Writers of Pro Football Prospectus 2008

21 Feb 2011

How Good Was Oakland's Offense?

So, in the column announcing the winners of the FO reader awards, I noted that people voted for Hue Jackson as coordinator of the year "simply for not sucking in Oakland." One of the readers pointed out that by official numbers, the Raiders were a lot better than just "not sucking." Believe it or not, the Raiders ranked sixth in the NFL in points scored this year. I honestly had no idea. I've been going through the whole season thinking of them in terms of DVOA, where they ranked 23rd.

I didn't go back to check, but I've got to figure this is one of the biggest gaps between DVOA and points scored over the past two decades. 17 places in the standings is a BIG difference. So what is DVOA seeing that a simple look at points scored doesn't show?

1) Oakland had an easy schedule, 23rd in the NFL by average DVOA of opponent.

2) Oakland had a great fumble recovery record, 15 of 23 fumbles on offense.

OK, but even without adjustments for these two things, Oakland would have 1.1% offensive DVOA. That would be 19th instead of 23rd. There is still a lot of room between that and sixth. What else?

3) Defensive touchdowns? Actually no, Oakland had just three, which is pretty average.

4) Oakland ran a lot of drives, because the defense was very good at forcing three-and-outs. The Raiders had 199 offensive drives, third in the league.

5) Oakland had a lot of very long plays. As you may know, in the DVOA formula, the value of each additional yard starts to drop after about 20 yards, because longer touchdowns are only partly about making better plays. They're partly about field position, because you can't run for a 60-yard touchdown when you are past midfield.

The Raiders had 11 plays of 50+ yards, 19 plays of 40+ yards, and 35 plays of 30+ yards. In all three categories, they were second to the Philadelphia Eagles.

6) Some of what ended up as points is actually field position from defense and special teams. The Raiders ranked higher in defensive DVOA (15th) than in points allowed (20th).

Of course, it's still hard to believe that all this is still enough to get us from 23rd in DVOA to sixth in point scored. Is it possible our formula mistakenly underrated the 2010 Raiders? Perhaps. On the other hand, I doubt that Jacoby Ford is going 71 yards on an end around again in 2011.

Posted by: Aaron Schatz on 21 Feb 2011

40 comments, Last at 30 Mar 2011, 5:33pm by jimbohead


by dmstorm22 :: Mon, 02/21/2011 - 7:19pm

Sebastian Janikowski going 12-19 from 40+ probably didn't hurt turn average drives into 3 points. He also hit a league high 33 field goals, again, drives that might not look great on DVOA because they ended in field goals were turned into points a high percentage of the time.

That said, I do think there was some underrating of the Raiders offense. They definitely did get a whole lot of big plays, but after a point, that is predictive isn't it?

Also, the never ending "the Raiders suck" routine gets tiresome when they are a talented team, and just went 8-8. There are a whole lot of other teams that deserve many more taunts and jests.

by Key19 :: Mon, 02/21/2011 - 7:22pm

It certainly helps when you get 98 points in two games against the Broncos...

Based on simply points scored, the story seems to be that Oakland simply tore up the AFC West (35, 59, 23, 28, 39, and 31 points scored in AFCW games). Their schedule also included Houston and Jacksonville, two absolutely abysmal defenses (24 and 31 points scored respectively in those two games). To cap it off, they also hit teams right when they were in funks (mid-season somewhat injury-ridden Seahawks - 33, late-season injury-ridden Colts - 26). That's 10 games right there, with three more games against the NFC West that I'm not even taking into account (although they did struggle against the 49ers). In their remaining three games against somewhat good-to-great defenses (Titans, Steelers, Dolphins), they scored 13, 3, and 17 points respectively.

I would estimate that the true value of the Raiders offense lies near the middle of the gap between DVOA and Points Scored. They were able to put up big numbers against easy competition, which shows they at least have some good ability. But they did struggle in their few games where the opponent was very good, so they're definitely not an elite offense (certainly not 6th-worthy).

by tunesmith :: Mon, 02/21/2011 - 7:36pm

Yeah, I read that in that 59-point win over Denver, Denver had played their guts out almost beating the Jets the week before, and then McDaniels didn't give them a break and just basically tore into the team all week afterwards, so Denver came out flat against Oakland. Also for Denver, Martindale's schemes were apparently problematic, which led to success for both San Diego and Oakland the second time they played Denver.

by justanothersteve :: Tue, 02/22/2011 - 12:07am

In the annual awards, wasn't McDaniels not included because of poor personnel decisions and not coaching? This seems to indicate otherwise.

by tunesmith :: Tue, 02/22/2011 - 5:15am

Yeah, that was definitely a subjective decision not to include him in the poll. What's probably the reality is that McD was a great coach in terms of scheme and preparation, but a poor coach in terms of keeping the locker room... you know, actual *coaching*.

by jimbohead :: Tue, 02/22/2011 - 3:14pm

Agreed. At this point, he's a fantastic coordinator without the ability to be the main guy. I hope for his sake that he uses his denver experience as a wake-up call to develop some humility and some people skills. He's too young to simply accept a permanent cap on his career.

by tgt2 (not verified) :: Thu, 03/03/2011 - 3:35pm

What does humility have to do with anything that was discussed? If he wants to be a coach, he needs communication skills and motivational skills, not humility.

by jimbohead :: Wed, 03/30/2011 - 5:33pm

Humility affects these things critically. It's hard to communicate with people effectively when you have a stick up your butt about how great you are, and an unbearable need to prove how tough a man, and how great a genius that you are. Pride is what causes a man to get rid of a DC like Mike Nolan, what causes "miscommunications" with Cutler (though that didn't really burn him that bad, considering what he got out of Orton), and what causes a guy to lose his team.

I'm guessing the misunderstanding between you and I, tgt2, is about the nature of humility. I don't mean that he should walk around saying "I'm no good." I'm saying he should make an effort to understand the perspectives of others, recognize them as valuable, and use them to relate, communicate and motivate. We cool? =)

by sundown (not verified) :: Wed, 02/23/2011 - 7:08pm

McD wasn't "great" at anything, scheme or anything else. They're not switching the D's alignment he put in place because it was a great move.

by Theo :: Mon, 02/21/2011 - 7:42pm

Raiderjoe already knew this.

by dafrk3in :: Mon, 02/21/2011 - 8:15pm

Look at their drive stats. 19th in pts per drive, 20th in drive success rate. Combine this with their soft schedule, and their offensive dvoa rating makes a lot of sense

by nat :: Tue, 02/22/2011 - 10:00am

To paraphrase the original article: Oakland's non-adjusted for opponent and situation all-inclusive cumulative scoring statistics don't agree with their opponent and situation adjusted predictive-events-only per-play-success rate statistics . What could possibly be the reason?

Opponents. Situations. Non-predictive events leading to scores. Lots of drives to accumulate points on. Oh, and did I mention random statistical variation? Just about the only word in my paraphrase that doesn't explain the difference is "Oakland".

by Bobman :: Thu, 02/24/2011 - 2:23am

Like Kevin Nealon once said in a Saturday Night Live commercial spoof for a bank that only made change (3 quarters, 2 dimes and a nickel for a dollar), "How do we make money doing this? Volume."

So their O is not a high-efficiency machine. You can say that that unit on its own is bad. But when combined with their D and ST, which provided many drive opportunities and good field position, as a whole, they were about average (see Parcells, Bill, "You are what your record says you are.").

They are, as a team and an offense, the anti-Colts. One team emphasizes running, stout D, and has black as a major uni color, the other does not. One has a high number of inefficient possessions, one does not. One consistently wins its division, one does not. But hey, both drafted Louisiana icon QBs #1... And there is that Zach Crockett/Dom Rhodes connection.

by Podge (not verified) :: Tue, 02/22/2011 - 7:33am

Sorry, this turned into a bit of a long rambling post that I can't be bothered to edit!

I think they just pass the eyeball test quite well, particularly in retrospect. I voted for Jackson, because I thought he had put a good improvement into a bad team, while LeBeau and Capers had kept an elite team at an elite level.

When I looked at the nominees, I thought back to watching Oakland's offense, and remembered big plays by Miller, McFadden, Bush, Ford and occassionally, hilariously, "full back" Reese and thought "yeah, they had at least a decent offense, instead of previously being terrible". You forget when you're thinking of one of those guys making a big play that its against a bad D, and you forget all the plays that were no gain, or losses. And I remember DHB catching at least one or two passes, which has to give Jackson some credit...

I guess I also forgot that the passing game was poor enough that Campbell and Gradkowski were both sorta kinda benched at one point or another, weren't they (IIRC, Campbell was hurt and not great, which upgraded hurt to injured so Gradkowski could come in, then the same happened with him)?

Maybe DVOA underrating the running game? Or people overrating it? I'm curious what sort of strength rushing DVOA has in overall offensive DVOA, compared to passing DVOA? 6th in rushing, 25th in passing leads to 23rd overall. I admit when I think of Oaklands O being good I think of the rushing game being good, not the passing game.

I think this might be one of those throwing the baby out with the bathwater things. Maybe rushing DVOA should have more weight in the offensive DVOA formula, but what's the correlation between rushing DVOA and winning compared to passing DVOA and winning (both in individual games and in the season as a whole)? I suspect passing is much more strongly correlated to winning.

Both Bush and McFadden had good DVOA but relatively average success rate, so maybe there's something there?

I think that maybe DVOA does underrate Oakland's offense a little bit, but I'm not sure that means the formula needs changing, as fixing it for Oakland could easily make it less accurate for other teams who have actual quarterbacks.

Another thing, on the awards, I know this might make more work for you, but would it be useful on "select from these nominees" ones to put a couple of stats after each one to support them? So Jackson would have been something like "Hue Jackson, OC, Oakland (#6 Rushing DVOA, #6 points scored)" or something?

by tuluse :: Tue, 02/22/2011 - 10:40am

I don't believe there is any weighting to to rushing and passing dvoa to create total offensive dvoa. It's based on what plays the offense actually called.

by Fan in Exile :: Tue, 02/22/2011 - 9:42am

As a Bronco's fan who needs hope this made me feel better. It looks at least to me like they promoted the wrong coordinator. I also like that San Diego lost Rivera, and that Weis is gone as well. Maybe we have hope.

by Mr Shush :: Tue, 02/22/2011 - 10:34am

"5) Oakland had a lot of very long plays. As you may know, in the DVOA formula, the value of each additional yard starts to drop after about 20 yards, because longer touchdowns are only partly about making better plays. They're partly about field position, because you can't run for a 60-yard touchdown when you are past midfield.

The Raiders had 11 plays of 50+ yards, 19 plays of 40+ yards, and 35 plays of 30+ yards. In all three categories, they were second to the Philadelphia Eagles."

The Raiders notoriously have a lot of guys on offense whose primary attribute is straight line speed. It seems plausible to me that there were a fair number of plays on which Raiders scored touchdowns where other players might have been run down for a big gain short of a score.

by swami :: Tue, 02/22/2011 - 11:07am

I wonder about the impact of discounting long plays in DVOA. Those yards did exist and were still valuable. Perhaps it's an area for DVOA debate/improvement. Would be consistent with the persistent over-valuing of the dink-and-dunk west-coast teams.

by Aaron Brooks' Good Twin (not verified) :: Tue, 02/22/2011 - 11:16am

Look at DVOA's ratings of Barry Sanders and Emmitt Smith to see that DVOA is the statistical equivalent of Woody Hayes.

by Staubach12 :: Tue, 02/22/2011 - 3:19pm

And what evidence do you have that valuing consistency is wrong? I do think Emmitt was the better RB. It's simply more valuable to have a guy who can move the chains every time than the guy who makes a handful of highlight plays and a bunch of negative ones. DVOA also doesn't account for the fact that Smith was probably the best pass-blocking tailback of all time. Blocking was always a weakness of Sanders's game.

by Aaron Brooks' Good Twin (not verified) :: Tue, 02/22/2011 - 4:07pm

Sanders played in an early shotgun spread offense, with no TE, immobile QBs, and a sub-standard OL. I don't think you can blame the Lions sack problems on Sanders.

Sanders also managed to thrive in the absence of an effective O-Line, when a DVOA-type back would have been putting up Thomas Jonesesque numbers. The ability to hit the home run kept pass rushers honest (over-pursuing DEs got murdered by Sanders) and bailed out an O-Line that couldn't block on power plays.

by JIPanick :: Sat, 02/26/2011 - 3:38pm

Can we lose the "absence of an effective O-Line" myth, please?

Sanders' lines are remembered as poor only by comparison with those that blocked for Emmitt & Terrell Davis. They were not bad lines, merely good-not-great lines.

by tuluse :: Tue, 02/22/2011 - 4:15pm

Remember, DVOA doesn't tell us that Smith was better than Sanders. It tells us that when the Cowboys ran the ball with Smith it was more effective than when the Lions ran the ball with Sanders. Which has a lot of implications about aspects of the team that are not the running back.

by MOAT? (not verified) :: Tue, 02/22/2011 - 10:46pm

I've often thought Sanders was overrated because he had too many negative plays, and I am in the camp that believes succesful teams need RB's who can consistently move the chains more than they need big runs. Previously, I hadn't given sufficient thought to the notion that Sanders had so many negative plays because of Detroit's o-line. Perhaps, the best way to check is to look at the number of negative plays he had in college? Maybe his negative plays could even be compared to Thurman Thomas's, who played with many of the same lineman?

by David Green (not verified) :: Wed, 02/23/2011 - 1:41pm

Emmitt Smith was probably the greatest pass-blocking tailback of all time? Are you insane? Ever watch Marcus Allen play? Easily more effective than Emmitt in protecting his QB. Emmitt had a lot of dive-ineffectually-at-the-defender's-legs blocks. He was a speed bump, which sometimes is all you need, but it's hardly the same as picking up the blitzing LB and taking him out of the play - as Allen often did.

Walter Payton was also an effective pass blocker. Drove hard into the defender, was very alert and quick at understanding the rush.

Don't know if I can put Allen as the #1 pass-blocking tailback of all time - though he's light-years ahead of Emmitt Smith - but he'd definitely get my vote as the best run-blocking tailback I ever saw. Simply phenomenal.

(And, let's not forget that there's no way DVOA can account for pass-blocking abilities of a runningback, so it doesn't help Emmitt's numbers over Barry's.)

by Kal :: Wed, 02/23/2011 - 6:07pm

And what evidence do you have that valuing consistency is wrong?

What evidence do you have that valuing consistency is right?

Tongue-in-cheek aside, a lot of the DVOA 'success' comes at looking at teams who did well before and seeing how they do. This tends to heavily favor a ball control, west-coast offense style with long methodical drives.

And that's not necessarily a better gauge of success than long bombs. The value of the clock, for instance, is not clear as a positive or negative thing, and long plays vs short drives that take only a few plays has really the only effect of giving more drives overall to both sides - which can be beneficial or detrimental depending on what the match is.

The biggest 'gotcha' of DVOA is how it arbitrarily assigns success rates on 1st, 2nd and 3rd down. I say arbitrarily because while it's based on looking at drive success and down & distance metrics it isn't a derived value that changes; it's always the same percentage regardless. Thus a team that barely gets what it needs to every time will not have any wasted yards by DVOA and will look better than a team that drives the same length of the field with the same result if they do it in fewer plays - or even if they do it in the same length of plays but had more negative plays mixed in with bigger positive ones.

My gut feeling looking at this and looking at S&P (which does something almost identical in methodology to DVOA) is that boom plays are undervalued as a descriptive tool, but possibly not as a predictive one. Then again, it seems like teams that do boom plays well get more of them more often reasonably - much like RBs that get them often do so predictably.

by bravehoptoad :: Wed, 02/23/2011 - 6:57pm

Tongue-in-cheek aside, a lot of the DVOA 'success' comes at looking at teams who did well before and seeing how they do. This tends to heavily favor a ball control, west-coast offense style with long methodical drives.

I'm not sure how you're reaching this conclusion. If "did well before" means "success," then what you're saying is that ball-control offenses are more successful. I don't know how you'd prove that, but then you contradict it right away with And that's not necessarily a better gauge of success than long bombs. There are way too many definitions of "success" flying around here.

I see this conclusion a lot on this site -- that DVOA favors ball-control offenses -- but I've never seen any support that isn't either anecdotal or a priori.

In any case, Aaron has directly contradicted this conclusion in the past. Long bombs are given more value than plays that barely make a first down, and of course he's trying to value those plays correctly. What's your evidence for thinking they aren't?

I don't see why there's anything inherent in the concept of measuring success-per-play that will over-value a high-percentage offense. It would be just as easy to give too much value to big plays and have the opposite problem.

by Kal :: Fri, 02/25/2011 - 6:20am

In any case, Aaron has directly contradicted this conclusion in the past. Long bombs are given more value than plays that barely make a first down, and of course he's trying to value those plays correctly. What's your evidence for thinking they aren't?

Long bombs are more valuable on a per-play basis than a play that is a success but nothing else. However, 60 yards of 'success but nothing else' is by their admission more valuable via DVOA than the long bomb; look it up.

As to what DVOA's based on and it favoring ball control, I mention that because the teams that were doing well were doing ball control. So naturally, you'd tweak DVOA to retroactively predict/describe teams that did better. You don't have to do this consciously, even; simply change things here and there and see which has better predictive value for prior seasons. Even without bias as long as ball control offenses do well, so DVOA will have soemthing of a bias.

I don't see why there's anything inherent in the concept of measuring success-per-play that will over-value a high-percentage offense. It would be just as easy to give too much value to big plays and have the opposite problem.

There's nothing inherently wrong with it. As we saw this year with Oregon, however, a team can be hugely successful with largely big play after big play and little ball control and a per-play metric would likely fail on this as a measure of a good team. It would be interesting to combine a per-drive metric for football with DVOA and see if it's more accurate.

One very clear 'value' on a team because of DVOA favoring ball control is the Eagles, who consistently played that style and consistently got higher ratings despite losing in sometimes catastrophically silly ways. GB is another team close to that - year after year high in the rankings, and often performing much worse than their DVOA would indicate or imply.

Ultimately I think big plays are a lot like 3 point plays in basketball. A team that relies on the 3 is going to sometimes wax teams left and right when they get good percentages. Or they'll have an offnight and lose horribly as they try low-percentage play after play. It does not go for a better consistency; the variance is going to be high. But it doesn't necessarily mean that it's a bad strategy or that it'll be less successful than another one.

by ChicagoRaider :: Tue, 02/22/2011 - 11:37am

First, Oakland is historically designed to be a big-play offense. So in a way, they design to underperform in DVOA.

I think it is fairly obvious that a good defense helps a good offense. Oakland was on the flip side of this for years, where the defense would get tired because the offense could not stay on the field. The Raiders were at 31:21 -- not a huge lead, but not the hole they had before. (Usually between 28:00 and 28:30) a 3-4 minute difference in playing defense is a significant, though not overwhelming, percentage of time. And given that they were facing weak defenses, I suspect that weak can collapse if overworked.

by drobviousso :: Tue, 02/22/2011 - 5:30pm

I think the idea the Oakland is built for a non-DVOA optimized strategy is it (along with the obvious opponent/situation adjustments as well.)

It's a bit of a chicken/egg problem, but for many years, it seems that most of the best teams were the consistent offensive teams, not the mad bombers. Maybe they were better because they were consistent. Maybe they were better because they had talent or coaching advantages. DVOA can't really tell us reasons, only correlations. Maybe as we learn more in future seasons, teams like GB and Oakland that Love the Bomb will start winning more, and DVOA will be recalculated.

by ChicagoRaider :: Tue, 02/22/2011 - 5:45pm

I think it is pretty clear that Oakland was an inconsistent team. It also appears that they lost more to great defenses and gained more from weak defenses than other teams.

Some of that was probably the unexpected emergence of certain players that affected game planning. Jacoby Ford, Marcel Reece and Darren McFadden: stand up and be recognized. The good defenses could handle the unexpected. The bad just got toasted.

by bravehoptoad :: Wed, 02/23/2011 - 12:56pm

I've never bought the idea that DVOA under ranks big-play offenses and over-ranks small-ball offenses. When this site was young, DVOA consistently under ranked Philadelphia, for instance, even though McNabb and Reid were running their same old offense.

DVOA is designed to give big plays their proper credit. If it were consistently under-rating them, it would be an easy problem to fix.

by DC in ATL (not verified) :: Tue, 02/22/2011 - 12:02pm

The answer is obvious. This is just Al Davis working some weird magic from beyond the grave.

by commissionerleaf :: Tue, 02/22/2011 - 12:36pm

This isn't too surprising. It's the flip side of why the Colts this year (and the Donovan McNabb pre-DeSean Jackson Eagles) are DVOA fiends. Short successful plays, especially short successful passing plays, are very good for your DVOA. DVOA is predictive, and therefore is keyed to what predicts future success; and what predicts future success is the ability to move the football and get first downs. Hence, Brady's DVOA is magical because he hits Wes Welker for seven million 5 yard drag routes every game. Hence, Donovan McNabb's Eagles were a great team, because they ran well, and passed frequently.

Long plays are not discounted (and I mean discounted, not NOT counted) in DVOA because they aren't predictive; being able to hit 10 successive pass plays is more predictive of future success than being able to score on a 50 yard bomb.

Thing is, partly due to sample size issues, it's hard to tease out what part of big-play success is predictive, and should be part of the formula. Certainly big-play success is part of why Ben Roethlisberger is an NFL quarterback instead of a registered sex offender. And Chris Johnson's success rate isn't going to make him a star. But they're valuable players because they consistently hit home runs even though, on any given play, you're better off betting against it.

Unfortunately, NFL careers are short enough, and seasons short enough, that the sample you'd need to use big play successes as a sufficient part of the formula to overcome the "consistency bias" in DVOA is probably larger than the sample that the actual seasons provide. Big play offenses are always going to look less impressive in DVOA than points scored, by ranking.

All that said, Darren McFadden is quite possibly the best running back in football.

by MJK :: Tue, 02/22/2011 - 12:42pm

Well said, all of this.

by Marcumzilla :: Tue, 02/22/2011 - 5:15pm

Are you sure you meant Colts as DVOA darlings this year? Their offense was 7th by DVOA and 4th by scoring.

(I agree with most of the rest; I didn't see enough of RunDMC this year.)

by Dan :: Tue, 02/22/2011 - 3:46pm

Oakland was 23rd in the NFL in offensive DVOA. But they had weak opponents and good fumble recovery luck - in VOA unadjusted for those two things, they were 19th. They also had a lot of big plays which are unfavored by DVOA (2nd most in the league) and a good field goal kicker (7th in FG DVOA), so in points per drive they did slightly better: 16th. But they had 199 drives - 3rd most in the league - so in total offensive points they were 10th. But they were also 3rd in non-offensive points scored, behind only Arizona and New England, with 7 return TDs (3 kick returns, 2 INT returns, 1 fumble return, and 1 blocked punt return) and 2 safeties, so in total points scored they were 6th.

by Dan :: Tue, 02/22/2011 - 4:08pm

In other words, it's mostly Aaron's points 3 & 4. The Raiders were 3rd in the NFL in the number of drives with 199, which was worth almost 30 points over what they would've scored with an average number of drives. And they were 3rd in the NFL in defensive/special teams touchdowns with 7 (only 3 defensive TDs, but 3 more on kick returns plus one on a blocked punt), which is good for another almost 30 points over what an average team gets. Those two factors are the difference between an average offense and a #6 rank in points scored.

by Jay raffa (not verified) :: Wed, 02/23/2011 - 6:32pm

Umm ya Oaklands offense was pretty good...great run game all year long and if we had a steady QB in there we would have been even more dominant,.....whatever formula you are using sucks sorry...Oakland 2011 look out .....very underrated dont forget we were 6-0 in our division last year

by BigDerf :: Wed, 02/23/2011 - 11:36pm

Generally, long ball offenses don't work but perhaps Al Davis' obsession with speed has finally collected enough guys that with a creative coordinator, it actually works.

It's basically the type of team works great in older Madden games. If everyone on the team is that fast but you can still run the ball inside, I don't know how a defense stops that.