Five different teams from last year's DVOA top eight rank in the bottom half of the league through four weeks of 2014. What can we learn from other teams with similar starts in the past?
25 Aug 2006
by Aaron Schatz
Time for another trip through the Football Outsiders mailbag. We get a lot of e-mail, and there are a lot of comments on the discussion threads, so I apologize if your question doesn't get answered. There simply are too many good questions that require well thought out answers. The best way to get your question answered at this point is to use the contact form. If it is a question not related to the DVOA stats, it is more likely to be answered if you send it to one of the other writers, not me.
Before we start, I want to apologize once again for the server problems that caused us to miss nearly all our e-mail for three weeks, only to have it all delivered on Wednesday. I can only imagine how frustrating it was for those people who purchased the KUBIAK fantasy projections, had problems downloading updates or recovering their passwords, and then seemingly sent e-mail after e-mail into a black hole. We never meant to ignore you, and we hope we can regain your trust. Suffice it to say, we've done some serious research on our next hosting company to make sure that this doesn't happen again.
So we had hundreds of e-mails on Wednesday, regarding all kinds of topics from technical problems with the KUBIAK download, to questions about specific projections, to questions about articles in Pro Football Prospectus 2006, to random questions like "can you tell me about former Jets quarterback Glenn Foley." I've spent the last two days answering as many as possible, and I'll do some more today, and we've tossed a number of the questions into this mailbag.
Many people have asked about an open Pro Football Prospectus 2006 thread. Feel free to use the comments on this article to discuss anything related to the book or the fantasy projections. If you want an answer from an FO writer, however, you are probably better off using the contact form to e-mail the specific author of the essay related to your question.
This is where I normally say that we reference plenty of our innovative FO stats here, not to mention their unfamiliar terminology, so if you are a recent addition to the readership you might want to read this description of our methods first, and that I know I need to update that page because it hasn't been updated in years. Guess what -- that's no longer true! It's finally updated! Hooray! On to our questions...
Joe Pina: Do you guys have a mock fantasy football draft that you post?
Aaron: Many of the original writers are in an auction league here in Boston and each year Scramble for the Ball analyzes that draft and writes about it. The auction is this Sunday, so I'm guessing the review will show up in next week's Scramble.
Michael: My Fantasy team uses IDP (individual defensive players). How do I change your projections spreadsheet to reflect that?
Aaron: We do not do projections for IDP. That's on the "to do list" for 2007, along with some way to create reasonable auction values. Frankly, I don't want to use the system that I use for my own league, because as you'll find out when Scramble reviews our draft, our league is insane. Last year, I think something like 80 percent of starting wide receivers went for either $1 or $2. It's absurdly running back heavy and so when I show up at the draft, I've tweaked the values to reflect this.
James G.: As I was scanning through the fantasy portion of PFP 2006 this morning, I noticed what I considered early draft positions for Ks and TEs. I assume these are based on VBD (value-based drafting) where the last starter is used as the baseline? I read an article at footballguys.com last year suggesting the baseline should be pick 100. Last year, using my league's draft history, I altered my baseline and had a very successful draft.
So my question: shouldn't players be rated vs. a replacement level rather than the last starter? If you rate them against last starter, kickers move up in value. But this fails to account for the lack of difference between the last K starter and the huge pot of undrafted kickers. Therefore, people wait longer to draft the late kickers, meaning you should also wait on the early kickers.
The same thing is true for tight ends, and I think if you change the baseline to replacement instead of last starter, you will even drop the draft position of premier tight ends like Antonio Gates. Instead of being the 13th rated player, as I think he is in the PFP rankings, he'd be more around 20th.
Aaron: The short answer is "yes." Sean McCall (designer of the KUBIAK macros) and I were already thinking about this when I saw your e-mail. We think we've come up with a good compromise solution. In the latest version of KUBIAK, there is an option to have the baseline for kicker and defense based on only half the number of teams in your league, rather than the baseline being based on all the starting kickers and defenses. This will move kickers and defenses much more into line with where they are generally taken in drafts.
Unfortunately, we were unable to do an easy fix with tight ends, because the tight end code is more complicated, related to the option to have TE count as WR in some leagues. I don't mind having tight ends appear higher in our rankings than conventional wisdom: the top tight ends are much more consistent than the top kickers and defenses, and the difference between Antonio Gates and a replacement tight end is bigger than the difference between the Bears and a replacement defense. But if you want to put our tight end values more in line with conventional wisdom, try this: on the Baselines page, change the baseline number for tight ends from the total number of starters in your league to 67% the total number of starters in your league. Then re-sort by Fantasy Points Over Baseline (FPOB). Don't use 50% the way we did for kickers and defenses -- that will actually make tight ends show up too low compared to conventional wisdom.
Pepper: WOW! How could you rate Mushin Muhammad so high after you RIP him HARD in your WR Profile section?!?!? 17th OVERALL among WR's!!!??? NO FRICKIN WAY!!!!
Roddy White?!?! 923yds? 6Tds? Biiiiig Stretch!!! VICK hasn't thrown for 18 TD's in his career, much less 3000 yds passing as you are projecting. He hasn't played a full 16 game season since he came in the league!
Another MYSTERY in your Projectionsâ€¦DEION BRANCH RANKED 11th among WR's?!?!!? The man is holding OUT if you haven't noticed.
So you would recommend based on your projections that I take White/Muhammad AND Branch beforeâ€¦
aaaaah, no thanks.
Guess that is why you give me the power to change the projections myself in the spreadsheetâ€¦and I THANK YOU for that. LOL. My draft is Friday evening, I'll let you know when your â€˜ELITE 3' went in the draft.
Aaron: Um, where I come from, "Elite 3" refers to Larry Johnson, LaDainian Tomlinson, and Shaun Alexander.
Despite this gentleman's, shall we say, pointed commentary, I do think it is reasonable to ask me about the reasoning behind the KUBIAK projections. Regarding wide receivers, people need to understand that wide receivers vary from year to year more than quarterbacks and running backs. Therefore our projections for receivers are bunched much closer together. The numbers are accounting for regression to the mean and the possibility that a guy will drop off like Joe Horn or Michael Clayton did last year. We are considering "forcing" the numbers higher next year so the projections look more like the actual year-end numbers. Doing so will cause the projections to be accurate on more of the top players, but they will be far more inaccurate on the players that we project incorrectly. And the order of the projections won't be any different, so I'm not sure how much it would really help anyone.
That out of the way, let's look at these players individually.
Muhsin Muhammad: Muhammad is addressed in the "Fantasy Risers and Fallers" chapter of PFP 2006. Chicago has a very easy schedule of opposing defenses this year. They will have a quarterback not named Kyle Orton. A huge drop in yardage like Muhammad had last year is usually followed by a little bit of rebound, and the top receiver on a team -- even a team with a bad offense -- usually ends up with more than four touchdowns.
Roddy White: The whole issue of the Atlanta offense and Vick's passing numbers is also addressed in the book. There are a lot of reasons to expect a little bit of improvement in the Atlanta passing game, and White and Michael Jenkins are two young receivers with breakout potential. It really comes down to which one is the number one option. For a while, we thought that was Michael Jenkins, then so many people told us it was Roddy White that we projected White as the top guy, Jenkins as the number two. Either way, there's reason to believe somebody is going to break out here. Vick threw 14 touchdowns in 2004 and 15 touchdowns in 2005, each in 15 games. Is it that hard to believe he could throw 18 touchdowns with another year of experience and a full season? And by the way, isn't an injury to Vick more likely to help White's receiving numbers, because Matt Schaub is a pure passer?
Deion Branch: Yes, Deion Branch is holding out. Everyone I have spoken to believes he is going to be back with some sort of agreement before the season starts. (The headline yesterday on ProFootballTalk.com was "Branch Caving Before September 10.") We've already reduced his projection slightly due to the holdout, but this guy is the number one receiver on one of the most powerful offenses in the league. You can also read this article by FO's Bill Barnwell on how Branch's 2003-2005 numbers are similar to those of a number of historical receivers who had big seasons the following year.
Now, let's talk about the players that Pepper thinks we don't want you to take.
Hines Ward: Ward is projected for only 40 fewer yards than last year. He's projected for only six touchdowns, compared to 11 last year. Then again, he had only four in 2005. Frankly, he could have four, six, or 11 this year. Touchdowns are much harder to project than yards and receptions.
T.J. Whosyourdaddy: His projection is 30 yards lower, and half a touchdown greater, than the average of his 2004 and 2005 seasons.
Derrick Mason: Remember, Mason is now 32 years old. Mason is projected for fewer yards than last year, but more touchdowns.
Javon Walker: Walker is a number two receiver coming off an ACL injury. We don't think there's anything wrong with a little caution.
Anquan Boldin: This one needs a little more discussion, because we've received a lot of questions about why our projections are so low for Boldin, Edgerrin James, and Neil Rackers.
Let's start with the fact that the Arizona Cardinals had 670 pass attempts last year, not counting sacks. Only two other teams in the entire league had more than 600: Green Bay (626) and Philadelphia (621). Not a single team had 600 pass attempts in 2004, and only one team (New York) had this many in 2003. Only five teams in history had more pass attempts than the 2005 Arizona Cardinals. The Cardinals threw all the time because their running game was horrible. Because they didn't want to try running, and were always losing anyway, the Cardinals had only 360 runs last year, which ranked last in the league.
In 2004, with the same coach and a similar losing record, the Cardinals threw just 533 passes and ran the ball 475 times. Which is more likely: That Arizona's pass-run split in 2006 will be as extreme as it was in 2005, or that the Cardinals will revert at least partway to the pass-run split they showed in 2004? Remember that this team just went out and spent huge free agent dollars on the running back who finished second in the NFL in total carries last year.
Wait, there's more. Let's assume that Edge not only leads to more Arizona running plays, but to more Arizona running touchdowns. After all, they only had two last year. All those drives that are now ending in rushing touchdowns used to end in something else: usually, it was either a passing touchdown or a field goal attempt. So again, something has to give. Either Rackers has fewer field goal attempts, or Boldin and Fitzgerald have fewer receiving touchdowns. I suppose that Arizona could turn into the 1998 Vikings, but if you've seen their offensive line in the preseason, you'll understand why I have my doubts.
Edgerrin James, Larry Fitzgerald, Anquan Boldin, and Neil Rackers cannot possibly all enjoy similar numbers to 2005. Something has to give. When we ran our numbers, the answer came out "everything but Fitzgerald." We actually used our subjective "expected role" variable to mark Fitzgerald and Boldin as equal co-number one receivers, but for various reasons, KUBIAK likes Fitzgerald more. The guy had 1,400 yards at the age of 22. I don't think people realize how amazing that is. Only two receivers in NFL history have gained more than 1,250 receiving yards before age 23: Fitzgerald and Randy Moss, who actually did it twice (he was 21 as a rookie). Other than Fitzgerald and Moss, the only 22-year-old with more than eight receiving touchdowns was John Jefferson with the 1978 Chargers.
One last note: Given a standard scoring structure, the current projections have Muhammad, White, Ward, Houshmandzadeh, and Boldin all within a narrow range of 127-132 fantasy points. (Donald Driver is also in this range.) Given the variation inherent in NFL stats, we're not saying that Muhammad and White are guaranteed to be better than Ward and Boldin this year. These players are basically tied. Mason and Walker are only a couple points lower. We don't intend for our projections to turn you into a robot. When player projections are that close, take the one you personally think has the best chance for a big year. Just don't take someone like, say, Joe Horn, whose projection is 20 points lower, a much more significant gap.
Weo: I'm watching the Rams / Colts pre season game (boring!) and noticed the top banner â€“ which reads, left to right: NFL FOX, IND XX, STL XX, time remaining. A symbol that looks like the #4 die, and (sometimes) down and distance.
My point? PLEASE add the QUARTER information (as in which quarter are we in?) to that banner!
Aaron: Believe it or not, that symbol that looks like a "4" on a six-sided die is actually the quarter marker. You'll see the number of yellow dots equals the quarter. It's horribly non-intuitive and an example of how broadcasters choose "style" over actually having understandable information. But it's nowhere near as bad as the ESPN Arkanoid score box.
David Brude: I was just curious what the fumble rate is for QB's on pass plays on sacks vs. designed QB runs or scrambles.
Aaron: Sure, David. That's a fairly easy question.
In 2005, there were 912 quarterback runs, not counting aborted snaps/handoffs. Aborted snaps/handoffs are considered runs by the league, but we split them into runs and passes depending on the type of play where there was an abort. 22 runs were fumbled, or 2.4% of quarterback runs. There were also 12 aborted snaps which we list as runs, i.e. aborted handoffs to running backs where the QB was charged with the fumble.
There were 1182 sacks in 2005, and 246 of those were fumbled. That's 20.8% of sacks.
Also note that 45% of the scramble/run fumbles were recovered by the defense, while 55% of the sack fumbles were recovered by the defense.
Jim McCombs: Are you going to show up and eat crow when Brooks turns out to be one of the best QBs in the league??? Unlike you, I have actually watched Brooks play. In case you are too busy or stupid to do a COMPLETE analysis of Brooks and the Saints, let me elaborate. He kept one of the worst teams in the NFL competative almost singlehandedly. Since arriving at the starting QB position, he has lost top notch offensive linemen almost every year, only to see these linemen replaced with second rate players. Stallworth has yet to live up to the hype and is on the trading block as we speak. The tight end position has been a bust and Deuce, when not hurt has been inconsistant. He would break the occasional long run, but could not be counted on to get the 3-4 yards when thay were desperately needed. Now for the defense. They haven\'t stopped a runner in 6 years. They have been in the bottom 5 teams in the NFL in scoring defense 4 of the last 5 years. In 04 they were sooooo bad that even after putting together a good 4 game run to finish the year 8-8, the defense was STILL #32 in the league!!!! In one game in 04 (Vikes) Brooks scored 24 points and scored 4 of the last 5 times he had the ball. On the other hand, the defense did not have a single stop after the mid-point of the 2nd qtr. Minn scored every time they had the ball except the last time. Time ran out before they could score. The fact is that Brooks had 18 4th qtr comebacks out of a total of 38 victories. Sounds like Brooks was capable of a lot of GREAT decisions while playing for the Saints.
Aaron: That's the Unprepared for Imminent Disappointment Hate Mail of the Week.
Nancy Poon: I purchased the KUBIAK system and was disappointed that there wasn't a projected 2006 run defense and pass defense rating for each team, like you did in the 2005 defensive rankings. Those projections of the defensive strengths players would be facing is what I was looking for, so I can determine who to draft, and who are facing the easiest respective defenses in my playoffs.
Can you please provide me to the link where I can find this, or which page in my Pro Football Prospectus? The general projected defensive rank doesn't really help, or the projected defensive points don't either. I am looking for some place that does it like you have in the link above, where a team's run defense is -15.7% and their pass defense is 5.9%
Also, I noticed there are no entries for Missed Time for any of the other players, except quarterback. Why is that?
Aaron: The reason there are no 2006 projected run defense and pass defense numbers for each team is that there are no projected run defense and pass defense numbers for each team anywhere. We simply don't project those separately. We only project total defense, which considers both. We could work on something to separate them for 2007, but unfortunately there is not time for the upcoming season.
What I can provide is a list of projected schedule strength split into offense and defense, and a list of projected schedule split into run defense and pass defense based on the 2005 DVOA ratings.
First of all, here is a list of schedule strength based on projected defense of all 16 opponents in DVOA. I'm listing two numbers here, mean and median. They can be very different: for example, Atlanta plays a lot of teams projected with moderately good defenses, and then two games against New Orleans basically throws the whole mean out of whack. I'm listing the teams in order of mean from hardest projected schedule of opposing defenses to easiest projected schedule.
Next, let's rank all 32 teams based on the 2005 DVOA of their 16 2006 opponents, split into run and pass. I'm using WEIGHTED DVOA here, giving more impact to games at the end of the season. Because run defense became stronger at the end of last season -- I need to study if this is something that happens every year, I'm not sure -- the run numbers are almost all negative, so you are better off looking at relative rank. What we have here is average opponent defense, ranked from hardest schedule to easiest schedule. I'll rank in order of run since running backs are the most important players in fantasy football.
Maybe Kevin Jones will bounce back after all. And yes, that really does say that Indianapolis will face the easiest schedule of opposing pass defenses, based on last year's weighted DVOA.
To answer your last question, about why the Missed Time variable only exists for quarterbacks: This is explained in the book. The reason I created the Missed Time variable is that it is silly to try to predict stats for backup quarterbacks. Unlike other positions, starting quarterbacks take every snap until they are injured or lose their jobs, and so you end up predicting every backup with something like 10 attempts. But at RB and WR, backups play throughout the season, so it's slightly easier to make predictions of playing time.
Anyway, because of this, we projected all QB for 16 games so we could get a more accurate idea of how backups would do if put into the starting role. However, this created a problem for the fantasy projections, because we can't rank Carson Palmer as if he's going to play 16 games when there's a damn good chance he won't play 16 games. Thus, the Missed Time variable was used to adjust QB rankings based on the likelihood, in our estimation, that they will miss time this year. By the way, the latest update to the sheet adds an option to remove the Missed Time variable and judge quarterbacks on projected stats over 16 games instead.
I do hope to expand these more objective playing time projections to other positions next year, but they're somewhat precarious as is and hard to figure out. No promises.
Ryan McKeon: Just received my copy of PFP 2006. A good read as expected. I was intrigued by the analysis of how historically strong the AFC West was last year. I had two questions: which are the weakest divisions historically? Is a weak division more/less likely to produce the Super Bowl participant for that conference?
Aaron: Hey, Mike Tanier wrote that article in the book so let's send this question over to him.
Mike Tanier: After reading your question, I may research the Weakest Divisions Ever and write up the results as a Too Deep Zone article. Just some thoughts after looking through the encyclopedias â€¦
Last year's NFC West must be among the weakest divisions of the past decade. I mean that as no disrespect to the Seahawks, but they got to play one of the worst teams in recent memory, a team whose head coach went on medical leave in midseason, and your basic 6-10 Cardinals squad. That's pretty easy pickings.
When we think of bad divisions, we often think of the old NFC West, when the Niners would go 14-2 beating up on the Rams, Saints, Falcons and later the Panthers. But through most of the late 80s to mid 90s, at least one of the also-rans usually finished 8-8 or 9-7. Take 1990, when the Niners finished 14-2 in a four-team division. The Saints were actually 8-8 that year and had the best linebacking corps in the league. Of course, in most years the other teams were so dreadful that a merely bad team could go 5-1 against them and pump up its record. Those Steve Walsh-Craig Heyward Saints weren't that good, but they went 4-2 in the division, beating a Niners team that already clinched the playoffs.
But what makes a "weak" division? Usually, we think of a powerhouse team and three or four patsies. There's also the case of four or five weak teams duking it out and going .500 by default. We tend to think of those teams as being battle tested when they are actually just mediocre. Take a look at the 1978 NFC Central, the fabled Black 'n' Blue division. The Packers and Vikings finished 8-7-1, the Bears and Lions were 7-9, the Bucs 5-11, their best season to date. The Vikings made the playoffs and lost 34-10 to the Rams. Fran Tarkenton threw 32 interceptions for that team, and they averaged just over 3 yards per rush in a season when offense was on the rise. They were the 35-year-old Purple People Eaters, and they allowed 12 more points than they scored. The other QBs in the division were David Whitehurst, Bob Avellini, Gary Danielson, and Doug Williams with a completion percentage under 40. This was a miserable division, but two teams finished over .500 because they played each other eight times.
Anyway, I hope to study this more in the near future.
36 comments, Last at 18 Sep 2006, 6:25pm by zlionsfan