Seahawks Insider

5 questions with Football Outsiders

Post by Eric Williams on July 31, 2012 at 6:45 am with 16 Comments »
July 31, 2012 6:45 am

Aaron Schatz, the creator of Football Outsiders, and Brian McIntyre, a salary cap specialist who works for, were both gracious enough to answer specific questions about the Seattle Seahawks prospects in the upcoming season.

Football Outsiders is a stat-based web site dedicated to providing football-specific information on the NFL. Seahawks general manager John Schneider is one of the many people in the industry who keep up with the detailed analysis Football Outsiders provides, and uses their information in their daily work.

Football Outsiders just published their annual Football Almanac, which provides a wealth of detailed breakdowns like personnel trends and running success rates on each team in the league. If you haven’t picked up one up already, it’s recommended reading in order to prepare for the upcoming season. You can find out more information on how to purchase this year’s almanac here in PDF form or in print. It’s also available at

McIntyre, who you probably remember from the excellent weekly personnel reports and contract analysis he provided for us over the years, wrote this year’s Seahawks chapter for the almanac.

The following are five questions I asked after reading over this year’s chapter

Q: According to this year’s Almanac, the Seahawks’ mean win projection is 7.2 wins, which I believe is the same as San Francisco. First off, how did you arrive at the number? And secondly, how did a team that finished 13-3 last season in San Francisco end up with such a low number. And lastly, based on your projection what will it take for Seattle to surpass the seven-win mark and compete for a playoff spot this season?

Aaron Schatz: The mean win projections are based on a complicated process. It starts with an equation that projects each team’s quality for the season based on a number of different variables – experience, offseason roster changes, different splits and various specific stats from the past two seasons.

We then simulate a number of different possible results from this equation, and then use the resulting ratings to play the season over and over. In the end, what we have are one million sample seasons that give us an idea of what we can expect from each team this year – what’s likely, and what’s unlikely.

I’ve written a lot about the San Francisco projection this preseason, since that’s the most controversial one we’ve made. Let me see if I can run down the issues quickly.

– The Plexiglass Principle: Teams that make a dramatic improvement tend to decline in year three, while teams that make a dramatic fall tend to improve in year three. Since 1978, eleven teams have improved by seven or more wins from one season to the next. Not counting the 49ers, the ten other teams declined by an average of 4.7 wins the following year.

– Offense is more consistent than defense from season to season, and defense is more consistent than special teams. The 49ers’ success was primarily built on defense and special teams, which are more likely to regress towards the mean than a similarly strong offensive team (say, the Patriots).

– The 49ers were superbly healthy, especially on defense, and that’s unlikely to be the case this year; you have to assume that teams will generally have an average number of injuries each year.

– Alex Smith doesn’t have a track record of being an above-average quarterback. He’s been in the league seven seasons, and last year was the first time he was above average, so it is hard to believe he’ll be above average again. (Objectively, that is; subjectively, we’re big believers in Jim Harbaugh’s coaching skills.)

As far as Seattle surpassing the seven-win mark, it could be anything – the range of results in the simulation is supposed to represent any number of possibilities. I think the most likely reasons for the Seahawks taking a big step forward would be: 1) Matt Flynn or Russell Wilson is better than expected as the starting quarterback; 2) Bruce Irvin has an Aldon Smith-like rookie year; 3) the non-Patriots of the AFC East are not as good as we are projecting, giving all the NFC West teams another win or two.

Q: With Chris Clemons recently signing a contract extension with the Seahawks, I thought the statistic of the Seattle defensive end dropping more in zone blitz coverage than any other defensive end other than John Abraham was an interesting nugget. How effective was the Seattle’s defense in those situations? And secondly, what are your projections for rookie defensive end Bruce Irvin in his first year?

Brian McIntyre: On those plays (36 in total), the Seahawks posted four sacks and an interception. Our game charters noted 12 other plays where opposing quarterbacks were under pressure and/or forced to scramble. Seventeen of 29 attempts resulted in completions, gaining 172 yards, eight first downs and a touchdown. Two additional first downs were gained by opposing quarterbacks on scrambles. Clemons was the nearest defender in coverage on two of those pass plays, and he had one pass deflection and allowed a six-yard completion.

Irvin’s remarkable sack production at West Virginia (22.5 sacks in 26 games) and impressive athleticism should have resulted in a high score on our SackSEER projection model, but the two seasons of eligibility Irvin missed and a lack of passes defensed lowered the projection to 11.1 sacks over five seasons.

Players can (and do) beat the projection. A great example of that is Jason Pierre-Paul, who also missed two seasons of eligibility and had outstanding athleticism. The initial SackSEER model had him for 3.8 sacks, a projection he beat in Year One (4.5 sacks as a rookie) and obliterated with 16.5 sacks in 2011.

As Seahawks fans know, 11 sacks is the same number that Clemons produced in each of his two seasons in Pete Carroll’s defensive system. Raheem Brock came close to double-digit sacks in this defense (nine in 2010) and Irvin is expected to play a similar role and similar number of snaps that Brock has the last two seasons.

Q: According to the almanac, the Seahawks had the third least healthy offense in the NFL last season, with a 53.4 adjusted games lost. And you also note that Tarvaris Jackson took 42 sacks last season, second only to Alex Smith. According to your charting, 17 of Jackson’s sacks were “long sacks”, where he held onto the ball for at least three seconds.

My question is, if Seattle names Jackson the starter, do you believe those numbers will improve with a healthy offensive line? Or do you believe that Jackson will continue to hold onto the ball too long? And how does a healthy Sidney Rice factor into that?

Brian McIntyre: If Tarvaris Jackson is named the starter, there is no question that a healthy and improved offensive line, and a healthy Sidney Rice, could reduce the overall number of sacks that Jackson takes in 2012. A healthy and improved line would be less dependent on Zach Miller in pass protection, which would give Jackson a nice safety valve in the passing game.

That said, Jackson is really good at following Pete Carroll’s mantra about ball security, which makes him susceptible to the “long sack.” Also, in very limited action with the Vikings in 2010, Jackson was sacked six times in 66 pass plays and had an average sack time of 2.7 seconds.

Q: You note that Seattle had dramatic splits in their performance before and after halftime. According to the Almanac, on offense the Seahawks went from 30th in DVOA before halftime to eighth after halftime. On defense, they went from 25th in DVOA before halftime to the best defense in the league after halftime.

First, can you explain what DVOA is, and what it measures? Secondly, I know you had a tough time figuring out the reason for the splits. I’ll offer a couple. Carroll is known for making pretty good halftime adjustments, particularly on defense, after seeing how a team is attacking him. Also, the Seahawks were outscored 103-58 in the second quarter in 2011, so usually they had to open it up offensively after starting the game conservatively. Lastly, I know the Seahawks finished +8 in turnover differential, good enough for fifth in the league. Perhaps a majority of those turnovers occurred in the second half, giving a bump to both sides of the ball?

Brian McIntyre: DVOA stands for Defense-adjusted Value Over Average. Every play is scored for its success based on the down and distance, then that success rating is compared to the league average adjusted for situation and opponent. It’s our main metric for both players and teams.

The Carroll halftime adjustments theory is an interesting one, but it doesn’t seem to be backed up by history. The 2010 Seahawks were 30th in defensive DVOA before halftime, 27th afterwards, not much of an improvement. The 1999 Patriots were 15th/16th. The 1998 Patriots did have a similar split: 18th before halftime, fifth afterwards. But the 1997 Patriots were the exact opposite: second in defense before halftime, 28th in the league after halftime.

You are definitely right about the turnovers though, especially on defense. The Seahawks had five picks in the first half of games, 17 in the second half. Again, we have no idea if this is a real meaningful trend or just random noise in the numbers.

Q: According to the Almanac, Seattle allowed 6.8 yards per carry and a league-worst 33.6% DVOA on handoffs to running backs out of the shotgun. That’s an interesting stat. What personnel grouping was on the field the most in those situations? And it appears that Seattle tried to upgrade that group by drafting Bruce Irvin, Bobby Wagner and Korey Toomer, along with signing defensive tackle Jason Jones. How much will those changes help Seattle’s nickel package?

Brian McIntyre: On 80 percent of those running plays (handoffs to running backs) out of shotgun formations, the Seahawks were in a nickel defensive (4-2-5). Irvin and Jones were brought in to get after the quarterback and overaggressive attempts to get to the passer could leave them prone to giving up big plays on draws.

As for the linebacker position, former linebacker David Hawthorne made over 25 percent of the tackles on these running plays. Whether it’s K.J. Wright, Wagner, Toomer or Malcolm Smith, the “Mike” in Seattle’s nickel packages may not have the experience to sniff out a draw play.

Leave a comment Comments → 16
  1. JMSeaTown says:

    Here’s to how much preseason projections change after week one. It’s why we love sports, fantasy football, and life in general. You can plan a pretty picnic, but you can’t predict the weather. (Especially here in southeastern Alaska)

  2. GeorgiaHawk says:

    I didn’t see The 49ers as a clearly dominant team over the Seahawks last year. I think the Seahawks just shot themselves in the foot with poor special teams play and poor qb play.
    That’s why I think Carroll and crew drafted several players this year, ( with good speed ) that can also excel on special teams if needed.

    And of course that’s why they upgraded the qb position. They knew what they needed to do to knock off the 49ers and they went out and got the players that can help make it happen.

    That’s why I think we will have more wins this year then the 49ers, provided that we don’t have a rash of injuries and cut down on silly mistakes.

  3. As if we don’t pay enough for the NFL as it is. Oh well data miners are longtail markets too.

  4. bbnate420 says:

    Good to see you back, Georgia. Seems you’ve been MIA.

    I think the points that SF is unlikely to be as healthy as in 2011 and to have as good of a turnover ratio are right on point. Time will tell if the Hawks are ready 2 take advantage.

  5. Hammajamma says:

    Thanks for this, Eric. Interesting stuff.

    This really clarifies the argument for SF falling back this year. They were, in fact, very fortunate with injuries while the Hawks were typically hammered. Most interesting to me is the sustainable Offense v. Defense analysis. You’d have thought it to be the other way around.

  6. Ahhh… Meaty! Thanks, Eric.

  7. Dukeshire says:

    “You can plan a pretty picnic, but you can’t predict the weather.” Well said and exactly why I put next to no stock into statistical predictions, regarding sports. For me, FO is a great source to use enhance scheme breakdowns and help illuminate what has happened on the field. But as a tool to look season(s) forward, they do very little for me.

    It’s definitely an interesting read, regardless.

  8. A couple of things stood out to me. “On defense, Seahawks went from 25th in DVOA before halftime to the best defense in the league after halftime.” … and… “Seahawks had five picks in the first half of games, 17 in the second half.”

    Random noise in the stats? I think not. Eric points to Carroll’s reputation for making good halftime adjustments on defense after seeing how a team is attacking him. I don’t think that can be underestimated. Carroll had to rely on some of the least experienced defenders in the league (two 2nd-year safeties, rookie CB and Sam). If it weren’t for great coaching adjustments I have to believe those numbers would have been reversed – with offenses shredding the youngsters in the second half once OCs had their scouts’ info the young players’ weaknesses. Didn’t happen.

    That says to me our young players were surprisingly coachable and that the coaching and halftime adjustments on D were great. And this also gives me some faith that Wagner can be coached up quickly to fill the MLB role.

    … off to camp… feels like spring football out there, this is July?

  9. Hammajamma says:

    I think I’d call this a probability model more than a Win/Loss prediction. There are always outliers in a data set or statistical run. It’s predicted SF will regress toward the mean, but it’s never assured. Play the season a million times and this is where you have the majority of your results. Any team can flash, develop new players that bring a new variable to a given result.

    If this were a discussion about baseball, Duke, you’d probably have a different take. I agree that football is much harder to handle statistically. Coaching, for one example, is much more important.

    There, I have it both ways.

  10. iHateHarbaugh says:

    Good read, thanks Eric #12thMan

  11. Wow. Excellent stuff Eric. This kind of in depth information is the stuff avid bloggers look for. Thank you

  12. GeorgiaHawk says:

    Eric- You make klm008 proud!

  13. GeorgiaHawk says:

    bbnate420- I haven’t posted much of late because I’ve been watching my friends house and pets on the other side of Atlanta while she is staying in Germany for two weeks.

    She doesn’t have a desk top so the only way for me to post is to use my phone and I am not about to do that! Lol. You would think that I really hit the bottle hard if you saw me post from my phone. Believe me I would make Dukeshire look good when it comes to phone posting.

    Anyways, Ive been able to get back over to this side of town, ( pets are driving me crazy ) from time to time.

  14. GeorgiaRay says:

    Interesting model, but impossible to prove out consistently from a statistical standpoint….I’m interested to see how this model compares with the reality of past seasons modeled…

    It’s all about what happens on Sunday; waaayyyy too many opportunities for missed plays, injuries, mis-reads etc….

  15. GeorgiaRay says:

    Hey GAHawk, Which part of town are you in?

  16. GeorgiaHawk says:

    GeorgiaRay- Canton. However at this time mostly staying in Duluth.

We welcome comments. Please keep them civil, short and to the point. ALL CAPS, spam, obscene, profane, abusive and off topic comments will be deleted. Repeat offenders will be blocked. Thanks for taking part and abiding by these simple rules.

JavaScript is required to post comments.

Follow the comments on this post with RSS 2.0