it's not just completion percentage. It's the same for every statistical measure. Stats wise he's maybe a fourth round pick. But Allen is legit as an early first rounder. So what gives?

What gives is that for all the bad tape (and bad stats) there is as much tape of Allen executing perfectly and when he executes perfectly he is able to get the most out of his physical skill set, which makes him stand out like an adult amongst children. Other athletically gifted prospects with big arms were never as impressive as Allen when he's on his game and doing things the right way. No EJ, Boller, Locker, Losman etc...have tape as impressive as Allen has. And there is enuf of it to reliably conclude that his demonstrated ability is not an accident. After even only a few preseason appearances, we have now all seen Allen make a couple of plays that we know only a handful of the best NFL starters are capable of. Add to that his intelligence, work ethic, coachability and apparent personality/leadership skills and one can easily understand why he would have been completely credible as the 1st pick overall. For sure he is an anomaly but folks can be forgiven for thinking that that he may just be a generational talent.

As Jahlen Ramsey has correctly pointed out, Allen's performances against better opposition were worse than against weaker teams. Not making excuses for him but that's sort of what you might expect. And I'm inclined to think that this has something to do with the increased gap between the relative quality of the teams. Allen is clearly a guy who takes it all upon himself and I can see him trying to make up for the differences in skill levels for all offensive position groups. In the pros he will be going against defenders and DCs the likes of which he has never seen, but he will also be playing with much more accomplished teammates (and it looks like he's got a good OC). Even if the Bills may occasionally be overmatched the gap will not be as pronounced as that between his 2017 WYO team and Iowa and he will always benefit from having much more talented options than he had in college. Logically (or analytically) a guy who looked bad against Iowa should look that much worse against a good NFL defence. But I don't think that will happen. I think he will look better, possibly a lot better.

[Fadingpain]

21 hours ago, musichunch said:

 

You must hate that we made the playoffs last year. I've been noticing your posts around this board along with a few other active accounts with a very similar tone. It would do the fanbase a lot of good for moderators to just ban you. I do not believe you are genuine in your negativity, rather you want to bring people down and incite emotion. 

Huh?  I loved that we made the playoffs last year.

 

I do hate mindless kool aid drinkers like you though who only participate here as part of a fake feel good exercise.  I see no value in that whatsoever. 

 

I prefer objective, rational conversation regarding all things Bills.

 

I was not a fan of the Allen pick and I think it was a big mistake.  I would have taken Josh Rosen in the same situation.

 

He's here now, and Rosen is not, so pretty soon we will all be rolling down the tracks on the Allen Express.  

[starrymessenger]

5 minutes ago, Fadingpain said:

Huh?  I loved that we made the playoffs last year.

 

I do hate mindless kool aid drinkers like you though who only participate here as part of a fake feel good exercise.  I see no value in that whatsoever. 

 

I prefer objective, rational conversation regarding all things Bills.

 

I was not a fan of the Allen pick and I think it was a big mistake.  I would have taken Josh Rosen in the same situation.

 

He's here now, and Rosen is not, so pretty soon we will all be rolling down the tracks on the Allen Express.  

 

I like Rosen too. A lot. We talk about Allen's high ceiling. I think Rosen will get better too. I think the 5 guys taken in the first all could transition as solid starters. I think it was a very good class.

IMO when he's on his game Allen is doing more and doing better than any of the other really good prospects.

[billsfan89]

I feel like scouts saw a player that had literally everything you would want out of a QB prospect physically and mentally. While the analytics and game tape showed a QB that missed some easy throws and had more bad throws than most top of the line QB prospects have coming out of college. I think in the end scouts saw the why of Josh Allen's inaccuracy issues and thought he could iron out the 2 issues that were causing him to be inaccurate (footwork and decision making) while analytics just saw a pattern of inaccurate college QB's not translating well into the NFL. 

 

Josh Allen won't disprove analytics if Allen blossoms into a top NFL QB. Allen will prove the limitations of analytics (you have to understand what is correctable by coaching) but analytics will still be a useful tool for NFL scouting departments. 

 

I think the scouts saw a QB prospect who had literally everything psychically you would want from a QB prospect. Cannon arm, great throwing motion, prototypical size, big hands, good intangibles, good work ethic, and smart. The biggest issues with Allen were his footwork and his decision making, both were the major source of his inaccuracy. Although footwork isn't easily correctable it is correctable and decision making can be improved with good coaching and experience. 

[Green Lightning]

On 8/24/2018 at 4:04 PM, musichunch said:

 

Until we lose again, then they come out of the woodwork. They feed on doubt and negativity. Banning is the best move IMO. 

Oh gosh, someone said something I don't like. Ban him!  Who needs free speech anyway? Damn constitution...