Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Futures: My Expansion Franchise

You've just been awarded an NFL expansion team and must build your personnel department. How would you do it? Matt Waldman takes on the exercise.

08 Mar 2010

Wisdom of Crowds in Review: QB

by Bill Barnwell

Before the 2009 season, I went on our then-new Twitter account to ask about 2,500 or so followers how they thought a variety of NFL players would perform in the upcoming season. We got somewhere between 20 and 50 predictions for each player in question, yielding a "Wisdom of Crowds" prediction.

Now that the 2009 season is over, it's time to review. I'll be looking back at what the crowd said about each of the players in question, what actually happened, and what both they and I should have seen to help improve each of our thoughts on the player. (For those of you who are on Team Potato, I'll also note where I foolishly disagreed with the crowd.)

Remember that the respondents were asked to give a stat line for the player in question assuming that said player participated in all 16 games and did not suffer any major injuries. You can find the original articles here and here.

First, the quarterbacks.

Matt Cassel, Chiefs

Predicted: 3,289 passing yards, 19 TD, 15 INT
Actual: 2,924 passing yards, 16 TD, 16 INT

Cassel was always going to be overrated heading into 2009; he was going from an elite offense to a terrible one, and from a really easy schedule to what ended up being just about an average one. Most people really underestimate how huge changes in context can be for players, and while Cassel's numbers don't look that far off (and the prediction would have been within 170 passing yards if Cassel had played 16 games at the same average YPG) from the predicted totals, the touchdown-to-interception ratio is off. I actually gave a predicted line of 3,400 yards with 18 TD and 20 INT here, which was too much volume.

Matt Ryan, Falcons

Predicted: 3,589 passing yards, 22 TD, 14 INT
Actual: 2,916 passing yards, 22 TD, 14 INT

Not a bad job by our respondents, getting those touchdown and interception totals spot-on, but Ryan also missed what amounts to three games with an injury. If you extend his average per-game performance to 16 games, he had 27 TD and 17 INT, but strangely, he hits... 3,589 passing yards. So our panel had it square on the nose one way or another. I said his totals would be lower than the predicted figures, but I can't claim that as a success because that only happened thanks to injury.

Tom Brady, Patriots

Predicted: 4,013 passing yards, 31 TD, 12 INT
Actual: 4,398 passing yards, 23 TD, 13 INT

Neither I nor our readers expected Brady's schedule to be so tough. I pegged his TD total at closer to 35, which just didn't happen. The interception figure is right on.

Brett Favre, Vikings

Predicted: 3,067 passing yards, 19 TD, 22 INT
Actual: 4,202 passing yards, 33 TD, 7 INT

No one saw that coming. Certainly, I didn't; I agreed with the readers, noting Favre's ridiculously easy schedule from the year before, while questioning his health. He was fine, and while he had a fluky INT rate for most of the year, he blew away anyone's expectations.

Tony Romo, Cowboys

Predicted: 3,754 passing yards, 26 TD, 16 INT
Actual: 4,483 passing yards, 26 TD, 9 INT

I pointed out that Romo's historical yards per game rates simply did not jibe with his predicted totals; thanks to the emergence of Miles Austin, Romo had his best seasonal YPG as a pro. He cut his historical interception rate in half, which should regress to the mean next year.

Jason Campbell, Redskins

Predicted: 3,220 passing yards, 20 TD, 12 INT
Actual: 3,618 passing yards, 20 TD, 15 INT

Sadly, no one went out and just predicted "pain". I suggested he'd go 3,450/18/12, but there's not much of a difference between that and the predicted total. When I do this next year, whoever just types "average" in reply gets a pizza trophy.

Eli Manning, Giants

Predicted: 3,418 passing yards, 22 TD, 15 INT
Actual: 4,021 passing yards, 27 TD, 14 INT

Eli seemed like an obvious candidate for regression, but thanks to the emergence of Steve Smith and the decay of the Giants' running game, he took his game to the next level and had his best season as a pro.

Jay Cutler, Bears

Predicted: 3,766 passing yards, 24 TD, 16 INT
Actual: 3,666 passing yards, 27 TD, 26 INT

I wondered how on Earth Jay Cutler was going to get to 3,766 yards with that corps of receivers. The answer, apparently, was "The offensive line falls apart and Matt Forte falls down." (More on him next time.) A crowd will never be able to predict an interception rate as high as Cutler's; too many people will regress his numbers to the mean to make that possible.

Ben Roethlisberger, Steelers

Predicted: 3,463 passing yards, 23 TD, 15 INT
Actual: 4,328 passing yards, 26 TD, 12 INT

I figured that Roethlisberger's yardage total was too high and that he'd struggle to get that many touchdowns, but he hit his TD peak from 2007 and his YPA went back towards his rookie and sophomore levels, blowing away even the predicted totals that I thought were high. KUBIAK is a long ways away from generation, but I'd imagine it will regress Roethlisberger a bunch heading into 2010, regardless of what happens to him off the field.

I'll be back later this week with running backs, including the one thing the crowd saw coming in 2009 that no one else did.

Posted by: Bill Barnwell on 08 Mar 2010

73 comments, Last at 02 Apr 2010, 11:02pm by Tim Wilson


by LouG (not verified) :: Mon, 03/08/2010 - 12:15pm

Nice HR reference w/ pizza trophy

by Theo :: Mon, 03/08/2010 - 12:17pm

I don't get it.

by Cro-Mags (not verified) :: Mon, 03/08/2010 - 1:13pm

What, no love for the Clubber Lang reference?

by Bobman :: Tue, 03/09/2010 - 1:15am

Ah, the prediction: Pain.

I didn't realize I got it until I saw your note.

by Theo :: Mon, 03/08/2010 - 12:16pm

Lesson learned:
crowd is right, but passing yards need to be added or extracted by experience.

by C (not verified) :: Mon, 03/08/2010 - 12:24pm

Barnwell, I think you are hyping up the schedule too much.... " nobody predicted his schdule would be this easy, or his schedule would be that hard etc.". Why can't it be that people were "right" or "wrong"? Some people might have looked factored in schedule.

I'm not bashing you, but it seems like the "wisdom of the crowd" did a better job predicting than you did.

I know Brady did worse than I'd expect ( I was wrong)
Cutler did worse than I'd expect ( I was wrong)
Favre did better than I'd expect ( I was wrong)

I was more Bullish on Aaron Rodgers ( right)
I was more Bullish on Romo than most people (right)
I was more Bullish on Matt Schaub (right)
I was more Bullish on Eli Manning ( right)

I was more negative on Jason Campbell ( wrong stat wise), but then again how many garbage yards did he pick up? His team is picking 4th in the NFL draft. How many people that predicted good stats saw good things happening for the Redskins?

Phillip Rivers had a better year than I would have guessed last year.

Next year I'm still putting my money on Aaron Rodgers. The Green Bay Packers will win 2+ super bowls in the next 10 years and Rodgers an MVP.

by MJK :: Mon, 03/08/2010 - 12:29pm

C, I don't recall you ever being negative about Jason Campbell...

by C (not verified) :: Mon, 03/08/2010 - 3:00pm

but that's the thing, if you look at his stats, they don't look THAT bad. But then consider he was on the 29th best team in the league ( with a weak last place schedule), lost to the Lions, and that his stats were heavily skewed and picked up when his team was losing...

It's like being on a crappy NBA team and picking up garbage points when the game is over. Or being a WR on a crappy team that picks up catches beause his team throws a lot, but can't do anything when the game is on the line.

His stats are weird because of his style of play. Ultra conservative QB, who picks up garbage stats ( but doesn't have THAT many bad picks but he does fumble). It's death by 1,000 cuts to lose a game, rather than a Cutler-like figure throwing 4 picks and making it obvious why his team lost. At some point, you have to take chances ( and not throw all checkdowns), and that's why your team losses.

That's probably why stats people ( Barnwell) like Campbell ( his stats weren't terrible), and I hate him so much ( his play WAS terrible).

by Pat (filler) (not verified) :: Tue, 03/09/2010 - 12:08am

I still think it's hilarious that you think a QB in the final year of his contract should "take chances" rather than go for options that the playcaller specifically puts in there.

C'mon, the Redskins playcalling situation was a disaster last year. Sherman Lewis's previous playcalling was Bingo at a retirement home, and that's not a joke. That's Art Shell Raiders-level incompetence.

Be fair and criticize the guy for something he should have done right - like his amazingly poor pocket presence.

by jebmak :: Mon, 03/08/2010 - 5:26pm

C, I don't recall you ever being negative about Jason Campbell...


by DavidL :: Mon, 03/08/2010 - 1:10pm

I admit, I thought this feature was just a gimmick, but it turns out to be pretty interesting. Nice idea!

by ChaosOnion (not verified) :: Mon, 03/08/2010 - 1:16pm

Is Team Potato derogatory to EMB fans posting in the TAtEr section? If so, I approve.

by bubqr :: Mon, 03/08/2010 - 3:28pm

TATE is taking over the world !

by HostileGospel :: Mon, 03/08/2010 - 9:35pm

Hah, no. It's a reference to this.

There's a place I want to be. It's the NovaCare Center. That's in Philadelphia. One NovaCare Way, where the Eagles practice and then they eat cafeteria food and they watch film and we eat and we have fun.

-Donovan McNabb

by sundown (not verified) :: Mon, 03/08/2010 - 1:26pm

A couple head-scratcher comments up there from Bill. He'd have been better off with Brady and Cassell just saying "pretty good job everybody."

"Neither I nor our readers expected Brady's schedule to be so tough."
What supports that statement? The prediction was actually pretty good--guessed high on TDs, but yards and interceptions were right there. (But note the crowd was still far more conservative than Bill's prediction of 35 TDs.) I'd argue the schedule had far less to do with it than the fact it was Tom Brady--give him a different name and everybody would have guessed lower on the TDs.

"Most people really underestimate how huge changes in context can be for players"
Again, he gives the respondents far too little credit; they seem to have done a very nice job predicting there'd be lots of struggles with the Chiefs. Reality basically split the difference between their prediction and Bill's.

by Theo :: Mon, 03/08/2010 - 2:18pm

Are you saying you expected self criticism from Bill?
Hahahaha. Good one.

by sundown (not verified) :: Tue, 03/09/2010 - 2:43am

And as somebody below caught, Bill couldn't even look up the numbers correctly--Brady had 28 TDs, not 23. So, the crowd was only high by 3. That's pretty awesome. So much for the BS about not being able to predict how hard the season would be, yada, yada, yada.

by Tim Wilson :: Fri, 04/02/2010 - 11:02pm

Also, Roethlisberger's career high TD level in 2007 was 32, not 26. So no, he didn't achieve it last year.

This piece really seemed sloppy.

by Jmagik (not verified) :: Mon, 03/08/2010 - 1:35pm

I have to say it's getting a bit old that every single post about a player that outperformed FO's expectations is qualified with "but he'll regress to the mean next year." First off, this is closer to tautology than prediction. Second, it comes off lazy half the time. You say we should look at context, which is true, but one should also pay attention to its evolution over the course of multiple seasons. For example, I highly doubt Romo is going to regress to his career average in interceptions next year. However, as FO's Audibles sections say, none of you guys watch many Cowboys games, which is fine since you have teams you prefer to watch... but Romo's interception drop had to do with more things than the emergence of a #1 wideout and good offensive line play. In fact, our o-line played a bit worse this year.

by Arson55 :: Mon, 03/08/2010 - 2:43pm

Exactly, Romo's interception totals dropped, in part, because he made the conscious decision to throw the ball away more and to willingly just take sacks instead of always trying to make plays when there was nothing there.

by GoVikes (not verified) :: Mon, 03/08/2010 - 2:58pm

I agree. It's tough to predict players will return to the mean when a) the QB is young and thus still developing, not to mention has a smaller sample of games to work off and b) the assumption is that the FO mean is correct to begin with. Many of the guys listed here - Cassell, Ryan, Romo - will likely keep getting better as they continue to grow and their offenses mature (barring injury of course). The only player listed who is likely to "return to the mean" would be Favre if he returns, and that has more to do with his age and the large body of work that can be evaluated to make a more accurate mean (we know he historically throws lots on INTs, so it's unlikely he'd throw only 7 again).

by C (not verified) :: Mon, 03/08/2010 - 3:05pm

I don't like that " we can't possibly be wrong" attitude, and the " it was all the schedule" attitude.

I think with QB's people often count last season as " what is"... when Kurt Warner was having weak seasons a few years ago, it was assumed he was done.. Favre was done... Byron Leftwich was "good" when his stats were "ok"....

Getting rid of TO was expected to be a death blow to Romo, but it let him focus on the process ( instead of a ball hungry player) that let him settle down in much the same way Eli Manning settled down when Plexyglass left and to a lesser extend Shockey.

Rather than having guys that you HAD to force the ball to, it allowed him to more so play QUARTERBACK instead of ball distributer which can be harder than it sounds. Tony Romo is a sharp guy. If he wasn't in the NFL I'm sure he'd be very successful at business, more so than your average guy. He's well read, funny, witty, smart etc. I still don't understand why Donovan Mcnabb and Tony Romo get so much hate.

by Jmagik (not verified) :: Mon, 03/08/2010 - 3:35pm

Among other things, relative to media expectations, McNabb and Romo have often come off as not serious enough about football. When asked if his team had what it took to win a SB, for the 3rd time in the interview, Romo wryly said "I don't know, I've never won a Super Bowl before." I think some wannabe Rick Reilly once wrote an entire article about how Romo is a piece of shit because he wears his hat backwards in interviews. We all remember the McNabb-partying-after-the-Championship-loss debacles.

One thing that would be interesting is if someone could write a comparative study of football players' attitudes towards football. Who is Sean Taylor, "we play a kids' game for a king's ransom"? Hines Ward, saying Roethlisberger should play with a concussion because it's an "important" game? Manning, always saying the right thing; Leaf, saying "I'm going to Vegas." Deion, dancing down the sidelines; Barry, flipping the ball. These things are often looked at just in terms of flamboyance or apathy or passion or whatever, but there's also this "seriousness" aspect. We all suspend disbelief and act like this is some REAL SERIOUS BUSINESS and forget that we're really just cuaght up. Some players are too, and some aren't. Some of those that aren't will always disappoint fans -- think Vince Carter in the NBA, someone who never played to his potential, but also probably has a few less miles on his knees for it, and who had more fun during the offseason enjoying his cash for it. Brett Favre is an interesting case... busts his ass in the season, does jack shit out of it. Success in championships will always be the metric the mainstream pays attention to though. Look up Lawrence Taylor's answer to the question, "what can you do that no other linebacker can?" (Spoiler: Answer wasn't football-related.)

Anyway, back to Romo. He got a lot more serious this year. I think that the media pressure and whatnot caught up to him. Either way, he certainly reeled in the "he's having fun out there" persona quite a bit. Media liked him more, commentators liked him WAY more. I watched every game, and he easily could've had 15 INTs again, but you know, last year he could've had 25. Few bad/good bounces change things... if his back doesn't give out (big if) and he plays 5 more years, I see him averaging 28 and 13, Brady-like numbers if not truly Brady-like production. But as long as he has his smirk and his backwards hat... he's just not a good stand-in for the BMOC masculine fantasies sportswriters have of all athletes. And who knows, maybe that lack of seriousness will keep him from breaking through, but... and I kind of wanna stab myself for saying this... he seems to be... having more fun than most players (when not writhing on the ground in back pain -- thanks again for keeping him in during garbage time vs. Philly, Wade!).

by C (not verified) :: Mon, 03/08/2010 - 4:24pm

I'd tend to agree that Romo's $hit grin, " I'm just having fun out there" hurts his perception, just like Mcnabb probably hurts his after a bad game when he's smiling, throws his hands up and casually jobs back to the sidelines. Just like Eli Manning's Aww shucks shoulder shrug hurts his perception.

I think both guys are very smart. Mcnabb is a film study buff and knows his stuff, Romo is a real assertive astute guy. I'll throw Eli Manning in there as well because he's smart too. The thing is I think these guys don't just show up and "get their game on", I think they more or less approach it as an academic TEST of X's and O's under stress. It's a thinking man's game, they did their homework, are expecting to see certain coverages etc, and then they execute. They aren't getting all hyped up on their emotions, they are trying to think clearly/rationally during the game.

Michael Vick on the other hand just shows up... gets real into the game ( because he didn't do his work Mon-Sat, and the crowd/announcers love seeing him try and wing it.

Drew Brees is doing his pre-game rah rah before the game and he's the best of both worlds ( he is studious) for the Media, same with Favre... During the season ( not off season).

I think fans read a lot into the body language, dress, perception of these guys a lot. I think with WR's you often get MORE credit for being a hot dog self promoter ( think Chad Johnson over Andre Johnson). With QB you seem to get the most credit with a strong "leadership" perception... not an immature punk like JP Losman.

by DeltaWhiskey :: Tue, 03/09/2010 - 3:04am

" I highly doubt Romo is going to regress to his career average in interceptions next year."

But I would expect him to regress towards his career average, which is what regression to the mean refers to. It is not predicting that he will move exactly to his mean, but simply stating that his numbers are going to move towards the mean.

by DaveRichters :: Tue, 03/09/2010 - 9:22am

But I would expect him to regress towards his career average, which is what regression to the mean refers to.

My nephew was exactly 3 feet tall in October, then 3 feet 2 inches in February. I expect he's regressed to his lifetime average height since then.

Also, the number of living dinosaurs on the Earth is zero RIGHT NOW, but wait until "regression toward the mean" takes place.

I don't think I've ever read the word "regress" so many times in a single article, but to the auteur's credit I don't actually believe that he thinks things never change between measurements save for random variation.

by Alex51 :: Tue, 03/09/2010 - 11:16am

Romo has been a starting QB for 4 years now, and in 3 of those years, his interception rate was above 3%. In 2009, it was 1.6%. So unless you think Romo, who was about average at avoiding interceptions for his first three years as a starter, has suddenly become the best in the NFL at that skill, his interception rate will likely be higher in 2010 than it was in 2009.

To put things another way, from 1999-2008, there were 41 instances of a QB starting a full season with an interception rate below 2%. In only 6 of those instances was that QB's interception rate below 2% again the next year. And the QB's that did repeat that low interception rate had lower career interception rates than Romo does.

Now, none of this guarantees a regression to the mean. It is possible for Romo to beat the odds and have a very low interception rate next year, but it wouldn't be wise to bet on it (otherwise, it wouldn't be called "beating the odds").

by JonC :: Thu, 03/11/2010 - 12:09pm

The issue isn't with the concept itself but the way the concept is used around here--that is, as a substitute for analysis. An unusually high or low streak should be the occasion for further investigation, rather than an off-hand "he'll regress to the mean next year, moving on...." The numbers themselves might *suggest* this, but they sure as heck don't explain why it is. Aren't we supposed to disentangle correlation and causation here? The best case in point is the Roethlisberger para.

by DeltaWhiskey :: Tue, 03/09/2010 - 12:05pm

Alex51, thank you for responding before my head had a chance to explode.

Another look at regression to the mean, on the opposite side of the INT spectrum, is Jay Cutler. Last year his INT rate was 4.7% against a career average of 3.5% For CHI fans sake, I (hope and) expect him to regress toward the mean.

Statistical regression to the mean does not refer to scores being above or below the mean, but refers to a tendency of outliers to move towards the mean on subsequent measures.

BTW, I don't think auter means what you think it does, but I like the idea of FO the movie.

by Temo :: Tue, 03/09/2010 - 1:38pm

Delta, while you are "more" correct than the poster above who attempted to refute you in an odd way, I don't think you're understanding the argument.

Typically, regression to the mean suggests regression to the TRUE mean. The argument put forth is that his true mean is now different from his career mean, because he is a different quarterback now. Whereas previously his play might resulted in a "true" interception rate of 3.2% (or whatever), his new interception rate may be 2.5%. Or it could all be hogwash, and his true rate is indeed 3.2%. But that's the argument framed above.

Obviously, Barnwell didn't specify either one, his statement was far too broad to say anything.

by Kevin from Philly :: Tue, 03/09/2010 - 3:05pm

Given the fact that the Cowboys won't be playing the AFC West and the NFC South this season, I'd say the odds of Romo's interception rate going up are pretty darn good.

by Temo :: Tue, 03/09/2010 - 3:46pm

First, they faced an above-average slate of pass defenses this year by DVOA.

Second, no one is disputing (thus far) that his interception rate will regress next year. His 2009 rate is unsustainable. The question is to what extent it will regress.

by Alex51 :: Tue, 03/09/2010 - 5:24pm

Second, no one is disputing (thus far) that his interception rate will regress next year. His 2009 rate is unsustainable.

Really? Than what the hell have we been arguing about? All Barnwell meant was that his 2009 interception rate was unsustainable and wouldn't be repeated.

The question is to what extent it will regress.

Barnwell didn't specify exactly how far it would regress, so I don't see why people are complaining about his statement unless they disagree with his basic assumption that Romo's low interception rate was largely due to luck. If he had said "regress towards the mean" instead of "regress to the mean", would everybody be ok with it?

by Temo :: Wed, 03/10/2010 - 8:43am


by DeltaWhiskey :: Wed, 03/10/2010 - 2:56am

"who attempted to refute you in an odd way"

to say the least

by DaveRichters :: Wed, 03/10/2010 - 9:00am

the poster above who attempted to refute you in an odd way

Am I the odd one? It's not clear.

Regression is well defined for things that have some "true" value that doesn't change between measurements. The odd thing is to think that everything you can measure multiple times is inherently stable. You need more than just the ability to measure things to expect regression, even for outliers. I'm the tallest person in my family no matter how many times we line up and check.

Typically, regression to the mean suggests regression to the TRUE mean

No. It refers explicitly to some measured value. You're thinking with logic, not with statistics. Don't fret, your way is better usually.

by DaveRichters :: Wed, 03/10/2010 - 9:20am

Statistical regression to the mean does not refer to scores being above or below the mean, but refers to a tendency of outliers to move towards the mean on subsequent measures.

Yes, but that tendency isn't always there. In some situations you may expect the outliers to be even farther from the mean. Let's say a class is given two tests and you need to get at least 70 on both to pass and then the passing grades are assigned on a curve. The people who did poorly may have less motivation for the second test, knowing they can't pass anyway, and the people who did well may have more motivation and do better. I'm not claiming that regression will not occur for measurements involving football or Romo, I'm just saying that regression is not a universal truth to be expected for all things that can be quantified multiple times.

BTW, I don't think auter means what you think it does, but I like the idea of FO the movie.

It is a French word that means "author" and it can be applied to the creator of artistic endeavors in basically any medium. But sure, it is most commonly used in English for movie-makers. FO the movie... I dunno, I'll wait till it's on cable.

by DeltaWhiskey :: Wed, 03/10/2010 - 11:32am

Sh*t, after reading through your responses four or five times, I finally understand that you seem to simply be arguing that sometimes, some measures (like one's height relative to the average familial height) there is not going to be regression to the mean. So, in the case of Romo, do you think his interception frequency will increase (regress towards the mean) or decrease (move even farther from the mean) next year and why? What factor for Romo will be the equivalent of grading on the curve in your example above?

by DaveRichters :: Thu, 03/11/2010 - 7:38pm

after reading through your responses four or five times, I finally understand that you seem to simply be arguing that sometimes, some measures (like one's height relative to the average familial height) there is not going to be [sic] regression to the mean

Just for parsimony, I read this four or five times.

What factor for Romo will be the equivalent of grading on the curve in your example above?

That's an odd question. I'm not actually making any prediction about Romo, but it is really easy to come up with plausible reasons why a football player might get better (or worse) from one year to the next. It might be that there is a general trend for interception rates to decline for players Romo's age because they get better. For Romo specifically, it might be that he changed his criterion and is now more conservative. Who knows? I'm not making any claim or prediction about Romo's stats. If your point is that it is reasonable for a person to believe that Romo will throw more interceptions because he was an outlier, I agree. However, I also think it was silly to point to every disagreement between predicted and actual stats and say they are all flukes.

by DeltaWhiskey :: Fri, 03/12/2010 - 2:46am

"However, I also think it was silly to point to every disagreement between predicted and actual stats and say they are all flukes."

It is silly; however, I don't think I ever said this, nor do I think it is true. If I created that impression, then I apologize.

by DaveRichters :: Fri, 03/12/2010 - 10:56pm

Oh, sorry, I didn't mean you. My original comment criticized this tendency obliquely and I was now making it explicit; I meant to direct that comment at the author of the article.

by Mr Shush :: Sat, 03/13/2010 - 11:16pm

That would be "auteur", old chap. "Auter", were it a word, would be pronounced "oh-tay", not "oh-ter". It would also probably be a verb, maybe meaning "to author", which the French would in fact most likely render as "ecrire"* (for literal uses) or creer* (for figurative ones).

And no, of course not everything measurable regresses to the mean, and yes, of course Romo is enormously likely to regress towards and perhaps even past his mean interception rate, be the rate in question historical or counterfactual.

*Apologies for lack of accents - I'm blowed if I know how to do them in HTML.

by DaveRichters :: Sun, 03/14/2010 - 1:16am

You should be correcting the person who misquoted me; I spelled it correctly.

by Kevin from Philly :: Mon, 03/15/2010 - 9:44am

So Buckwheat was French? Who knew.

by The Other Ben Johnson (not verified) :: Mon, 03/08/2010 - 4:50pm

I seem to remember from the initial publication of the responses that you threw out "troll" answers, right? Any idea what those troll answers would do to the averages? I could see a diehard Favre fan or two trolling this question and pushing the crowd wisdom closer to actual. "Wisdom of crowds" also includes outliers. It turns out we need a buncha yokels and sarcastic weirdos to make us smart as a whole.

by Kevin from Philly :: Mon, 03/08/2010 - 5:06pm

So RaiderJoe's input would give a better estimate of JaMarcus Russell?

by Bobman :: Tue, 03/09/2010 - 1:20am



and, might I finally add, No.

Pretty sure rj's eternal optimism/blind spot would have resulted in a prediction along the lines of 3,600, 24, 15. In fact, I am pretty sure I saw something like this.

Not so outlandish that you fall down laughing, but enough for you to say "Closing time, and the Sierra Nevada tap's got air in it."

by Nathan S (not verified) :: Mon, 03/08/2010 - 4:51pm

What happened to the high/low predictions? Those were interesting bookends for the average.

by RickD :: Mon, 03/08/2010 - 5:30pm

"A crowd will never be able to predict an interception rate as high as Cutler's; too many people will regress his numbers to the mean to make that possible."

I predict he'll have 2,123 interceptions this season!

by Bobman :: Tue, 03/09/2010 - 1:21am

well, there goes the average, shot to hell.

That prediction also probably makes the median double or triple, too, right?

(yes, kidding)

by Dean :: Mon, 03/08/2010 - 5:42pm

I guess my big gripe is the same one I had in September. I just don't care about conventional wisdom. Is that not what we're measuring here?

I come to FO to get AWAY from conventional wisdom.

Having said that it looks like it's a fairly popular idea, so I'm not saying stop it just because one person doesn't give a shit. If it's getting people to click on the link instead of the little X in the upper right...

by sundown (not verified) :: Mon, 03/08/2010 - 6:17pm

Then you came to the wrong place, as Bill's projections demonstrate. He was right there with the crowd. The problem, as always, is that awesome historical stats don't mean you can see the future. It makes you one helluva Monday morning quarterback, but it probably won't get you banned from betting in Vegas.

I haven't seen this in a bit, but a couple years back Aaron used to pretty routinely respond to posts like this stressing how these weren't ever intended to be predictive tools. But, of course, many (including, rather oddly, some of the FO staff) wants to use them that way.

by tuluse :: Mon, 03/08/2010 - 8:00pm

Don't you need to have a conventional baseline to see if your unconventional wisdom is actually superior?

by AnonymousMe (not verified) :: Mon, 03/08/2010 - 5:46pm

"I figured that Roethlisberger's yardage total was too high and that he'd struggle to get that many touchdowns, but he hit his TD peak from 2007"

I'm not sure I understand this comment. He threw 32 TD's in 2007.

by Treima (not verified) :: Mon, 03/08/2010 - 8:39pm

Yeah, and Brady had 28 TDs this season, not 23. A lot of the information here is off.

by The Guy You Don't Want to Hear (not verified) :: Mon, 03/08/2010 - 6:06pm

That was an excellent tease for the next article at the end.

by Bobman :: Tue, 03/09/2010 - 1:25am

I'm not sure it even remotely makes sense (or were you being sarcastic?).

How can a crowd (not an infinite number of people, but a pretty big number, right?) do anything that no one else does, if, by implicationm the crowd sort of represents a sloppy mish-mash of everybody....?

Bill, did you mean "none of the so-called experts" rather than nobody? That would make more sense.

by sundown (not verified) :: Tue, 03/09/2010 - 2:34am

It makes zero sense. It's the same as saying "Nobody saw this coming except for everybody."

by Mr Shush :: Tue, 03/09/2010 - 10:06am

That's just not true of this particular "crowd", though, is it? This is 20-50 people - quite a small sample and, crucially, by no means a random one. It shouldn't be too surprising if the collective opinion of a small group of intelligent, statistically-oriented hardcore fans produced a much better prediction than almost anybody else in at least one of the several things they predicted. Admittedly "almost nobody" would probably have been more accurate than "nobody" - there's bound to be someone else out there who thought the same - but I'd call it mild hyperbole rather than outright nonsense.

by sundown (not verified) :: Tue, 03/09/2010 - 1:54pm

But if this article is any indication of how the RB one will be structured, he's not talking about the FO crowd versus outside commentators (who could be total morons and/or lacking key statistical references) but rather against himself and possibly other FO staff. If that sort of insular crowd accurately predicted something that absolutely blindsided FO staff, that's going to either be chance or a case of the staff making a really bad guess despite having the same tools at their disposal.

by sundown (not verified) :: Tue, 03/09/2010 - 2:02pm

I sort of got my tongue tied up in all that explanation, but to put it simpler, I'd liken it to the class predicting something the professor absolutely couldn't see coming. That's either going to be luck or a really bad effort by the professor.

by The Guy You Don't Want to Hear (not verified) :: Tue, 03/09/2010 - 5:45pm

I assumed that Bill meant, as you say, "none of the so-called experts."

I think it was a good tease, because I am now looking forward to that article to see at what he was hinting. I believe that is the point of a tease.

Plus, there seems to be an awful lot of Bill Barnwell bashing here lately, so I thought it would be nice to be positive one time.

Finally, I concur with Mr Shush's comment. I would have written something similar, but he already has.

by Miles but not from Austin (not verified) :: Tue, 03/09/2010 - 8:22pm

Maybe he is getting bashed but miss-posting multiple stats in this short an article is pretty sloppy. And they weren't even simply typos--he uses them in his arguments like that whole part about nobody being able to predict how tough the Pats schedule would be. Ouch. And how long does it take to scan the posts and realize you messed up? The mistake was caught yesterday and Bill's been posting Extra Points today.

by DeltaWhiskey :: Tue, 03/09/2010 - 3:17am

If only on a statistically based website there were some statistical methodology that could be employed to determine the accuracy of the crowd's predictions.

by Will Allen (not verified) :: Tue, 03/09/2010 - 2:37pm

Romo's largest problems in previous years were inexperience and crappy pass protection. He, obviously, had more experience in the most recent season, and his protection was somewhat better. Strangely, when, in the playoff game against the Vikings, his pass protection collapsed, his stats went straight into the sewer.

I didn't think Favre would remain healthy, and was quite happy to be wrong. I hope I'm wrong again next season.

by Independent George :: Tue, 03/09/2010 - 8:23pm

Will, I'm very disappointed that you didn't join the Will Allen discussion thread.

by Will Allen (not verified) :: Wed, 03/10/2010 - 2:08pm

Eh, I came to it kinda' late; a certain overtime game 6 weeks ago temporarily dampened my enthusiasm for all things NFL, so I've been dropping in here a little less frequently.

by Jeff Fogle :: Wed, 03/10/2010 - 12:42pm

Bill, can you do the same thing with KUBIAK?

Or, line up
for each of the guy's listed in TD/INT/Yards? Would be interesting to compare and contrast the different approaches I think. A chance to celebrate either the quality of Kubiak or the quality of your readership.

by just your average joe (not verified) :: Wed, 03/10/2010 - 7:48pm

Regarding Brady, he actually threw 28 TDs (not 23 as is stated in the article). That brings both the crowd and Bill much closer to the mark. Particularly close for the crowd, who did a great job overall.

Also, the author's talk about not being able to predict the toughness of the schedule Brady faced seems a tad irrelevant given his track record. Throwing out his first year as a starter and the crazy-good 2007, he's been remarkably consistent: 23-28 TDs vs. 12-14 INTs. While I certainly wouldn't say it was unreasonable to predict 31 TDs (or even 35 as the author did), he had eclipsed 30 only once before in his career.

by Mr Shush :: Sat, 03/13/2010 - 11:22pm

Ok, but Brady's supporting cast in 2009 was far more similar to his supporting cast in crazy-good 2007 than in any other year of his career. Throwing 23-28 TDs to Caldwell and Gaffney ought to suggest more than 28 TDs to Moss and Welker.

by tuluse :: Sun, 03/14/2010 - 5:04am

What's your point? This article is about how well the crowd could predict stats, and they predicted well. You seem to be arguing that they shouldn't have done so well, but it's irrelevant because they did.

by sundown (not verified) :: Wed, 03/17/2010 - 10:22pm

No, his point is that Bill put the wrong numbers down which pretty much invalidated his comments. He took the crowd to task for being off by way more than they really were.