Another Statistical Analysis of the Broncos with Tebow

[Joel]

I found it at NFL.com just now (as long as there are no games in progress it USUALLY doesn't crash my browser) but freely admit I found it interesting mainly because it informed me someone is actively tracking Win Probability now. I never expected that because it's such a situational stat, but 25 years has made it far easier to track an entire seasons (or multiple seasons) worth of plays via computer like Carroll, Palmer and Thorne did back then (it's still more fun to use the redhead though. ;)) Between use of Win Probability and the reference to Football Outsiders Defence-Adjusted Value Over Average you could almost call this article "The Hidden Game of Tebow." Be warned, Broncos fans, the news is not good:

http://harvardsportsanalysis.wordpress.com/2011/11/26/a-statistical-analysis-of-the-miracles-of-tim-tebow/

[Thnikkaman]

I find the comments more interesting than the article itself. And regardless if what the stats say, I'm still going to watch every game.

[Northman]

Pretty interesting article, made some good points. May have to look at that more later.

[camdisco24]

Interesting article. They were doing pretty well until they compared Vince Young to Tebow. IMO, the two couldn't be any different when it comes to leadership and work ethic. While they do have similar play, Young was never able to improve his game and motivate those around him. I think with a real offseason, Tebow can really improve his throwing game.

It's a wait and see right now. At the end of the day though, I'm going to enjoy every one of these wins!!

[Joel]

I find the comments more interesting than the article itself. And regardless if what the stats say, I'm still going to watch every game.
Each to his own, I suppose; it seemed to me most of the comments consisted of people arguing either for or against the stats and/or ignoring them completely on the grounds Tebow can do no wrong. Debating the latter position is as pointless as debating any religions adherents.

As to the stats, two of them were developed by fine statisticians using every play of the '84-'86 seasons (skipping the '87 stats for obvious reasons) in The Hidden Game of Football, which virtually defined NFL statistical analysis and continues to do so; anyone who hasn't read it should stop reading this and do so NOW. Go on, I'll wait....

The article compiles those stats the same exhaustive way and uses them in conjunction with a third even older and simpler stat duly cited and amply vindicated by that same book (specifically, the one in this 1970 article from "Operations Research:" http://www.jstor.org/pss/169286; if it helps, THGoF notes study co-author Virgil Carter played QB at BYU back when college quarterbacks actually had to go to class.)

Suffice it to say I consider the stats themselves above reproach, though the articles use of them may not be. I think it generally is though, because the article essentially says an extended streak is still a streak. Tebows miraculous wins are impressive and memorable solely because he allowed the team to get far enough behind with so little time left that their probability of victory, based on the results of every other game where a team had the ball with the same down, distance, score and time remaining, was nearly 0. That may seem esoteric and abstract at first glance, but it's really just precisely quantifying things everyone already knew vaguely.

When we were down by 15 against Miami with just over 5:00 left we all knew chances were slim Tebow or anyone could lead us back; when we trailed the Jets by 3 inside our 10 with the same time left we knew the odds weren't much better: Win Probability simply compares the outcomes of every other game when a team was in that situation and tells us the precise probability of victory. At this point, we're skewing our own numbers. :tongue:

Obviously I'll keep watching the games and rooting for our team, but the trouble with creating high expectations is that you have to meet them. Denver fans are talking playoffs again, but that means finishing even with the Bengals (who are one game ahead of us) and Jets, and ahead of the Titans (whose remaining schedule is garbage.) That can only happen if Tebow proves himself more than an admittedly large statistical anomaly.

[Joel]

Interesting article. They were doing pretty well until they compared Vince Young to Tebow. IMO, the two couldn't be any different when it comes to leadership and work ethic. While they do have similar play, Young was never able to improve his game and motivate those around him. I think with a real offseason, Tebow can really improve his throwing game.

It's a wait and see right now. At the end of the day though, I'm going to enjoy every one of these wins!!
Agreed on all points; I continue to feel Tebow has the physical and mental ability, as well as the diligence, to become as dangerous with his arm as he is with his feet. I think the articles chief relevance is for the remainder of this season, because he won't learn how to QUICKLY read defences and go through progressions in the next month or two. I am hoping he doesn't plateau or implode, but to get into the post season he must surpass his already remarkable improvement as a passer, not next year, but now.

[vhatever]

Then, yielding to fan pressure, they inserted Tim Tebow at quarterback and went 4-1 over the last 5 games in rather spectacular fashion, producing a flurry of ridiculous media coverage, spurring Denver fans to start wearing Jesus jerseys and creating a whole narrative around Tim Tebow’s Miracles.


It sounds like a really informative analysis based on facts and free of bias.

[NightTrainLayne]

The article doesn't go into too much detail on how these stats are derived, but they seem to leave out turnovers.

Turnovers, or the lack thereof, are by far the primary reason for our winning streak.

[MOtorboat]

This is, essentially, more to substantiate that this team's winning is more of an illusion than a trend, much less a long term trend.

Kind of like the defense during the 6-0 streak two years ago.

[SM19]

The article doesn't go into too much detail on how these stats are derived, but they seem to leave out turnovers.

Turnovers, or the lack thereof, are by far the primary reason for our winning streak.

Expected points added and win probability added certainly include turnovers. A fumble or interception is a play that changes how many points a team can expect to score, and how likely they are to win, just like any other.

AdvancedNFLStats.com explains WPA here (http://www.advancednflstats.com/2010/01/win-probability-added-wpa-explained.html) and EPA here (http://www.advancednflstats.com/2010/01/expected-points-ep-and-expected-points.html).


Interesting article. They were doing pretty well until they compared Vince Young to Tebow. IMO, the two couldn't be any different when it comes to leadership and work ethic. While they do have similar play, Young was never able to improve his game and motivate those around him. I think with a real offseason, Tebow can really improve his throwing game.

It's a wait and see right now. At the end of the day though, I'm going to enjoy every one of these wins!!

Young's not the only quarterback they compared Tebow to, though. There were nine others that fit a profile similar to Tebow's and Young's, including Tom Brady and Aaron Rodgers. The point isn't that Tebow's another Vince Young, or that he's another Aaron Rodgers, or that Aaron Rodgers is another Vince Young, it's that all of these quarterbacks who started out with a large gap between their EPA and their WPA eventually saw that gap narrow.

[Joel]

Then, yielding to fan pressure, they inserted Tim Tebow at quarterback and went 4-1 over the last 5 games in rather spectacular fashion, producing a flurry of ridiculous media coverage, spurring Denver fans to start wearing Jesus jerseys and creating a whole narrative around Tim Tebow’s Miracles.

It sounds like a really informative analysis based on facts and free of bias.
The beautiful thing about stats is that, in themselves, they admit no bias (though they are subject to massaging by those who do.)

The article doesn't go into too much detail on how these stats are derived, but they seem to leave out turnovers.

Turnovers, or the lack thereof, are by far the primary reason for our winning streak.
The detail is lacking, I believe, because they borrowed stats derived by others; they didn't compile or crunch the numbers, let alone do the research to discover an important revealing stat.

Make no mistake, however, they are important revealing stats, and not especially new or controversial ones. Win Probability is fairly straight forward and simply quantifies precisely things everyone vaguely knew already. It builds on Carter and Machols 40 year old statistical work that basically boils down to expected points with 1st and 10 asymptotically approaching -2 at our goal line and 7 at theirs (AdvancedNFLStats uses the value of 6 from THGoF, which I consider a big tip of their hand that they're going off that book rather than the original "Operations Research" article.")

It must be noted that "expected points" does not mean the team with the ball is likely to score (or be scored on, if near their goal line) on the next play or even the current drive: It is just the expected value of the next score--WHENEVER it happens--to that team (obviously EoH resets everything.) Think of it as a quantified value for field position, because that's basically what it is; it's just more obvious that 1st and G on their goal line is worth nearly 6 points than it is that 1st and 10 at midfield is worth 2 points. In terms of probability, both are equally true; if 1st and 10 at your 1 inch line is worth -2 points and 1st and G at THEIR 1 inch line is worth 6 points, obviously 1st and 10 at midield is worth midway between them. Nothing too controversial about that, right?

It does lead to a few surprising conclusions, like "going for it on 4th and 5 at midfield is statistically the right call" and "turnovers are equally bad wherever they occur." Since you mentioned them I'll focus on that example:

A turnover at their goal line improves their situation no more than a failed 4th and G, which we already know isn't horrible because it pins them deep. Your expected points drop from 6 to 2, a net of 4; they go from -6 to -2, an equal gain. Flip it around to YOUR goal line stand and the same thing happens; your -6 is now -2 and their 6 is now 2, still a net change of 4. In other words, a turnover on EITHER goal line nets you -4 and them 4 expected points, because they were already low when you were pinned deep and remain high when they are. Turns out that net 4 point swing is present ANYWHERE you have a turnover: With 1st and 10 at midfield you have 2 expected points, but turn it over and THEY do, still a 4 point swing.

An interesting side note Carroll, Palmer and Thorne observe in their book is that, a team pinned deep should NOT get conservative unless they "can run out the clock before having to punt. Otherwise they're just tap dancing until the other team scores. Remember, a turnover causes the same 4 point swing in point potential no matter where it happens, so why condemn your offense to failure because it might make a mistake? Call the play that will get you out of the hole!"

That illustrates the difference between expected points and expecting to score on a given drive: The -2 points at your goal line does not guarantee a safety, but a "safe" three and out then punting gives them 1st and 10 at midfield. You've done an awful lot of work to end up giving them the same 2 expected points they had when you first got the ball. As most of us already knew, a single first down is very valuable for you there is very valuable, even if you still end up punting. HOW valuable? About 1 expected point. ;) Two first downs is worth about another expected point, meaning 1st and 10 on your 20 gives both teams 0 expected points, and a touchback makes each team equally likely to score. The surprising thing here is not that the statistic actually seems valid but that THE LEAGUE MAY ACTUALLY HAVE GOTTEN SOMETHING RIGHT! :eek: I'm as shocked as anyone.

Now, if you want to know why the Broncos should've gone for the kill up by 10 with 1st and 10 on the Redskins 13 in SB XXII, get the book (but I'll give you a hint: That field position was worth 5 expected points; three "safe" plays protecting the eventual FG only scored 3.)

Win Probability is a bit more complex and about 15 years newer (at least in football) but every bit as rigorously grounded as expected points, from which it is derived. You can read Brian Burkes description of it at AdvancedNFLStat here:

http://www.advancednflstats.com/2008/08/win-probability.html

and/or try to follow my take on it, but they should be essentially the same in terms of mechanics. All WP does is compile the score, time, field position, down and distance for each interval of each game in a given sample, and spit out the probability a team with the ball wins in that situation:

Times a team with the ball, that score, remaining time, field position, down and distance WON
Times a team had the ball with that score, remaining time, field position, down and distance

When Carroll, Palmer and Thorne introduced WP to the NFL in The Hidden Game of Football 25 years ago, online NFL stats were virtually unavailable, and the computing power to access them little more so. Thus they only used three minute intervals from '84-'86 (they threw out the '87 numbers, which have "statistical anomaly" written all over them.) Today virtually ALL NFL stats are available online, and AdvancedNFLStats has compiled the score, remaining time, field position, down and distance for what appear to be one minute intervals over five (rather than three) seasons worth of games (including post season.) Statistical reliability improves with sample size, and that's a MONUMENTAL sample size (the creator of a WP model at ProTrade posted to Burkes description above, stating THEY used around 2,000,000 samples in their model.)

Win Probability has limitations; for one thing, it is still JUST a probability, not a certainty. It is also useless for predicting a final BEFORE the kickoff: Until someone has the ball at some field position with some down and distance to go and some score with some amount of time left you're dividing by 0. It is one of those rare stats MOST useful to coaches and LEAST useful to fans (and bookies.)

Because of the nature of stats, it is most reliable for common situations (where there is a large sample size) and least reliable for uncommon ones (where there are few samples.) Fortunately, that is largely compensated for by the fact common situations like 1st and 10 at midfield up by 3 with 10:00 left in the 2nd quarter (which AdvancedNFLStats says has a surprisingly large WP of 67%) are the ones where we need help. We know a team at midfield down by 23 with 10:00 left in the 4th loses 99 times out of 100, without calculating anything (though that team actually has a 2% chance of winning if they have the whole 4th quarter; thank you, Tony Romo. :rolleyes:)

Unfortunately, just as WP becomes more reliable as scores go up and remaining time goes down, it is less definitive when games are tight and/or have a lot of time left: Knowing that winning the coin toss gives us a 50% (or 55%) chance of victory doesn't tell us anything we didn't already know. Knowing a team down by a FG with 1st and 10 at its own 20 and 5:00 left in the 1st quarter has a 40% chance of winning is only slightly more illuminating. The good news is that AdvancedNFLStats offers a handy WP calculator that lets us punch in score, remaining time, field position, down and distance and get WP, expected points (which we shouldn't need to calculate; it's just distance from our 20/12.5) and probability of a TD and FG (it doesn't say whether that's for the play or the drive, but it appears to be the latter.)

http://wp.advancednflstats.com/winprobcalc1.php

Having said all THAT, according to AdvancedNFLStats methodology page:
http://www.advancednflstats.com/2007/09/nfl-win-prediction-methodology.html

they do go a bit further when specific (not generic) teams are involved, modifying basic WP based various efficiency stats specific to the teams in question. That gets a lot more complicated because you're forced to do things like rate the teams performance against League averages and power rank them. I dislike power ranking. Intensely, despite recognizing its undeniable worth. Unfortunately, power ranking is a recursive process, because calculating the first set of power rankings changes the INITIAL power rankings on which they were based, requiring the whole process to be repeated indefinitely, until they stop changing. I was never willing to do that for 28 teams, let alone 32, but anyone with that kind of time and tenacity is welcome, with my blessing.

For generic WP it does not matter much, and unless your John Fox I wouldn't sweat it. Denvers WP up by 4 with 2nd and 7 at Minnesotas 32 and 13:47 left in the 4th isn't much different than any other teams would be against any other team in that situation, and even if you take the time to figure out all the power rankings for all 32 teams so you can factor it in (and you've got, what, 4 hours?) you'll just have to redo it all again next week and factor in this weeks scores. Like I said, you're welcome to it, but I've probably spent way too much time on the subject already. :tongue:

[Chef Zambini]

Interesting article. They were doing pretty well until they compared Vince Young to Tebow. IMO, the two couldn't be any different when it comes to leadership and work ethic. While they do have similar play, Young was never able to improve his game and motivate those around him. I think with a real offseason, Tebow can really improve his throwing game.

It's a wait and see right now. At the end of the day though, I'm going to enjoy every one of these wins!!tebow should have improved his throwing game LAST OFF SEASON, lock out or not!
he should have used some of his 6 large up front to HIRE a QB coach ! instead he did commercials and circumscisions.
he came to camp unable to take snaps from under center, and every side of a barn safe from harm.
a back shoulder throw remains out of the question.
and many would compare him to a 2 dollar whore.
"ugly but he gets the job done"
( put down those pitch-forks and torches)
as apasser, TT still needs ALOT of work.
But as a leader and a teammate, he is the essential flashpoint to our teams success.
his will to win makes us winners.
its 'storybook", not playbook.
and thats why statisticians remain so busy trying to justify the tebow phenomenon.

[HORSEPOWER 56]

Stats only take you so far and we all know that having a Kyle Orton/McDaniels led offense looked great on paper, but was a paper tiger.

The only stat that matters is wins. Are we winning? Yes. Guys kept saying that our offensive scheme wouldn't work in the NFL. They were wrong. Now they are saying it's not sustainable. We'll see. The same things were said when all the teams ran power-run football and the passing game started gaining in popularity - it'll get figured out, it's not sustainable, etc. The same was said about the 4-3 defense and the 3-4 defense. It's more about who you have running it than what the principles are.

Why, if schemes are so easily figured out, can nobody stop Aaron Rodger's "scheme" this year? How come they haven't really been able to blueprint how to stop Manning and Brady consistently - even after a decade of film to watch? Because there is no blueprint for stopping any scheme. At the end of the day, your players just have to play better than theirs and their coaches have to be smart enough to put their players in a position to succeed. You can game plan against certain players, but if schemes are executed correctly by those players, all of them are tough to stop.

Even though newer rules favor the passing game, it doesn't mean you can't win with a run-first offense and a stingy defense. If anything, our team is built perfectly to to counter those pass-happy teams. We run the ball to keep their pass happy offenses off the field and have a pretty darned good pass rush to pressure the opposing QB. So far, it's working. Just because it's not the norm doesn't mean it won't work or continue to work.

[Cugel]

Statistical reliability improves with sample size,

And that is exactly why ANY "statistical analysis" of Tim Tebow is utter garbage! It makes about as much sense as an analysis of Kyle Orton -- based on the first 6 games of 2009. :coffee:

If Tebow plays a significant number of games in the next couple of seasons we can THEN look at his passing and do a "statistical analysis."

Until then we get to listen to Tebow-heads who wouldn't know the basic tenets of statistics if they came up and bit them on the ass. And equally ignorant Tebow detractors using STATS. The guy has thrown fewer than 100 passes. Let's not talk nonsense about his QB rating or passing percentage.

The most we can learn is simply to watch his throws. From that most experts conclude he's a lousy passer. Not a controversial observation really, except that the Teboners can't accept it without much gnashing of teeth.

Does that mean he can NEVER learn to throw the ball accurately? No. He'll have to change his mechanics but he might be able to do that in the off-season.

Is it inevitable that he WILL learn to throw accurately and become a great NFL QB? No. :ranger:

[Bullgator]

Every week there is another article that tries to dismiss the success Tim and the Broncos are having, causing doubt that the success is real. And every week Tim goes out there and wins the game in the 4rth quarter.

People all over the sports world speculate how long this success will last. What difference has that made? They are irrelevant. Their words are empty.

And clutch win after clutch win there are still some here who need more proof. Fight it till the end will we?

What will it take before you believe? A superbowl win? Isn't that a tad late to START investing some faith in your team?

[Joel]

And that is exactly why ANY "statistical analysis" of Tim Tebow is utter garbage! It makes about as much sense as an analysis of Kyle Orton -- based on the first 6 games of 2009. :coffee:

If Tebow plays a significant number of games in the next couple of seasons we can THEN look at his passing and do a "statistical analysis."

Until then we get to listen to Tebow-heads who wouldn't know the basic tenets of statistics if they came up and bit them on the ass. And equally ignorant Tebow detractors using STATS. The guy has thrown fewer than 100 passes. Let's not talk nonsense about his QB rating or passing percentage.

The most we can learn is simply to watch his throws. From that most experts conclude he's a lousy passer. Not a controversial observation really, except that the Teboners can't accept it without much gnashing of teeth.

Does that mean he can NEVER learn to throw the ball accurately? No. He'll have to change his mechanics but he might be able to do that in the off-season.

Is it inevitable that he WILL learn to throw accurately and become a great NFL QB? No. :ranger:
You make a very valid point, and I have argued the same in the past. Tebow probably won't even have enough passes this season (his second) to qualify for the League passing title, so we can't make sweeping predictions about his future based on the current tiny sample size. However, the ONLY predictions we can make about his future are those based on the data set we have, however limited, and it is growing. The more it grows the more reliable those predictions become, but they are certainly still tentative; that is true even for prototypical pocket passers until they've started for two (and preferably three) years.

What I get from the article is that the stats say he can't keep winning doing what he's done. Most people who watch as many games on Sunday as Saturday would agree that's a reasonable conclusion. We must be careful to avoid circular truisms; we can't simply say every win means he did something different and every loss means he didn't (and the nice thing about stats is that no sample size will support that,) but the basic premise is sound. It's not about pro/anti-Tebow or anything else except the fact one dimensional offenses don't work in the NFL: If you can't run they'll blitz you ever down and/or double cover your good receivers; if you can't pass they'll load the box and tackle everyone who MIGHT have the ball. Fans of the option love it for its "flexibility" and "versatility" but that is an illusion: Whether you run the halfback or the quarterback you're still RUNNING, and that doesn't fool NFL defences.

Tebow and McCoy mixed it up some today, and had an oustanding passing game with a near perfect passer rating. I don't want to get into one of those correlation/causation debates, but Tebows running game was pretty flat on the few times he went to it. I may not be able to definitely say going to pass won the game, but it sure didn't hurt, both in terms of yards and points and in terms of opening things up for McGahee by forcing the Vikings to defend the pass.

Every week there is another article that tries to dismiss the success Tim and the Broncos are having, causing doubt that the success is real. And every week Tim goes out there and wins the game in the 4rth quarter.

People all over the sports world speculate how long this success will last. What difference has that made? They are irrelevant. Their words are empty.

And clutch win after clutch win there are still some here who need more proof. Fight it till the end will we?

What will it take before you believe? A superbowl win? Isn't that a tad late to START investing some faith in your team?
No, just our QB. Welcome to Denver. ;)

[silkamilkamonico]

No, just our QB. Welcome to Denver. ;)

Crazy to think of just how ignorant people (who don't know anything about the Bronco organization) are about Denver and the QB situation.


The general true fanbase is going to doubt it until there is another SuperBowl. That's what being spoiled by the G.O.A.T#7 can do to a fan.

[wayninja]

Tebow and McCoy mixed it up some today, and had an oustanding passing game with a near perfect passer rating. I don't want to get into one of those correlation/causation debates, but Tebows running game was pretty flat on the few times he went to it. I may not be able to definitely say going to pass won the game, but it sure didn't hurt, both in terms of yards and points and in terms of opening things up for McGahee by forcing the Vikings to defend the pass.

This is spot on. Unless you have a ridiculously good passer and equally good guys on the receiving end, the best thing you can do to help even the odds is deception. If we can get teams to treat Tebow like a true passer, we have already won the deception game. That simply isn't going to happen though if teams KNOW we are going to pound the ball 75% of the time. The playbook needs to get opened earlier. I don't care if we waste the few first couple of drives trying to lull/fool them, but McCoy is carrying it too far. We need to start mixing up earlier and more often if we are really going to enjoy success. Tebow and DT proved today that it can be done, just a matter of getting better at it against better secondaries. If we can open this up earlier and have success, we really help out our running game with the occasional option and McGahee making us a real dual threat.

[Joel]

Crazy to think of just how ignorant people (who don't know anything about the Bronco organization) are about Denver and the QB situation.

The general true fanbase is going to doubt it until there is another SuperBowl. That's what being spoiled by the G.O.A.T#7 can do to a fan.
If they stick around they'll get it, eventually; if they don't, it won't matter. Elway set a perhaps impossibly high standard (how many QBs have started in 5 Super Bowls? One) but, reasonable or not, it's a fact of life for Broncos fans, and our QBs. Ask Danny White how much fun it was taking over from Staubach. When Steve Young won his (only) Super Bowl his sigh of relief was audible throughout the country. It's the nature of the beast.

The trouble with creating expectations is that you have to keep meeting them. Perhaps losing the SEC Championship rematch to Alabama was good for Tebow in that respect, reminded him of both the importance of humility and hard work as well as the frustration of losing big games. I hope something taught him those lessons, because at 6-1 as a starter on a 7-5 team dangerously close to becoming a playoff favorite, he has already added to the high expectations that come with being Denvers QB. Sooner or later he will fail to meet them one way or another, and face the consequences, so I hope he's prepared to weather the storm, shrug it off and get back to winning. Otherwise he could end up less like Staubach and more like Romo, constantly as uncertain as everyone else whether he's the GoAT, or just the goat.

This is spot on. Unless you have a ridiculously good passer and equally good guys on the receiving end, the best thing you can do to help even the odds is deception. If we can get teams to treat Tebow like a true passer, we have already won the deception game. That simply isn't going to happen though if teams KNOW we are going to pound the ball 75% of the time. The playbook needs to get opened earlier. I don't care if we waste the few first couple of drives trying to lull/fool them, but McCoy is carrying it too far. We need to start mixing up earlier and more often if we are really going to enjoy success. Tebow and DT proved today that it can be done, just a matter of getting better at it against better secondaries. If we can open this up earlier and have success, we really help out our running game with the occasional option and McGahee making us a real dual threat.
It's odd to think of "running to establish the pass" morphing to "passing to establish the run," but as much as the League passes these days a strong case for that can be made. In many places it seems like teams have forgotten WHY, much less how, to play ball control offense.

Some bedrock rules persist, however, and one of them is that the offenses greatest advantage, even greater than the ability to extend drives or even games with defensive penalties, is having the initiative on every down. Offenses decide when plays begin, defences when they end. We talk about "defensive play calling," but except on the line most defences run formations and assignments, not plays. That's inevitable, because they must be SOLID; they cannot allow offenses to find/create even a single hole to plow through for a score.

Defences must PREVENT first downs and scores; if they fail that task and do everything else right they still fail. A good example is our goal line stand against the Jets a few weeks ago, where the D fought through all the blocks, blew up the play and even forced a fumble--and while they were jumping around slapping each others backs for that a guard fell on the ball in the end zone for a TD: FAIL.

The flip side, as that play also neatly illustrates, is that offenses need only ACHIEVE first downs and scores; accomplishing that task makes failing all the others irrelevant. The Holy Roller was such a badly executed play it prompted a new NFL rule, but it scored the game winning TD, so nothing else mattered (not even the Raiders subsequent admission they deliberately batted the ball forward in violation of an existing rule.)

What all of that means is an offense surrendering its precious initiative has already lost; once the D knows where they'll drill, preventing it is ridiculously simple. Offenses still routinely get killed by blitzes they know are coming, just because they can't stop them, but the days when Lombardi could diagram the Packer Sweep for opponents then successfully dare them to stop it are long gone. It's not even deception that's important so much as UNCERTAINTY: We can line up in a jumbo set, execute the running play for which it's designed and still succeed as long as the defence has good reason to believe the unlikely pass is nonetheless a real possibility. They can't cover everything at once, and we know where we're concentrating our forces for an overwhelming breakthrough, so we win. Tell them where we're going and it turns into a stalemate the defence wins after four downs.

Deception and trick plays only work a few times, until teams study game tape and learn your tendencies; uncertainty works all the time because they can't learn tendencies you don't HAVE. That's Offensive Coordinator 101, so I'm sure McCoy knows it, and as he becomes more familiar with and confident in Tebows repertoire I "expect" (read: require) we'll see more of it. String together a few more games like today and teams will scramble to shut down Tebows passing just as they have his running, at which point we start pounding the crap out of them running him and McGahee again. At a certain point we'll reach an equilibrium where they don't sell out on either of the equally likely possibilities, and every time we call a run OR pass we'll be stacking our whole team up against whatever portion of theirs is responsible for that area. Pretty obvious who wins that argument. ;)