There has been a great deal of research on the effects of variance in determining game strategies. Brian wrote a post a few years back about how underdogs need to use high variance strategies and Dean Oliver discussed the underlying statistical concepts in basketball terms. The main point is this: Below average teams must use riskier strategies to win. If a team's performance is below league average, then consistency is a flaw.
That brings us back to Orton v. Tebow. If the Broncos were a league average team or better, Orton would indeed give them a better chance to win. Yet, the fact that the Broncos are dismal should be more of a reason for Tebow to play. Picture Orton and Tebow as follows:
This is a rough sketch, portraying a Denver offense run by Orton (green) and Tebow (blue) versus the league average (red). Notice that Tebow's curve is wider, meaning he has a higher variability, and as a result, the area under the blue curve that intersects with the normal curve is greater than the comparable area under Orton's curve.
I tried looking at standard deviations in EPA/P from last year, which supports this notion (Tebow = 1.54 > 1.45 = Orton), but it is not convincing. More importantly, I don't feel comfortable using these numbers due to small sample sizes (Tebow had less than 100 passing plays). Yet, if we add in the running game, it is clear that Tebow presents a much wider range of opportunities for the Broncos. Tebow regularly has the potential to add numbers to the running game of which Orton could only dream.
The Broncos win this week over the Dolphins is a perfect parallel. Tebow played like pure garbage for three and a half quarters, much worse - I imagine - than Orton would have in similar situations. Then, Tebow prayed a little and showed off the upper portion of his blue curve, turning on Tebow Time and showing his high risk/high reward potential.
In any one game, Tebow gives the Broncos the best chance to win. That being said, every win is a loss in the Andrew Luck sweepstakes and a step toward perpetual mediocrity.
Keith Goldner is the creator of Drive-By Football, and Chief Analyst at numberFire.com - The leading fantasy sports analytics platform.
Bill Walsh and Vince Lombardi both said the single wing could win a lot of games in the NFL. The pros don't use anything like it for fear of having a zillion-dollar throwing arm on a (relatively) small and delicate QB get crushed on a running play, and few pro QBs can run like a running back anyhow. But Walsh later said he'd always regretted not putting in a single wing package for Young.
Well, Tebow is a 240-pound linebacker of a QB without a zillion-dollar throwing arm to damage. He can run like a real pro RB, hit the 10-yard pass decently, has a good chance of hitting a longer pass to a WR left open by a D covering extra running threats, and let us assume he has the brains, leadership and all the famous "intangibles" to run an O that operates within his abilities and plays to his strengths.
If we want a high-variance strategy, why not put in a single-wing package for this guy during the last half of the season and see what happens? Extra advantages: It'd be a surprise weapon ... they could practice and develop it constantly, each opponent would have to prepare for it on a "WTF is this?" basis within a week ... even if it doesn't work out, it'd give all the fans and commentators something new and fun to enjoy and yack about while Elway and Fox figure out the future.
What do the Broncos have to lose? I *want* to see the single wing in the pros again. And it's now or never. Bring back the single wing!
I think the "many people want Tebow to fail" is a MASSIVE Red Herring. I mean, I'm sure you can find some forum comments out there by Tebow haters. But this idea that there is a significant number of people who want him to fail because he's a devout Christian or something is, as far as I can tell, a complete load of crap.
Now, there are plenty of people who EXPECT Tebow to fail, and who say as much, but that's not the same thing. I suppose that there are people who would rather be right about their predictions, just because people like to be right, but that's not quite the same as out-and-out rooting against him.
For what it's worth, I don't think Tebow will ever be an above-average NFL quarterback. His mechanics are still terrible (and very slow), his accuracy is still erratic at best, and doesn't appear willing to go through basic progressions.
Given that it doesn't look like those things are going to change, embracing his style and putting him in a Wildcat/Single Wing system for the majority of his snaps would be pretty reasonable. It's not as though you can't throw out of that formation.
In that graph, you have Orton and Tebow with the same mean.
Jim-
I would love to see Tebow in the single wing and think it is the perfect example of a high-variance strategy. That being said, I think the speed of modern defenses could wreak havoc on the single wing. All else equal, like you said, what is there to lose? The problem, like Brian often points out, is that coaches and those who run teams try to hard not to lose rather than trying to win.
As for anonymous' comment about the graph, I believe I wrote that it was a "rough sketch". Don't have any reliable data to make accurate curves.
I understand it is a rough sketch, but a bit inconsistent with your previous comments in the same post that Orton > Tebow. If you give them the same mean, then your point is obvious, go with Tebow. But since you think Orton is better then it becomes a tougher decision, doesn't it? If Orton was better than league average, then it would be obvious that starting Orton is better, even with the lower variance on a bad overall team. There should be a break-even point where the variance difference is canceled off by the mean difference.
I agree with the basic point of this article, but I have to wonder about the run/pass thing and whether Orton might actually be the higher variance. Passing will always be much higher variance (and higher reward). Since Orton is a better passer, the Broncos are much more likely to use him to pass the ball than Tebow to run it. Using Tebow effectively basically means using him to run (or at least threaten to run). Running in general isn't that useful to teams that are likely to find themselves behind.
I actually sort of see Tebow's style as being very useful to a team that is already fairly likely to win without him. In truth, this might have been why University of Florida was such a perfect fit for him.
Anonymous -
If it pleases you: http://i.imgur.com/eZ2yh.png
Jeff -
Passing in general may have a higher variance, but between Tebow and Orton, Tebow will have a higher variance in production in both the passing game and running game (relative to themselves, rather than one another). As for Tebow's style being useful for a team that is fairly likely to win without him, I disagree. Due to his high volatility and the fact that his strength comes in the running game, in today's NFL, passing is essential for success. A "team that is already fairly likely to win without him" will want as much consistency as possible, especially in the air.
This is easily the most interesting discussion of Tim Tebow I've ever read.
(Having trouble posting this, might be because of word count so I split it up)
Let's assume for a moment that it's true that the inferior team needs to use high variance strategies to increase their chance to win. Is that still the case in your view when they're the superior team in a matchup?
Denver entered the Miami game with the better record. Denver ranked 20th in Brian's rankings compared to 26th for Miami...and that may have overshot "Miami with Moore" because Henne probably represented about 70% of Miami's QB time at that point. The game closed at pick-em in the markets...in Miami...which suggests Denver is slightly superior because home field gets some accomodation in the pointspread (even in spots where home teams have been slumping).
If the game's a coin flip, do you still believe Tebow gives them a better chance to win? If Orton's a consistent B-minus, and Tebow's a volatile C, who do you want playing on the road against a C-minus?
Calling Denver "dismal" might be a bit extreme for 20th in Brian's rankings at kickoff.
Regarding the true range of Tebow's skill set, it's very tough at this point at this point statistically because he's caught some breaks in terms of schedule and meaning of the game.
Start One: at Oakland, Raiders were 6-7 at the time but still had an outside chance to make the playoffs. Tebow passed for only 138 yards and the Broncos were 2 of 12 on third downs.
Start Two: vs. Houston, Texans were 5-9 and way out of the playoff picture. Houston also had the league's 30th ranked yardage defense last year. Big stat game for Tebow against a lame duck defense playing a late season game at altitude. The bulk of the "Tebow's stats are better than you'd think" stuff that you hear on TV is largely a result of big numbers vs. a bad, lame duck defense at altitude.
Start Three: vs. San Diego, the week after San Diego had been eliminated from the playoff picture, also played at altitude.
Start Four: at Miami, rallying very late to forge a regulation tie with an 0-5 opponent that was starting a backup QB.
I think we all agree we need more samples vs. more motivated defenses in "need to win" scenarios. Will be fun to watch the home game this week against Detroit with the Lions coming off two losses. (Is Suh ready for altitude?)
There's a danger that evaluating Tebow at the moment catches him at his best vs. defenses at their worst. If you assume that he's going to be awkward to deal with at first until the league adjusts to his tendencies...then we have:
*A fresh Tebow
*Defenses still adjusting
*No starts yet against a winning team in a meaningful game
If you assume that trying to run the ball a lot is going to wear him down or lead to an injury (which is very common to QB's who run), that's going to create:
*A tiring Tebow
*Facing defenses who have learned his tendencies
*In games where the opponent's are at peak intensity
Can't draw a graph of a ski slope in the comments. But, that may be just as relevant a graphic as the one above. For the talk of Tebow improving from his coach...there would seem to be more immediate downside potential for that style given what we know about the health issues for running QB's, what we know about how NFL defenses have adjusted in recent years to running QB's (even back to Kordell Stewart, who was supposedly revolutionizing NFL offense until he didn't), and the unlikelihood that Tebow will run into many more teams who have just been eliminated from the playoff picture. (The next several weeks are pretty brutal in terms of defenses on the agenda).
Tebow's skill set may not actually be as broad as "also able to run" might suggest given his shaky decision-making and established weakness on third downs (11 of 42 conversions outside the Houston game). And, he may not be more volatile than Orton. That may be an illusion created from a relatively friendly schedule. Tebow plays on fire, is extremely well conditioned by all reports, and he does battle until the very end. So far he's faced one winless team and two teams who were already out of the playoff picture. He may be more of a one-dimensional bulldozer who's about to run into some well-reinforced brick walls the next several weeks.
for what it's worth, Tebow's YPA is higher than Orton's YPA this year and last year. However, he had a lower completion % in both years.
Wouldn't this mean that Tebow's passing has a higher risk and reward (i.e. you get more yards per completion, but you will get more incompletions as well)?
That sounds right, which mirrors the idea that he is a higher variance passer. But again, sample size is a huge issue in measuring his performance (and the quality of the samples as Jeff Fogle points out).
As for Denver being "favored" against Miami, if they were definitely a better team, I would agree that they should use a more consistent strategy. But, they are/were not definitely the better team. If it is a coin flip, as you say, than variance means nothing. In general, however, when playing in a league where you are below average, the higher variance will translate into more wins.
This is piggy-backing off of Keith, but maybe each team needs to have a high variance QB and a low variance QB, depending on the quality of the team they're about to play. Or maybe depending on whether they have the lead or not!
I suspect Tim Tebow has never played the single wing in his life, and if you look at the line spacings in a classic single wing, they're not well suited to a wide open modern football game.
He is, however, one of the best spread option quarterbacks ever, and totally capable of running out of that formation.
Read Option! I want to see if Tebow can outrun a linebacker, and then throw it over the safety's head when he cheats up playing run.
On Sunday Tebow had negative WPA and a success rate of all of 35%, FOers says he produced -54 yards below replacement level. In light of which...
1) While the point about "high variance" is well taken, it seems to be an assumption that Tebow is higher variance than Orton. Worse on average does not equal higher variance, it may mean just plain worse. (If they put me in at QB I'd be worse with a lot lower variance!) My image of a "high variance" QB is one who can throw of 40-yard rope but isn't very accurate about it -- maybe a Testeverde. If you're lucky with the ropes then you get points. But if a QB simply can't throw, where is the high variance supposed to come from?
2) My idea about putting in a single-wing package was half-joke, half-serious. The entertainment side is, wtf is Denver going to do this year anyhow? If this year in a losing season the fans want to see the rushing-and-passing combo Tebow they saw in college, then give them the Tebow they saw in college. And *I* want to see a real option offense in the pros! It's now or never.
The serious side is, if one wants to get high variance out of Tebow, what better way to do it? The wildcat was successful enough for a while, and the single wing is super wildcat that fits the tools Tebow has both running- and throwing-wise. If one wants to get high-variance out of him, one has to play to *what he can do* via running, improvising and short throwing. It'd provide a real preparation advantage every week. And a single-wing package used on 3rd and 4th downs could create a true 4-down (to heck with punting) offense such as we here would like to see.
That's all high-variance. More so than trying to make Tebow a normal drop-back QB, ISTM. And ... what is Denver going to do this year otherwise? If I were them, I'd really have done it, back in camp, prepared a package for Tebow that they could use selectively and expand during the season. Too late now, I fear.
If you were building a basketball team, would you start by collecting 3-point shooters? The high variance argument says "yes!"
The problem is the long run. Over the long run, mean performance is far more important than variance.
Read Option! I want to see if Tebow can outrun a linebacker, and then throw it over the safety's head when he cheats up playing run.
Yeah, that's it.
I suspect Tim Tebow has never played the single wing in his life, and if you look at the line spacings in a classic single wing, they're not well suited to a wide open modern football game.
Oh, I'm not speaking about the 1950s classic -- more an upgraded wildcat, which is a version of the single wing, with a real "double threat" at QB instead of the RB most teams used, and not weakened by shifting the QB out to flanker or wherever. An option offense with someone who can both seriously run and throw at QB that is not weakened at another position could cause problems to Ds. But they'd have had to prepare it in the offseason, so we're not going to see it.
He is, however, one of the best spread option quarterbacks ever, and totally capable of running out of that formation.
And the spread option is derivative of the single wing, evolved from it. That's my point. Letting Tebow do what he can do, while reducing the calls on him to act like a standard drop-back QB, is the "high variance" strategy, no matter how amatuerish it may look to pros like Elway and Fox.
Hey, I'm just imagining this for fun. Even if they did it, I'd expect them to get a "real" QB for next year and be done with it then. But why shouldn't NFL teams run experiments when they've little to lose and they actually might make some sense?
Someone already referenced this, but I think the discussion could be served well by expounding on this more - do Denver fans want a QB that will win them some more games when they're outclassed but fewer when they genuinely obtain talent and intelligent schema or do they want a QB who will lose more when they're outclassed but win more when they obtain that talent? Presumably, this is a franchise that tends to increase in quality, which would decrease the amount of teams that could outclass them.
It IS a long-view/short-view problem - while you are rebuilding (if you want to win now), put in the high-variance QB. Then get rid of him as soon as you are an average team with a favorable schedule. The problem is, no fan would want to see that, because the narrative would likely be this: "Wow this team is terrible, but this QB is so good, he WILLS them to win! Sometimes, when his defense/receivers don't let him down." The losses probably wouldn't be called his fault (unless his INT numbers are high) and the wins would be credited to him. Also, I feel coaches view the game the same way.
At the end of the day, the hidden argument of your post seems to be that Tebow will probably not lead them to macro-level success, he will likely just lead them to "not failure," which is curiously the opposite of the micro-level philosophy of employing the high-variance strategy as an underdog (playing to win instead of playing to "not-lose") in individual games.
Jim,
To the point that the spread option is merely the single wing reborn, Chris Brown of Smart Football has touched on that notion, and he doesn't agree.
I don't see it either. Multiple 1940s formations had a quarterback away from center (double wing, short punt), and the spread option appears more like Bill Yeoman's veer to me than 1930s power football.
D-
David, the Wildcat is a version of the single wing, simplified and modernized, as any number of web sites and single-wing coaches explain (for instance, plus, and). As it exists in the NFL today, there is no need to go back to 1940s schemes to put in a more expansive package of it utilizing the skills of a player better suited than others to run it today.
Jim,
I've been talking about the spread option, not the wildcat. As Chris explained in the link above, they're not the same.
I think you're talking extemporaneously, and not paying attention to what I say. Given that, I'll answer you more fully on my own blog.
David.
Jim, I've been talking about the spread option, not the wildcat.
David, I've been talking about the single wing since my first comment up at the very top.
As Chris explained in the link above, they're not the same.
Who said they are? What are you arguing against?
Look, if you want to talk about things other than Tebow running the wildcat-single wing, such as the origins of the spread option, that's fine. But the logic of saying the spread option is different than the single wing, and that somehow this is an argument against using the single wing with a player who fits it unusually well, escapes me.
The only relevance I can see is your saying Tebow *ran* very successfully out of the spread option, and my saying good! Your own source there notes the similarity of running from the spread option to doing same from the single wing: shotgun formation, option running schemes, emphasis on misdirection and fakery. Tebow's long familiarty with same can only be an advantage running in a single wing. Right?
As to your: "To the point that the spread option is merely the single wing reborn..." What point? Who ever said such a silly thing? I sure never did.
Obviously the purpose of the single wing is power running controlling the line of scrimmage using an option scheme with opportunistic passing.
How do you compare the spread option passing game to this, as a criticism? To put Tebow in a "wide open" passing game in the NFL this year would be a disaster.
Your source says: "coaches who developed today’s modern offenses, like Rodriguez and Malzahn, did not spend their time meticulously studying the single-wing tapes of yesteryear".
Sure. Obviously. Who cares? Did somebody somewhere say they did do that?
I think you're talking extemporaneously, and not paying attention to what I say.
I've paid great care and attention to everything you've written. I'm just waiting for you to say something relevant to Tebow running the wildcat-single wing.
this article is completely meaningless.
the graph is not real, and even then it is wrong, and even then it does not address the question, what maximizes GWP.
one would hope this would be a site about statistical analysis, not blah blah my opinion based on fantasy blah blah
this comment is completely meaningless.
the pseudonym "anonymous" is not real and even then its wrong to be a troll, and even then the comment adds no value to the discussion.
One would hope people had one ounce of decency, but instead they let their personal biases drive them to troll the internets looking for opportunities to get upset about an article in the slightest way suggesting that playing Tebow in certain situations might be a plausible idea.
Go back under your bridge.
Chris said...
this comment is completely meaningless.
the pseudonym "anonymous" is not real and even then its wrong to be a troll, and even then the comment adds no value to the discussion.
One would hope people had one ounce of decency, but instead they let their personal biases drive them to troll the internets looking for opportunities to get upset about an article in the slightest way suggesting that playing Tebow in certain situations might be a plausible idea.
Go back under your bridge.
endquote.
do you have a comment on the facts I posted? I am sure namecalling is quite fun for you, but you fail to address the fact that this post, and your comment are equally meaningless.
This site contains many very good articles, with keen statistical analysis. I would hope in the future there is some editorial control applied to prevent articles like this one.
Chris said...
this comment is completely meaningless.
....
Go back under your bridge.
endquote.
do you have a comment on the facts I posted? I am sure namecalling is quite fun for you, but you fail to address the fact that this post, and your comment are equally meaningless.
This site contains many very good articles, with keen statistical analysis. I would hope in the future there is some editorial control applied to prevent articles like this one.
You did a poor job of presenting your "facts". You were extremely rude in your comment. You want to affect change, then bring something to the table. Clearly you have no interest in that, because you are using stupid derogatory remarks to the table. That's all I have, unless you actually want to raise your game a bit here.
(By the way, I have no problem in general w/ someone disagreeing with something written on the blog, but lets be civil in doing so, and actually offer some value ourselves when disagreeing, rather than saying...."this article is completely meaningless")
For those of you who followed the links in Brian's roundup this week....my complaint here is that anonymous brought a bottle of ketchup to the potluck, without bringing any hamburgers. :-)
So, I think the takeaway going forward, is that the appropriate response to similar commenters in the future is, "Where's the beef?"
Thanks for your support, Chris, glad you're enjoying my stuff. I thought the article prompted a pretty good discussion on the topic of high variance strategies for underdogs.
Always open to suggestions that can make my writing and analysis better, anonymous, would love to hear your ideas.
Thanks for reading
Yes, this is logical.
Lets build a team around a QB that has no accuracy and runs around like he is playing school yard football. Tebow is actually making the right play whenever he runs the ball b/c he can't hit the broad side of a barn when he tries to throw it. He has played maybe 10 minutes of decent football with ok decision making as a starter in the past 2 1/2 games.
Let's not get high on a guy who found a way to beat the Dolphins who always find a way to lose.
I honestly feel bad for the kid and hope he finds some type of H-Back hybrid position in the NFL. It's not his fault the Broncos are idiots and wasted a first round pick on him. He is great for punching it in the end zone and I'm sure they can find other ways to utilize his skills but he is never going to be a successful NFL starter.