Writers of Pro Football Prospectus 2008

18 Dec 2013

Scramble: The Losing Ways of Winners

by Tom Gower and Mike Kurtz

Mike: I saw someone assert, recently, that baseball had the most parity in US professional sports.

Tom: This person is clearly wrong, unless he's referring to gross difference in winning percentages, something I would attribute to baseball's 162-game season resulting in compressed records, especially given that pitchers vary in caliber the way starting quarterbacks do.

Mike: Winning percentage is probably what he is referring to, but it might be interesting to look at champions, because while baseball fans whine about the Yankees and Red Sox and whathaveyou, our NFL fan bretheren also have the Patriots and the Steelers on our side.

Tom: The NFL playoffs are now a ridiculous crapshoot. Only two of the last 13 Super Bowls have been won by teams that had the No. 1 seed in their conference.

Mike: I am not sure that is a recent development.

Tom: The team with the best DVOA in the league has won one of the last 13 Super Bowls. From 1975 to 1999, the Super Bowl winner was not the top seed of its conference only seven times (1997 Broncos at No. 4, 1992 Cowboys at No. 2, 1990 Giants at No. 2, 1988 49ers at No. 2, 1987 Redskins at No. 3, 1980 Raiders at No. 4, 1978 Steelers at No. 2). It should be noted that prior to 1990, the NFL prevented divisional rematches in the playoffs before the conference championship game. For example, in 1989 the 49ers had the best record in the NFC but hosted NFC North winner Minnesota in the conference semifinal because the wild card winner was their divisional mate the Los Angeles Rams.

Mike: A 72-percent success rate for the top seed seems pretty good. That is also an 88-percent success rate for the top two conference seeds.

Tom: Before 1975, the conference championship game host alternated by divisions, which is really screwed up and deeply weird if you think about it. So the unbeaten 1972 Miami Dolphins played the AFC Championship Game on the road, in Pittsburgh.

Mike: Football was pretty screwed up back then.

Tom: It's a low-key kind of weird how intelligently the league is actually set up now.

Mike: The NFL is by far the best run and organized league in sports.

Tom: Then you look at old things like this, and you recognize how good we have it today.

Mike: In any case, the old pattern was that a top seed, or more accurately the top two seeds, were almost certain to win the league championship. That has not borne out, recently.

Tom: Beginning with the 2000 Ravens, everything went to heck. I do not see an obvious explanation for that. The coming of the salary cap leading to greater turnover makes sense, but the cap took effect in 1994.

Mike: Well, let's look at DVOA. We have DVOA back to 1989. In the 90s, the top seed won a little over every other year.

Tom: Of the Super Bowl winners from 1989 to 1999, three were not the top team by DVOA (1994 49ers at No. 3, 1995 Cowboys at No. 2, 1997 Broncos at No. 2).

Mike: So, the top team by either DVOA or conference seed won every year in the 90s except 1997. For the 1990s, that leaves us with two pretty good predictors that could probably be combined into one very good predictor. That is not true of the 2000s.

Tom: And in 1997, the No. 1 team in DVOA played the No. 2 team in DVOA in the Super Bowl. And then we get to the 2000s (team ranking according to DVOA):

  • 2000 Ravens, 3d
  • 2001 Patriots 11th
  • 2002 Buccaneers 1st
  • 2003 Patriots 4th<
  • 2004 Patriots 2d
  • 2005 Steelers 4th
  • 2006 Colts 7th
  • 2007 Giants 14th
  • 2008 Steelers 4th
  • 2009 Saints 6th
  • 2010 Packers 4th
  • 2011 Giants 12th
  • 2012 Ravens 8th

The bar for Super Bowl champion, by DVOA, is much lower in the 2000s than in the 1990s. And yet, we've seen the Patriots and the Steelers both win multiple titles.

Mike: So I suppose our next question is, do the teams of the 90s have a higher average DVOA than the teams of the 2000s?

Tom: The Super Bowl winners, you mean?

Mike: Yes. Well, and the actual No. 1 DVOA teams. Theoretically, the 90s DVOA leaders are so much better than the average team that both the No. 1 seed and the championship immediately follow. Also theoretically, the 00s DVOA leaders are closer to average and therefore more vulnerable than their 90s brethren.

Tom: The best team in DVOA from 1989-1999 averaged 35.9%. The best team from 2000-2012 averaged 34.3%. That's not a noticeable difference to me.

Mike: More importantly, considering the small sample size it is nowhere near a statistically significant difference.

Tom: Absolutely not.

Mike: So the best teams are still just as much better than the competition. Why do they keep losing?

Tom: I wish I knew. The one thing that does jump out looking at the numbers is there were four teams that had the best DVOA in the record with a DVOA under 30% in the past thirteen years, compared to two in the decade of the 90's.

Mike: That's still just 5% away from the average of the top teams, however. Again, probably not significant. Liberalization of passing rules is one possible culprit, but that is only the second half of the 2000s. And the top offensive team by DVOA has not fared well, either, with only the 2006 Colts topping the league in offensive DVOA and winning the Super Bowl. So having the best offense in an offense-friendly environment also does not seem to yield championships.

Tom: I do a couple radio spots, and both of them in recent weeks have asked about who's going to win the Super Bowl. The best I feel like I can do is give a list of teams that are maybe good enough to win it. And talk about last year's Ravens and the 2011 Giants, teams that showed very few signs of particular greatness until playing well enough in the postseason to come away with the Lombardi Trophy. You read or listen to Aaron talk about it, and you can tell he's frustrated that the best team in the league has stopped winning the Super Bowl. As an analyst, I'm just as frustrated as he is.

Half a dozen years ago, we tried to look if there were anything teams that actually did win the Super Bowl had in common. We tried it in 2006. The Colts came out as awful by that metric. They won the Super Bowl. We tried again in 2007. The Patriots and Giants came out badly. They, of course, met in Arizona. We have not tried again, nor do I see a particular reason to.

I consoled myself with the subjective knowledge that the team that won was always a team with a demonstrated theoretical capability of playing at a high level on both offense and defense. Then Baltimore last year won. I could sort of see Joe Flacco winning the deep pass lottery for a few weeks, like an inconsistent Eli Manning did play well for a few games in a row for the first time in 2007, but defensively? Their two best games of the season, by a mile, came completely out of the blue, in the postseason against Denver and New England. And it wasn't like they got a great player back so it made sense narratively, like with Indianapolis and Bob Sanders.

I'm basically at the stage where I'm (a) pulling my hair out and (b) throwing my hands up in the air and giving up, which is a really awful position to be in.

Mike: My only other thought is perhaps that while the difference between the top team and the average have remained unchanged, the difference between the top team and other playoff teams has shrunk?

Tom: The 1991 Redskins, the best team in DVOA history, won the Super Bowl. The 2007 Patriots, second-best, did not.

Mike: With a similar spread between the 91 Redskins and the 49ers, as compared to the 07 Patriots and Colts.

Tom: Right, and neither team faced their top conference competitor in the playoffs.

Mike: Back of the napkin math refutes this hypothesis, also; the top of the league was roughly as dense in the 1990s as it was in the 2000s.

Tom: In the Super Bowl, though, the Redskins faced the AFC's best team in the Bills, while the Patriots played an average Giants team that wasn't in the top five in the NFC by DVOA. The Redskins won. The Patriots lost. It's so non-obvious it feels random. Maybe it is, or maybe I'm just not smart enough to come up with the right explanation for what's going on.

Mike: It does feel random, but it stands in stark contrast to decades where the process of selecting a champion was almost entirely non-random. Something changed around the turn of the century that snapped a very long trend.

Tom: Maybe, just maybe, any year now, the best team in the league will start winning the Super Bowl again. But I will not believe that until it actually happens.

Live From Santa's Spanish Galleon

Mike: I don't get who Santa is supposed to be in this commercial. It is like he is some kind of horrible combination of Steve Jobs and Billy Mays. It feels like it's referencing something, but without an actual reference the clip is just bizarre and annoying.

Tom: Keep in mind all of these people are Santa's employees. Steve Jobs, maybe, or even more cult-ish. The elves are also really eager to believe Santa's claim that his watch just took a picture, despite them perhaps having no knowledge about a watch's ability to do so nor anything beyond Santa's word as evidence. Maybe Santa did project the picture he just took on the screen, but we're not shown that.

Mike: The issue is that Jobs' presentations were much less emphatic than Santa's in this commercial.

Tom: And when Mrs. Claus calls, we get the projection of the "watch" with the call signal from Mrs. Claus. But it's about as realistic as the Duke Nukem Forever trailer.

Mike: He made the product speak for itself, almost, whereas here Santa is wildly gesticulating and shouting. Perhaps it's another edition of 'Samsung thinks Apple is stupid and evil and everyone who likes Apple is stupid.' This would all be less of a problem if the Galaxy Gear weren't such a thoroughly useless device, based both on reviews and Best Buy's reported 50 percent return rate.

Tom: One of the issues is they introduced a wearable device at a time where the specific wearable device is as unpopular as it's been in several centuries. The "it's a miracle" line at the end seems like it's a traditional thing to say, about whatever new gimmick Santa comes up with that year. Or at least the way the elves say that makes me think so. But what the heck is going on with Carly Rae Jepsen there, aside from the obvious "call me"/phone connection? Santa already seems to have his employees pretty well brainwashed, so what does he need to spend money on song licensing for, especially when his apparent business is non-revenue?

Mike: Aw, I thought Eli Manning single-handedly led a resurgence of the wristwatch!

Tom: Unstoppable, just like Eli Manning throwing interceptions!

Loser League Update

Quarterback: After last week, we had a return to normalcy with the only negative quarterback score of the week, -3, belonging to a New York quarterback who threw five interceptions. The only irregular part was it was Eli Manning and not Geno Smith, the latter of which put up 14 points.

Running Back: Well, Eli did not have a good day throwing, but I am sure the Giants were at least able to run ... oh, hello there, Andre Brown atop the running back standings with 1 point. The runner-up trio with 3 each consisted of Stevan Ridley, Robert Turbin, and Danny Woodhead.

Wide Receiver: Somehow, though, no receivers from that Seahawks-Giants game finished atop the Loser League scoreboard this week as both Victor Cruz and Golden Tate managed a point too many. Mike Brown offset his receiving yards with a fumble for 0 points, while Santonio Holmes, Andre Johnson, Darius Johnson, Andre Roberts, and Tiquan Underwood each had 1 point.

Kicker: Really, though, this might actually be the first time a single team has had a player put up the worst score of the week at three different Loser League positions. Josh Brown did not get to attempt a kick for his 0, while Garrett Hartley missed two field goals to offset his made field goal and extra point, and is no longer employed by an NFL team.

To see how your team did, check out the Loser League results page.

Awards!

KEEP CHOPPING WOOD: The Denver Broncos went four-and-out on their first possession of the second half, but Andre Caldwell managed to down the punt at the 1 and the defense forced a third-down incompletion from Philip Rivers to Eddie Royal. Bring on the punt team, meaning good field position for the Broncos was likely after a punt from the 7. They still had 23 minutes remaining for a comeback from a 24-10 deficit. Until Nate Irving goes into the neutral zone with the punt team out, giving the Chargers another first down, that is. San Diego would not score on that possession, but they would run off another 6:48 of clock between the Irving penalty and their actual punt, and Denver would get the ball at their own 11 instead of at midfield.

MIKE MARTZ AWARD: The best way to win in the NFL, as a general rule, is by throwing the ball. There are exceptions. One of those is, perhaps, when facing a terrible run defense, and you are having success running the ball already ... oh, wait, never mind, that was last week's write-up about how the Dallas Cowboys failed to run the ball and it helped cost them the game. An NFL team would never make the same mistake like that twice in as many weeks, would they?

Really, though, we will instead harken to Greg Schiano and let him explain the Buccaneers' ill-fated kickoff return handoff that gave the 49ers a clinching touchdown. Greg, take it away: "It was a called play. We were going to run a reverse if the opportunity presented itself the right way. It didn't, yet we still ran it, but it's okay. We made a mistake. Guys make mistakes sometimes." Yes, they do. Here is a hint: When a kicker is putting most of his kickoffs deep into the end zone, as Phil Dawson was that game, just accepting the touchback is probably a good idea instead of trying to do something tricky.

Todd Haley Mystifyingly Job-Saving Lock of the Week

Tom: As a reminder, all lines are courtesy of Bovada and were accurate as of time of writing, while all picks are made without reference to the FO Premium picks.

I thought I had it last week. The San Francisco 49ers scored to go up 23-21 with over four minutes to play, giving the Tampa Bay Buccaneers enough time to pick up a covering touchdown before valiantly losing. The Bucs being the Bucs, they instead collapsed and a 20-14 49ers lead became a 33-14 final. Not the 38-0 I feared, but nor did Tampa cover the 5-point spread. You, meanwhile, chose the Patriots, who lost.

Mike: Yes. Both of those things happened. I've given up trying to figure out the Patriots.

I think I have the Chiefs figured out to some extent, inasmuch as they are really good at stomping bad teams. Indianapolis, as its weighted DVOA indicates, has been playing like a bad team recently. Part of it is injury, part of it is regression. All of it is fodder for Kansas City's stompin' roadshow. Will they win the Super Bowl? Well, as we pointed out above, that is basically impossible to figure out. Will they cover against the Colts in Kansas City? Yeah. Kansas City Chiefs -7 vs. Indianapolis Colts.

Tom: The Miami Dolphins are playing a December road game in Buffalo, and are favored. The Dolphins are a pretty average team. The Bills are worse than that, but they're not that bad. Yes, Buffalo is eliminated from playoff contention and Miami is tied for the lead in the chase for the sixth seed in the AFC, but what would the Bills do, play a rookie quarterback? They've been doing that already. Buffalo is not so bad that Miami should be favored, let alone by a field goal. The early forecast is not particularly cool for December in Orchard Park (48 degrees), but an 80 percent chance of rain (per weather.gov) would make conditions plenty miserable on their own. I'll take the Bills and the points. Buffalo Bills +3 vs. Miami Dolphins.

Posted by: Mike Kurtz and Tom Gower on 18 Dec 2013

98 comments, Last at 02 Jan 2014, 4:02pm by Duke

Comments

1
by Scott Kacsmar :: Wed, 12/18/2013 - 3:21pm

The difference comes at the bottom of the playoff pool. The Wild Card teams are better now, so there's more competitiveness in January.

•From 1990-2001, the playoff success for the six seeds ranged from 22.6 percent to 69.1 percent (46.5 percent gap).
•From 2002-2011, the playoff success for the six seeds ranged from 40.6 percent to 56.1 percent (15.5 percent gap).

The Chiefs and 49ers are likely to be Wild Card teams this season and at the rate they're playing, both would be up to the caliber of many teams who have had a first-round bye in the past.

5
by Jerry F. (not verified) :: Wed, 12/18/2013 - 4:01pm

Could this then be related to the division realignment that happened in 2002? I'm not sure why that would be, because it seems to me that this means more crappy division winners getting in at the expense of quality wild card candidates. But it's something that changed.

21
by Perfundle :: Wed, 12/18/2013 - 5:14pm

But it also means that those wild card candidates have a much better chance to win their first games against those division winners. Look at this year, for instance: you have three wild card teams in the top 5, and potential division winners as low as 19, 20 and 21.

86
by tuluse :: Thu, 12/19/2013 - 6:38pm

It makes sense, going from 3 divisions to 4 increases the chances that a division winner won't be very good and that a wild card team will be good.

73
by RoninX (not verified) :: Thu, 12/19/2013 - 10:32am

Is the bottom of the playoff pool worse by DVOA though? The article seems to argue it was not.

93
by skeptic3 (not verified) :: Fri, 12/20/2013 - 4:43pm

How about, for the last few years:
passing is more important;
therefore team success depends more on a single player, the QB;
therefore team game to game variance is larger;
therefore upsets are more likely.

Put differently: a QB can suddenly get hot or cold more easily than a whole team can.

2
by MilkmanDanimal :: Wed, 12/18/2013 - 3:28pm

The most amusing part of the whole issue with the Bucs fumble on the called reverse on the kickoff return is not only did the guy drop the ball and fall down, but he managed to hurt his foot and go on IR in the process. That's a nice trifecta of failure.

3
by dmstorm22 :: Wed, 12/18/2013 - 3:38pm

A couple theories:

1.) Admit that the current NFL playoff results aren't really stranger than before, but just stranger to the other extreme. The dominance of 1-2 seeds in the 70-90's is statistically odd. They shouldn't have won as much as they did. The #1 seed in the NFC won every divisional round game from 1988 to 2006. That not very statistically likely. In a sport with just a 16-game season, and the shortest playoff system possible (1-game series), it's not all that strange to get crazy results every now and then. Because of this, the NFL playoff results should vary more in the NFL than NBA, NHL, MLB.

Anyway, outside of that here goes:

1.) Over-reliance on more high-variance passing game has fueled recent top teams by DVOA, and these are high variance strategies, more likely to have a bad game, more likely to be effected by weather (though I can't think of many times that it has).

2.) Increasing top-of-roster wages has hurt depth, making injuries more impactful now than what they used to be.

3.) Random fluctuation (probably at least 50% of the reason).

Let's go case-by-case. There really are four buckets for the Champions from 2000-2012.

Bucket #1: Conventionally Great Teams: 2002 Bucs, 2003-04 Patriots, 2008 Steelers.

These were Top-4 DVOA teams that had 1st round byes. Two had to win a Title Game on the road, but the Bucs were better by DVOA than the '02 Eagles, and the '04 Patriots were basically as good as the Steelers.

Bucket #2: Not Great, Great Teams: 2009 Saints

This is a small bucket. They're just the 2nd team with the #1 seed in this period to win a Super Bowl, but they were #6 by DVOA. Some of this has to do with them resting their starters in Week 17 (after a shock, meaningless loss to the Bucs in Week 16), which pushed their DVOA down. They probably belong in that 1st bucket.

Bucket #3: Great DVOA, low Seed Teams: 2000 Ravens, 2005 Steelers, 2006 Colts, 2010 Packers

These were all very good teams by DVOA (or in the case of the Colts, a great offense with a defense that when healthy was very good, as indicated by their defensive DVOA in 2005 and 2007) that had lower seeds because other teams won more games. Twice a team likely lost their division title because of in-season injuries to a starting QB (Packers in 2010, Steelers in 2005). The Ravens had a great team in their division. The Colts had some ridiculous losses, like falling victim to a 60-yard field goal by Rob Bironas. Win that, they get the #2 seed, and it seems a lot less strange that they win.

Bucket #4: The Outliers: 2001 Patriots, 2007 & 2011 Giants, 2012 Ravens

These are the ones that make your head hurt. Now, the 2001 Patriots were a #2 seed, but that was in a watered down AFC. Still don't know how they did it, but there was incredible amounts of luck. 1st, with the Tuck play (rule was right, but Brady was fortunate the rule came into play), then 2 special teams TDs in the AFC Title Game, and then they capitalized on the Rams biggest weakness, turnovers, to win a game they were outgained by 160 in. The other three are the real crazy ones. The 2nd Giants team and Ravens team had similarities. Defensive issues late in season were cured by returning players and health. QBs suddenly stopped throwing picks. Both had some fortunate wins. Both beat some great, great teams. The '07 Giants were even crazier.

Having four odd Champions in 13 years isn't that strange, but having them makes the more explainable ones (Bucket #3) look stranger than they are.

In the end, I have no idea. But I don't think it is all that bad. If you want to crown the best team, have each team play each other home and away and have something like the EPL. American's don't like that, so this is what we're given.

9
by volatile (not verified) :: Wed, 12/18/2013 - 4:20pm

My first thought was that it probably has to do with higher variance strategies as well. A team with average DVOA but very high variance is more likely to win a super bowl than a team with average DVOA and low variance. Similarly a team with high DVOA and high variance is more likely to lose a super bowl than a team with high DVOA and low variance. Is the average variance of teams higher in the 2000s than it was previously? Are the winners on the lower DVOA side particularly high variance teams?

37
by Bobman :: Wed, 12/18/2013 - 6:53pm

That Bironas FG killed me. And the Jags got a similar kick a year or two later....

78
by PaddyPat :: Thu, 12/19/2013 - 1:05pm

I still believe that the real trend here is home teams losing games in the divisional round. If you look at the long eye of history, that never, ever, ever happened. I have made myself hoarse trying to talk about this on Footballoutsiders without garnering a lot of interest, but before around 2004, the annual divisional round upsets were something like .7 out of 4 games. Since then it's more like 1.6. One can argue this in terms of weaker rosters and depth and stuff related to free agency, but I really believe it's all about the passing game dominance.

I believe that passing is not as reliable week-to-week as running games and defense were. I know that this assertion is not born out by current numbers, but the game has changed. In the past, a good ground game and a great defense generally saw you through. Add to that that the best two teams in each conference got a bye and a home game, and conference championships were almost invariably matches of the 1 and 2 seed. Go back and watch some old super bowls, you'll be astounded that the great dominance of the teams is their running backs.

4
by bingo762 :: Wed, 12/18/2013 - 3:42pm

Need some fantasy football help. Cameron or Pitta? Streater or Jennings? thanks

75
by ALauff :: Thu, 12/19/2013 - 11:13am

It looks like Cameron might not play (concussion), so your decision appears to be made for you. FWIW I have to roll with Pitta in one of my championship games (Gronk owner). For your other choice, I would use Streater over Jennings (assuming you meant Greg): despite their recent string of injuries in the secondary, Cincy is no joke in pass defense while SD is, well, just that.

Good luck!

94
by Tom Gower :: Fri, 12/20/2013 - 7:37pm

Cameron's missed practice, as noted, so Pitta by default it seems to be. Streater's facing the Chargers, which makes him the choice to me.

6
by Tony Heath (not verified) :: Wed, 12/18/2013 - 4:01pm

The Format of the Playoffs Changed in 2002 when the league realigned to its current format. Could that be part of the reason?

7
by nat :: Wed, 12/18/2013 - 4:12pm

While it's true that only 3 of the last 13 Super Bowls were won by a conference top seed, it's also true that an impressive 5 of the last 15 were, as well as a very good 3 of the last 10.

Way to use an arbitrary (and obviously cherry-picked) cutoff to push a point, guys!

38
by Jimmy Oz :: Wed, 12/18/2013 - 6:54pm

Thanks. Glad I'm not the only one rolling my eyes

67
by Whatev :: Thu, 12/19/2013 - 6:13am

Way to assume that choosing a big round number as the cutoff was an intentional attempt to skew the results, guy.

And whether they put the cutoff there or 2 years earlier or 3 years later, it won't make the period extending from the chosen cutoff to the present have the same rate as the period from 1975 to the cutoff.

8
by dk240t :: Wed, 12/18/2013 - 4:19pm

Maybe, the unsatisfying answer is, splits happen.

Flip a coin 20 times.

HHHHHHHHHHTTTTTTTTTT is just as likely as HHTHTTHHTHHTTHHTTHTT, but one "looks" random and the other doesn't.

14
by RickD :: Wed, 12/18/2013 - 4:50pm

But if you look at summary statistics, the odds of "10 heads in a row + 10 tails in a row" is far lower than the odds of "at most 2 heads or 2 tails in a row". There are a lot more sequences that give you the latter behavior.

10
by RickD :: Wed, 12/18/2013 - 4:34pm

With stuff like this, you want to compare the DVOAs of the Super Bowl winners with the DVOAs of the teams they beat along the way. Ordinal ranks are not terribly informative. A team with the 5th best DVOA could have 99% of the DVOA of the team with the best DVOA, in theory. That would be less interesting than if a team with the 2nd best DVOA upset the team with a higher DVOA, if the latter team's DVOA was 40% higher.

57
by mshray63 (not verified) :: Wed, 12/18/2013 - 11:48pm

I think this is a good point. The more I read this the more I thought that the real comparison is to the NCAA basketball tournament. Lots of championships by non-#1 seeds to start with, and almost never a match up of two #1s or even a 1 vs.a 2. But the team that knocked a number one out in an earlier round rarely goes all the way (the glaring exception was that '97 Arizona team that beat 3 #1 seeds).

Basically my question is, is this perhaps more about high DVOA teams increasingly finding ways to lose before the Super Bowl (or even their conference championships), thereby opening the door for the lower DVOA teams, and not just about lower DVOA getting hot/lucky?

11
by theslothook :: Wed, 12/18/2013 - 4:40pm

I think the narrative of "playoffs are funky" is strengthened when used in combination with pff findings. For instance - the ravens corners(all three) were rated negatively in the regular season. Reed was basically above average, but pollard was actually slightly below in coverage. Basically, this screamed liability. Of course, then all 3 corners have great postseasons, including pollard playing well, and its reed whos numbers dip down.

Contrast that with the 49ers who had 5 members of the secondary grade well in the regular season. THen in the postseason, nearly all except I think carlos rodgers or maybe it was brown graded poorly.

This phenomenon of massive improvement come playoff time is true for the 06 colts, 2007 giants, 2008 cardinals, and the 2011 giants. There is simply no good explanation for this.

Personally, I like it this way. Basketball playoffs feel completely meaningless until we get to the conference championship because very few teams have a true chance of winning. At least in football, once you get in, you can win. That makes every game interesting and gives everyone hope that you don't necessarily need to be the best team in order to win.

13
by RickD :: Wed, 12/18/2013 - 4:47pm

The NFL playoffs are very short in terms of the number of games played. No other major professional sport decides a playoff with a single game. Upsets are far more likely to happen in a 1-game series. And by chance alone, there's a higher probability of some team going on a three-game streak of playing above its head.

Why has this happened more in the past 10 years than it did in the 80s and 90s? I think the salary cap has leveled the playing field. The gap between the .500 teams and the league leaders isn't as large as it used to be.

15
by theslothook :: Wed, 12/18/2013 - 4:50pm

But the salary cap started in 1994. Why specifically has this phenomenon happened only in the recent decade?

19
by RickD :: Wed, 12/18/2013 - 5:10pm

Well, the Broncos kept their team together for their Super Bowl wins by violating the salary cap.

So there's a bit of fuzziness in terms of when we should expect the leveling to start. Between 1994 and 1998 there were some salary cap shenanigans. And the first "hot middling team to go on a run to the Super Bowl" was the Pats of 2001.

51
by sycasey (not verified) :: Wed, 12/18/2013 - 7:56pm

Teams started pro-rating their contracts and making them heavily back-loaded to avoid the cap and sustain success with loaded rosters. The 49ers were particularly good at this.

But eventually teams employing this strategy have the pay the piper, which perhaps started happening around 2000.

87
by tuluse :: Thu, 12/19/2013 - 6:42pm

Wouldn't you expect a delayed response to the salary cap though? They didn't rip up every contract in 1994 and start from scratch.

56
by jonnyblazin :: Wed, 12/18/2013 - 10:59pm

Ravens win is a lot less flukey when you consider that 2 dominant defensive players (Suggs and Ngata) got healthy for the playoffs and they inserted a super-intelligent ILB (Lewis) back in the lineup, plus they reshuffled their offensive line and put their best LT on the field. Their regular season DVOA was not indicative of their overall skill/talent level.

12
by RickD :: Wed, 12/18/2013 - 4:42pm

FF championship game

for my flex spot, options are Andre Brown, Fred Jackson, Andre Caldwell (picked up after his hot game last week), Roddy White, and Tavon Austin.

Opinions?

76
by ALauff :: Thu, 12/19/2013 - 11:17am

I'd probably use Fred Jackson. Fifteen carries (plus GL work) seems like a near certainty, he's at home, and he's playing against a team that doesn't defend the run very well. Andre Brown is a close second, but I just really hate the matchup.

95
by Tom Gower :: Fri, 12/20/2013 - 7:47pm

Concur with Fred Jackson, though that may be my pique with Brown after last week. Caldwell is a Patriots-esque play-at-your-own-risk for me-could hit double digits, could go 2 for 17. Austin is in the same category. If I had to pick a WR, it'd be White for sure, but Jackson over him.

16
by Will Allen :: Wed, 12/18/2013 - 4:54pm

The effect of the salary cap took a few years to fully manifest itself, especially since two of the Super Bowl winners from that era were cheating on the cap.

17
by theslothook :: Wed, 12/18/2013 - 4:56pm

But I fail to see how the salary cap explains the ravens, who were essentially average on offense and below average on defense, for most of the year suddenly became world beaters at both. Or how the 07 giants, a below average at everything team, managed to upset a team that was 2nd in dvoa history. Or how the colts in 06 can go from one of the worst rush defenses in history and somehow not get pile driven by the run in the playoffs.

20
by Scott Kacsmar :: Wed, 12/18/2013 - 5:11pm

The salary cap did take a few years to have an impact, which is why we had loaded teams like the 94 49ers, 95 Cowboys and 96 Packers winning those titles. The 1997-98 Broncos did cheat the salary cap.

The period of 1999-03 is one of the strangest in NFL history. Teams were winning Super Bowls with unheralded QBs taken low in the draft or out of the grocery store like Kurt Warner. They did it with great defenses. A team like the 1999 Rams had everything fall into place in a hurry with a low-cost Warner putting on one of the greatest seasons ever, the Marshall Faulk trade, Torry Holt was a rookie and Orlando Pace was in this third year. The salary cap wouldn't have been a huge problem at that point. That team also did not sustain long-term success.

Then the referees were reminded illegal contact exists in 2004 and we entered into this pass-happy era where teams with a HOF QB can challenge for the perfect season, but usually watch things end miserably in January. We've also seen Nick Harper, David Tyree, Tom Brady's torn ACL, Kyle Williams and Rahim Moore make dramatic things happen to change the outcomes of those seasons.

I think the narrative about the 2006 Colts turning their run defense around ("Bob Sanders!") overlooks the simple fact that the Chiefs and Ravens were not very good offenses with rather one-dimensional running games and no passing threat. The Colts were able to tee off on those teams and played with the lead throughout the game. New England ran the ball very well in the first half of the 2006 AFC Championship, but everything but Ellis Hobbs' kick returns collapsed in the second half that day. In the Super Bowl, Chicago also ran the ball well, but no one acknowledges it because everything about that game gets chalked up to Rex Grossman's awfulness.

There are similarities in the Giants and Ravens of recent years. As for pulling off playoff upsets, there were signs in the regular season. The 2007 Giants handed the Patriots their largest deficit (12 points) of the season in Week 17. We've seen the Giants play them will since, sweeping them in 2011. The Ravens also beat the Patriots last season in Baltimore and swept them with the playoff win.

Baltimore's 34-17 loss to Denver in Week 15 looks bad on paper, but it was Jim Caldwell's first game calling plays and the game really swung on a pick-six before halftime thrown by Flacco. In that game the defense held Manning to a season-low 204 passing yards and forced Denver into a season-worst seven three-and-out drives. So there was hope. Go to the playoffs and the Ravens got 14 early points because the refs botched pass interference twice. That defensive foundation (Reed, Lewis, Ngata, Suggs) was in place only in the playoffs. Pollard and Ellerbe also missed the Week 15 game and returned for the playoffs. Denver guard Chris Kuper returned for the playoffs and killed his team with three costly penalties. The Ravens could not pressure Manning early, but got after him nine times once Knowshon Moreno left injured. Even after all of that, Denver lost control of the win after the Rahim Moore play.

Most things that have happened can be explained, except for the 2001 Patriots not getting blown out by a superior team.

24
by RickD :: Wed, 12/18/2013 - 5:32pm

"Most things that have happened can be explained, except for the 2001 Patriots not getting blown out by a superior team."

Mike Martz vs. Bill Belichick.

28
by dmstorm22 :: Wed, 12/18/2013 - 6:00pm

It's really not even that. It's: 3 to 0. Turnovers for the Rams to Turnovers for the Patriots.

The Pats got 17 of their 20 points on those three turnovers, including a pick-6 and the TD before halftime after a fumble (by Proehl, iirc) near midfield.

It's become about Martz's arrogance, but if Warner doesn't throw wildly off-target twice and Proehl doesn't fumble, the Rams win that game easy.

35
by RickD :: Wed, 12/18/2013 - 6:34pm

But the turnovers were part of who those teams were. The Rams had 44 turnovers in 2001, while the Pats had 28.

I'm considering the simple fact that Martz didn't make good in-game adjustments to the kind of defense that the Patriots were playing. And he didn't have the Rams ready to compete at the beginning of the game. He didn't use Marshall Faulk much, even though Faulk was the best offensive weapon in the NFL. That the Rams didn't score a TD until the 4th quarter was ridiculous.

40
by dmstorm22 :: Wed, 12/18/2013 - 6:59pm

I'm not saying the turnovers were a fluke, but Kurt Warner throwing two awful passes isn't on Mike Martz.

Running Fualk may have helped, but the Rams got like 360 yards passing. That wasn't the issue.

45
by Scott Kacsmar :: Wed, 12/18/2013 - 7:34pm

Vrabel hit Warner in the face on the pick-six. It should have been a penalty in 2001 and it would almost certainly be one in today's game.

46
by dmstorm22 :: Wed, 12/18/2013 - 7:38pm

Warner threw it first, iirc.

That second pick was just terrible though, an awful pass.

58
by mshray63 (not verified) :: Thu, 12/19/2013 - 12:15am

"We've also seen Nick Harper, David Tyree, Tom Brady's torn ACL, Kyle Williams and Rahim Moore make dramatic things happen to change the outcomes of those seasons."

You missed the most obvious one of all: James Harrison's 100 yd INT return in XLIII. Arizona has 1st & goal at the 2 with 18 seconds left. They get a FG or TD & it's either a 10-10 tie or 14-10 AZ at the half. Even the Cardinals can't score at all, or better yet, if Larry Fitzgerald catches Harrison one yard sooner, it's still 10-7 PIT at the half because time had expired when Harrison crossed (just barely) the goal line. Keep the other 59 minutes and 42 seconds of that game exactly the same & Arizona wins either 23-20, 26-20 or 30-20.

59
by Scott Kacsmar :: Thu, 12/19/2013 - 12:40am

But Arizona winning that game would have been another case for the "highly improbable champion" list.

68
by steveNC (not verified) :: Thu, 12/19/2013 - 9:38am

"Keep the other 59 minutes and 42 seconds of that game exactly the same"--really?

So the second half is played independently of what the score is, with the play calling identical? Really?

63
by Edge (not verified) :: Thu, 12/19/2013 - 2:35am

How long do the Rams have to remain in the mix for it to be deemed "long term"? They maintained their success as long as the McCarthy Packers or Payton Saints currently have, that doesn't count as "long term"?

88
by RickD :: Thu, 12/19/2013 - 7:26pm

I would argue that he Rams were an elite team in three seasons: 1999, 2001, and 2003. They had a relatively down year (10-6) in 2000 and the 7-9 2002 season was underwhelming.

I don't think anybody would cite either the recent Packers or recent Saints as teams that have been consistently successful, year to year. Let's make as a minimal qualification for such a status to be "wins division at least two years in a row". Neither the Saints nor the Rams have done that, and while the Packers have, their playoff record is very similar to the Saints (one Super Bowl run surrounded by a many early exits).

89
by theslothook :: Thu, 12/19/2013 - 9:43pm

I think your definition is far too harsh. Were the eagles and ravens successful long term in the 2000s? Both missed the playoffs a number of times but were also there a bunch, just like the saints starting in 2006 when brees arrived. They made in 06, 09, 10, 11, and now 13. That's 5 appearances in a span of 8 years. Packers I would say would also qualify as long term, starting in 08(rodgers first year), they've made it 09,10,11,12,and possibly 13 - that's potentially 5 out of 6 years.

23
by JIPanick :: Wed, 12/18/2013 - 5:32pm

Does a 1-game series really increase the chance of an upset?

Increased variance favors the underdog, sure, but I am not at all convinced that it favors the underdog more than the increased homefield advantage (100% of all games in the series, rather than ~57% in an NBA/NHL/MLB seven game series or 50% in a soccer-style home-away series) favors the (usually stronger) home team.

EDIT: Collecting numbers from whowins.com.
Across all three 7-game series sports, 798 teams hosting a 7-game series won the series, while 795 won game 1. (1187 series)
Removing baseball (because different starting pitchers provide increased game-to-game variance not present in other sports) it's 688 winning the series vs 705 winning game 1. (1029 series)

Some of his numbers don't seem to quite add up so I wouldn't take this as gospel, but I think it's enough to say that "Upsets are more likely in a 1-game series than a 7-game series" is a statement that needs to proven, rather than what we should be assuming by default.

26
by Perfundle :: Wed, 12/18/2013 - 5:40pm

I'm not sure that's the best way to argue that point. Teams in an elimination game don't behave the same way as teams that know they've got up to 6 more games to make up for an opening loss.

47
by JIPanick :: Wed, 12/18/2013 - 7:49pm

I completely agree, and I hope someone smarter and with better information available than me will go back and do the job right.

29
by RickD :: Wed, 12/18/2013 - 6:03pm

It's not an assumption, it follows from a simple mathematical analysis. Let p be the probability of the favorite winning the game. Presuming p > 1/2, the expected number of wins for the favorite in n games is n*p, and the variance is n*p*(1-p), so the std dev is sqrt of above. The longer the series, the more likely the observed winning percentage of the favored team is closer and closer to p.

In a 3-game series, the favored team wins with probability p^2 + 2*p^2*(1-p) =
p^2*(3 - 2*p) > p because p*(3-2p) > 1 for 1/2 < p < 1. So then the odds of winning a 3-game series are better than the odds of winning a single game. I suspect an induction proof could do this for an n-game series.

So this brings us to the question of home field advantage. This is obviously only interesting if the home field advantage is enough to make the series underdog a true favorite at home. But we have to also account for the home field advantage of the favorite. If you postulate an equal effect of home field advantage for both teams (say, that the favorite wins at home with probability p + delta and on the road with probability p - delta), then the expected number of wins actually increases for the favorite with the introduction of the extra variable. So, basically, home field advantage is more likely to flip a game in favor of the favorite than the underdog, simply because there are more games at the favorite's venue.

This isn't rigorous, but I think it suggests what a rigorous proof would be. My approach would be to compare the two distributions.

Of course, if you argue for a case where the series underdog has a much greater home field advantage than the series favorite, and also postulate that p is fairly close to 1/2, then we could easily see a series where the "series favorite" isn't expected to win a majority of the games any longer. But I think that's a fairly contrived counter-example.

44
by theslothook :: Wed, 12/18/2013 - 7:21pm

Ah, but he's saying P is not a constant probability, but now a variable across all games. He's suggesting p is a function of the number of games n. I presume he means if both teams are aware that its a 7 game series, probability might then be p in game 1, then higher for the losing team in p2, then even again in p3. Essentially, its not exchangeable.

62
by RickD :: Thu, 12/19/2013 - 2:25am

If p is wandering all over the place with no rhyme or reason, there's no reason to expect anything.

I think from the example of the NBA, that it's reasonable to model varying p as a home court advantage. And I don't see how, in any case, the favored team would end up with a lower probability of winning a loner series than a single game. Letting p vary from game to game, based on the standing in the series, would just require a more elaborate argument, again, unless somehow there is some factor that will favor the underdog more than the favorite.

64
by theslothook :: Thu, 12/19/2013 - 4:10am

That's fair. Now that I think of it, there's really no compelling reason why p would necessarily change over a long series versus 1 game. Even we by the idea that adjustments happen game to game, both teams theoretically have coaches who adjust.

69
by steveNC (not verified) :: Thu, 12/19/2013 - 9:54am

Agree. As an example, in the rare case that the favored team loses game 1 and, as a result, psychologically collapses ("I can't believe we lost"), making their probability of winning each subsequent game in the series much lower than their probability of winning game 1, then their chance of winning a series can be lower than their chance of winning one game. But this may rarely happen in practice.

70
by steveNC (not verified) :: Thu, 12/19/2013 - 9:57am

Re the NBA: it seems like some of those series go on forever. If so, perhaps this is a "backs to the wall" effect, rather than an "extend the series for more TV money" effect. In any case, it would be interesting to see if the average length of these series is longer than expected, given a simple model with p varying only due to home court advantage.

71
by Jeremy Billones :: Thu, 12/19/2013 - 10:10am

For series that go 7 games, does the home team win game 1 at the same rate they win game 7?

There's a little selection bias there and a much smaller sample size, but that might shed some light on some of these issues.

72
by pm :: Thu, 12/19/2013 - 10:24am

In the NBA, from 1985-2012 and 2nd round or later, Game 7 was won by the home teams at an astonishing rate. 39-8 (.890) record for the home team. That's way better than the record early in the series. Overall the team with Home court advantage for the series wins 75% of the time from the 2nd round or later. I think in the NFL the home team wins like 60% of the time in the playoffs.

77
by dmstorm22 :: Thu, 12/19/2013 - 11:22am

(P-y = probability of home team winning game y in a 7 game series) absolutely does vary from game to game, especially if the conditions of the series entering the game are different.

For instance, in the NBA especially, the home teams wins Game 7 more than they win Games 1,2,5. This probability combining the home team and the series standing is factored into betting lines for each game. People have made a lot of money before Vegas started factoring this (like the swing theory, where a road team is more likely to cover Game 2 if they lost Game 1).

18
by nflenthusiast :: Wed, 12/18/2013 - 5:09pm

I read a lot of assertions that the "best" regular season team (or close) not winning frustrates you, though no reasons why? Is it cognitive dissonance causing mental stress when your model-based expectations don't pan out in reality?

22
by Perfundle :: Wed, 12/18/2013 - 5:24pm

I don't think it has that much to do with their model, because no model would've predicted some of these recent champions. I feel it's more: what's the point of trying your best in the regular season if teams can just muddle their way through and catch fire in the playoffs? You wouldn't want to see that precedent decrease the quality of play, which would happen if a team down by a couple of scores decides it's not worth it to give it their best the rest of the game, what with the risk of injuries and everything, and try again the next game.

25
by JIPanick :: Wed, 12/18/2013 - 5:37pm

Because we want good teams to be rewarded with championships; the Giants, Giants, and Ravens were emphatically NOT good teams, ergo seeing them win Super Bowls is frustrating.

Fans of fair officiating (which everyone at least ought to be) were also done with the Ravens in particular after the pass interference debacle in Denver.

27
by Dennis :: Wed, 12/18/2013 - 5:57pm

Interesting way of looking at it. I want the the teams that I am rooting for to win, regardless of whether they are the best team. I find it frustrating when the teams that I like don't win, and the teams that I don't like do win.

32
by kgjftydytfgj (not verified) :: Wed, 12/18/2013 - 6:27pm

a) Feeling that the best team should win most of the time
b) Feeling upset when one's favorite team loses, regardless of whether that team sucks or not

Those two are not mutually exclusive.

48
by JIPanick :: Wed, 12/18/2013 - 7:53pm

It only applies to me when my favorite team (the Cowboys) aren't involved in the championship picture.

Unfortunately, since it's the Cowboys, that's every year.

65
by Bjorn_ (not verified) :: Thu, 12/19/2013 - 4:51am

Hey, if you can manage to avoid losing to the eagles on a fumbled kneeldown or some other Cowboyesque way of snatching defeat from the jaws of victory you might get your chance this year.

30
by RickD :: Wed, 12/18/2013 - 6:05pm

Come now, the Giants may not have been a good team, but the Giants and the Ravens certainly were.

36
by Hurt Bones :: Wed, 12/18/2013 - 6:40pm

The unusual thing about the Ravens was that they got a significant health benefit at the end of the year. Just as DVOA can’t really get a hold what the Patriots will do without Gronk, it can’t show what adding players who hadn’t played would do to the Ravens performance. It just measures the performance to date. It’s descriptive not predictive. Were the Patriots one of the best teams in the league earlier this year? Yes. Are they now? I don’t think too many people think that they are. The Patriots got worse as the year has gone on.

No team is static. Everyone hopes their team improves, but it doesn’t always work that way. Expecting the Patriots to play the way now that they did before the injuries is stupid. Just as expecting the Ravens to play as poorly as they did with injuries once the injured players have returned. Were the Ravens a great team during the season? No, but they were a good team (8th in DVOA, 3rd in the AFC). They were a better team than that at the start of the playoffs.

41
by Bobman :: Wed, 12/18/2013 - 7:00pm

Fair point and it suggests the Seahawks are actually better than DVOA says, since they've gotten healthier the past 3-4 weeks, particularly on the OL. Whereas teams like the Colts (who were already banged up when they beat the banged up Hawks in Week 6) have gotten more injuries piled up on them as the season progresses. Two teams heading in opposite directions in terms of record, DVOA, and health. There may be a connection!

43
by Perfundle :: Wed, 12/18/2013 - 7:14pm

Regarding Seattle's OL, it's been a mixed bag. It seems that Wilson has better pass protection, although it's still mediocre (just as it was last year), but the run blocking has looked worse. I don't know if this is because the linemen are not entirely healthy yet, or if defenses are specifically keying on Lynch (the Saints did this, but I don't know about the other teams), but the running game outside of Wilson scrambling has not been very good recently.

33
by BBB (not verified) :: Wed, 12/18/2013 - 6:28pm

There was nothing near as egregious as the end of the Saints\Vikings NFCC

50
by JIPanick :: Wed, 12/18/2013 - 7:55pm

That game drove me nuts too, but I was restricting my commentary to the weaker Super Bowl winners of recent years. Saints not in that category.

31
by RickD :: Wed, 12/18/2013 - 6:08pm

People need to recognize that, the larger the tournament, the less likely it is that the "best" competitor will win. And understand that being "the best" and "the champion" are two different things. The Patriots were certainly not the best NFL team in 2001-2002, but they were the champions.

39
by dmstorm22 :: Wed, 12/18/2013 - 6:57pm

I would love if someone did a statistical breakdown of how likely the results of the playoffs from the 70's-80's were, because I think the strong alternate result (the home team/best team winning an overwhelming majority of the time back then) is skewing how we view what happens now.

My guess is it was just as unlikely that all but a handful of the first 30-odd Super Bowl winners being 1 or 2 seeds, as the varied winners we have had in recent years.

A sport with a 16-game schedule and one-game playoff system shouldn't have such dominance by high seeds.

52
by JIPanick :: Wed, 12/18/2013 - 8:15pm

Actually, going back that far, I think people's perceptions of who the best team was are pretty colored by who won the Super Bowl that year. The 90s were the decade where we had the odd run of the best team winning almost every year.

Editorializing a bit, but:

'70 Colts - best team in a very weak conference, basically the Falcons in an Eagles-free 2004 NFC.
'71 Cowboys - probably the best team, at least after picking a QB. Might not win DVOA though.
'72 Dolphins - one of the best teams, '72 was a mess at the top (they were underdogs in the Super Bowl)
'73 Dolphins - at least as good as any other team, probably best of '73
'74 Steelers - Kind of a solid second tier team, think a NE/CAR/NO (but not Seattle/Denver) this year kind of a team (with the John Madden Raiders as the top team)
'75 Steelers - best team of '75
'76 Raiders - It's weird that this is the Raiders team that actually won it, kind of like the '06 Colts. They had only the 5th best point differential in the AFC (waaaaay behind the Steelers and the Bert Jones Colts) and didn't really outclass the NFC elite either.
'77 Cowboys - Probably bottom edge of the top 5, but had the weaker conference all to themselves.
'78 Steelers - Near the top, but behind the Cowboys (who played a nightmare schedule). If Denver beat Seattle this year, it would be similar.
'79 Steelers - Top 2 along with Air Coryell.
'80 Raiders - ho-hum wildcard
'81 49ers - Unimpressive for SB champs, but '81 had no dominant teams. Wouldn't surprise me if they lead the league in DVOA but had the worst 1st place score ever.
'82 Redskins - best of '82
'83 Raiders - Solid division champ, but not on the same level as the Redskins, Dolphins, or 49ers. Imagine the Bengals winning this year.
'84 49ers - best team of '84
'85 Bears - best team of '85
'86 Giants - probably best team of '86
'87 Redskins - fully a peer of the '07 Giants, '11 Giants, and '12 Ravens
'88 49ers - Ho-hum regular season team, but I think Montana was hurt for a while? Best in league '87 and '89.
'89 49ers - best team in league '89

It wouldn't surprise me if fewer than half finished #1 in DVOA, although all but the '87 Skins will likely be above the sudden-run-from-the-middle-of-the-rankings crowd of recent years.

1970 cutoff helps too, because neither the Jets nor the Chiefs were even the best AFL team, and the '67 Packers were on the decline.

53
by dmstorm22 :: Wed, 12/18/2013 - 8:33pm

You may be right. I was mainly talking about seed rather than the point differential/potential DVOA of the team. I think standardizing the argument matters. Are we talking about high seeds not winning, or high DVOA teams?

Anyway, thanks for the work. I definitely don't know as much about 1970-80's football than many others on these boards (though I do recite the Super Bowl Champions to prove to myself I'm not drunk - to mixed results). I think the NFL is what it is now. It's a 16-game season, with injuries and variable performance and one-game playoffs. Upsets will happen. Every Super Bowl Champion outside of the '72 Dolphins lost a game, some to bad teams. Just for a long time not in the playoffs.

The quicker we accept this the quicker we as a society can analyze sports better. If we just agree that the playoffs is random, Joe Flacco doesn't get a $120MM contract for four good games. People don't think Eli Manning is better than Peyton. In so many ways we have gotten smarter as a society in analyzing/talking about sports, but not when it comes to RINGZ. THe quicker we realize the playoffs in every sport is random, and more so the single-elimination NFL, the quicker we can get past easy tropes.

96
by Tom Gower :: Fri, 12/20/2013 - 8:03pm

Pretty much concur with this. I'll be curious to see where the 1987 Redskins and 1988 49ers come out. 1988 seemed like a bit of a mess to me, while of course the strike screwed up 1987. My guess is the Redskins look more like the 2001 Patriots than the Giants/Ravens, but we'll see.

Valid point on the first four Super Bowls. If there was an arbitrary cut-off in the article, it was stopping our thinking at the merger.

34
by Kevin M (not verified) :: Wed, 12/18/2013 - 6:32pm

The 2007 Giants were capable of playing at a high level, but never did so consistently for one reason or another. (The 2011 team was more consistent.) As was mentioned in the article, Eli Manning didn't string together consistent performances until the postseason. And the defense obviously improved too. Perhaps because of those inconsistencies, the Giants lost relatively close regular season games to the Cowboys, Packers, and Patriots... teams they would later defeat in the postseason.

I think the best evidence that 2007 postseason wasn't a fluke was what the Giants did in 2008 with few additions to their roster. They were (I believe) the top rated DVOA team for most of that season until Plaxico Burress shot himself in the leg. That was despite Strahan retiring and Umenyiora not playing a single down.

As far as 2011 goes, the Giants were a quality, yet flawed team elevated by the play of a small group... Manning, Cruz, Nicks, and JPP. Why would anyone be surprised by what they did in the postseason when you consider their regular season matchups with the Packers, 49ers, and Patriots? All 3 of those games went down to the final possession before the games were decided. With the exception of the postseason rematch against the Packers, the other 2 postseason games went down to the final possession too.

The fact that they played close games during the regular season against future postseason opponents shows that not a whole lot was needed to change the outcomes of those games in a rematch. For example, the Giants played better in their 2007 postseason rematch with the Packers than they did in the regular season game. OTOH, the Packers played far worse in the 2011 postseason rematch than they did during the regular season game. The turnover battle played a huge part in the 49ers 2011 regular season win. It also played a huge part in the Giants' postseason rematch victory.

42
by Bobman :: Wed, 12/18/2013 - 7:11pm

The 06 Colts D winning because of Bob Sanders's late-season return has been debunked here if I recall. An equal contributor was probably benched former 1st rounder Rob Morris replacing Gilbert Gardener at Sam in week 15 or so. Two positional upgrades for a D that was, as Dungy claimed, just a half step late or inches out of position to make tackles. (I dunno, coach, I never quote bought that....) I still think it was playoff flukiness as much as anything. They walked into a perfect storm of RBs (KC, Balt) who should have run roughshod all over them and the Colts just stone-walled them. Instead of the predicted 150 yards, Holmes got, what, 30? Just a month after MJD and co ran for 375 on them and said "the only thing that stopped us was the end zone." Harsh, but true. In Balt it's not like the Colts put up huge offensive numbers forcing the Ravens to pass--it was an all-FG-fest (15-6 final score?).

That's why I am just glad to get to the "second season" and let it unfold as it may, without worrying about predictions or making sense of it all. It does make the fan/media/HOF voter fixation on SB rings maddening because getting the Lombardi is more random than people expect, yet having more rings than your neighbor seems to entitle you to a better afterlife. And more ignorant troglodytes loudly chanting about Lombardi-based supremacy. Really, would you rather have Rodgers, or Roethlisberger or Eli? Peyton or Eli? Trent Dilfer or Fouts/Marino? Montana or... uh, where was I going with this again?

49
by Jerry :: Wed, 12/18/2013 - 7:54pm

The idea that home field was arbitrarily rotated rather than based on record before 1975 isn't as absurd as it might seem. Go back to the two-division setup before 1967. Teams played everyone else in their division home-and-home, as well as two games against (arbitrary) teams from the other division. If the bulk of one division was significantly worse than the other, the champions' records would be more likely to reflect the difference in divisions than the relative strength of the teams. So it made as much sense to just rotate the site of the Championship Game between the two divisions as it would to use records, or the result of the All-Star Game.

As more divisions were created, and intradivision games became a smaller part of the total schedule, comparing teams' records made more sense. We're willing to accept that a team can be seeded higher due to the strength of their schedule.

97
by Tom Gower :: Fri, 12/20/2013 - 8:11pm

I see their thinking, but even the current schedule permits similar apparent record mismatches. The Chiefs and Patriots this year have 5 common opponents and 11 dissimilar ones. When the Steelers hosted the 14-0 Dolphins, the majority (8) of Miami's regular season games came against Steelers opponents. That may well have been above-average for the period, and we can get truer comparisons now (e.g. NE-BAL, divisional matchup, same place finisher, 12 of 16) even if it's not necessarily likely. Still seems weird to me.

54
by quest4six (not verified) :: Wed, 12/18/2013 - 8:50pm

I think a couple of things:

1. DVOA I think in general is more reflective of what has happening the past, versus what will happen in the future.
2. Due to parity, teams are so close to one another, so hence whey we see so many one possession SB. These are coin flips by their nature, so introduce more variance
3. DVOA has no concept of team health. Wild card teams that made their runs were significantly more healthier than their counter parts

Even though Aaron won't like to hear this, perhaps the problem is DVOA itself is flawed, so hence why it's not predictive enough.

One last point: Perhaps SB match ups should be measured more along the lines of the last 2-4 games played (excluding resting players situations), versus year long or weighted formula.

55
by Marc (not verified) :: Wed, 12/18/2013 - 8:59pm

It seems to me that a contributing factor might be the following combination: 1) a narrowing of the roster-wide physical talent gap (at least among the top half of the league), due both to salary cap rules and also more data-driven/aggressive scouting methods aided by the explosion of available data in the internet age, combined with 2) more sophisticated schemes that require greater precision/harmony on both sides of the ball for success (rather than a pure talent advantage) than in prior decades. I can imagine that "evolutionary" pressure for the emergence of #2 has arisen in part due to #1. Perhaps Walsh's invention of the West Coast offense and Buddy Ryan's 46 defense are early progenitors of this phenomenon that has only more recently become more prevalent(?).

The second of these factors (more complex coaching schemes) might be responsible for more variance in the playoff results, as in any given game one coaching staff would be more likely (than in the past) to outfox the other in such a way that has profound effects on the game outcome (while a talent gap would be more reliably realized). In a one-game-per-round tournament, this effect wouldn't in general translate to the next game (where the week of preparation, including that week's tape and the different opponent, would allow such advantages to be neutralized in many cases). I acknowledge that lesser teams such as NY and BAL "getting hot" for 4 games in a row flies in the face of this narrative. We are dealing with small samples and multiple factors, though.

I have no data to back up this hypothesis and wouldn't have much of an idea how to quantify the variables involved reliably. Nor have I tried to apply this thinking to specific years/teams. (Also, even if it is a factor, it would be just one among many.) Instead I offer it as fodder for discussion.

90
by Ryan W (not verified) :: Fri, 12/20/2013 - 2:48am

My hypothesis is similar to #2. Something that I presume became a big factor about 15 years ago is technology. Teams are probably starting to have complete access to film of their opponents, advanced statistics, and can come up with much more specific game plans. Way back in the day, it was probably more likely that coaches tried to develop a offensive and defensive strategies that would work for every game. I have to imagine that now it's much more specific to each individual opponent.

I wonder if other sports see a similar change around this time, especially college football (though I'd expect a slight delay tech wise). Though since there isn't a playoff system, that might be harder to assess.

91
by tuluse :: Fri, 12/20/2013 - 3:31am

It was really cool reading Blindside. I actually did care for the parts about Oher, but the history of the left tackle parts were amazing.

Bill Walsh essentially had a guard pull to the outside on passing plays to neutralize LT in a playoff game. He knew it was a one shot deal, once he did it in a game, the Giants would come up with a counter. So he saved the plan all year and waited for the playoffs to unleash it.

92
by Marc (not verified) :: Fri, 12/20/2013 - 1:56pm

Consistent with your comments, I believe opponent-specific planning has been mentioned as a significant contributor to Belichick's long-term success, particularly in more recent years (though I haven't independently tried to assess it). It's pretty clear to me that this type of factor would contribute to playoff variance generally.

60
by fitzy (not verified) :: Thu, 12/19/2013 - 12:50am

My hypothesis -

There has been a growing variance between how officials call the games during the regular season and how they call the game during playoffs. The regular season has become a bit closer to flag football, allowing for high flying offenses. But refs tend to let players play during the playoffs. There is a lot more physicality with receivers in the playoffs, and many more tough hits that just don't get called. Many of the examples of high DVOA busts were examples of teams with high-flying offenses that ground to a halt against defenses that were playing tough and dirty (2001 Rams, 2004 Colts, 2007 Patriots, 2011 Patriots, 2012 Patriots).

61
by nat :: Thu, 12/19/2013 - 12:59am

During the thirteen years the article looks at, top seeds had 26 chances to win two games in a row against playoff teams to advance to the Super Bowl and did so 11 times. If the games were fifty-fifty, they would expect to advance just 6 or 7 times. Top seeds were actually doing very well in their conference playoffs.

So the real question is "Why did top seeds go 2-9 in the Super Bowl when they got there."

Four of the losses were by less than 7 points. One of the losses was to another top seed. So no explanation is needed beyond "shit happens". Better yet, you can say "The Patriots happen" since they were involved in the four close losses by top seeds.

Anyway, it's small sample theatre.

66
by TimK :: Thu, 12/19/2013 - 5:21am

Good call on the KCW this week. The Broncos' consistently awful field position to start drives felt like it was a factor in their defeat. It seems as though the playcalls get more conservative in extremis when backed up, and that a lot of the 3&out drives are starting deep, giving the opponents little yardage to go to get some points.

As a general observation it does seem that there are a lot more fair catches being made now inside the 10 yard line, what happened to the returner waving for a fair catch then getting out of the way and hoping the ball would bounce through into the endzone - that often seemed to work, and slow down the tip-back-&-ground players a bit as well. Is it no longer allowed for some reason, or is all the commentators talk of punts with backspin really meaning that punters have got dramatically better at stopping balls inside the 10 over the last few years? Has anyone got stats on spot of ball landing vs touchback/grounding/fair-catch?

74
by Dr. Mooch :: Thu, 12/19/2013 - 10:57am

As a trivial note, the Bills are sitting Manuel this week and starting Thad Lewis. Not that it would make any difference at all for the pick.

79
by In_Belichick_We... :: Thu, 12/19/2013 - 1:34pm

I didn't read all of the comments here so I apologize if these were already discussed.

There are some comments about the salary cap and I believe this is definitely a factor. Many times it is the least injured team that survives the playoffs. The salary cap likely makes injuries more significant. In years past, the best teams would have back up players that would be starters on other teams. Depth has dwindled due to the salary cap and free agency.
Injuries also skew DVOA ratings. The Pat's offensive DVOA was sub par all year. Gronk and Vareen got healthy and they started clicking. What if these two came back one week before the playoffs? The pre-playoff DVOA wouldn't be a good forecaster of the offensive playoff performance.

I would be willing to bet injuries are related to play off upsets quite often. Either because the favorite recently lost big players or the dog recently got healthy.

I also agree with realignment being a factor. There is a greater chance in having a wildcard team that is flat out better than a divisional winner. Who will the favorites be if SF travels to Philly and Carolina travels to Detroit in the first round of the NFC playoffs?

One more explanation may be modern day game planning and video availability. I'm sure coaching staffs are larger than they used to be. There are coaches specifically for video analysis and game planning. Football is definitely a sport where an inferior team can beat a superior team because of a better game plan. If I remember correctly, the 2002 Combine that with a single elimination tourney and there is potential for many upsets. I doubt you would see very many wild card superbowl champs if each round was 2 out of 3.

80
by Eddo :: Thu, 12/19/2013 - 1:42pm

I wonder if the biggest difference nowadays is the time and effort put in by coaching staffs.

Now that coaches have much better film, it's possible they can more easily adjust to the seemingly "dominant" opponents than they could twenty years ago.

81
by Scott Kacsmar :: Thu, 12/19/2013 - 2:06pm

This top-seed failure thing goes beyond the NFL, by the way.

http://sportsillustrated.cnn.com/2012/writers/the_bonus/05/11/upsets/ind...

82
by Duke :: Thu, 12/19/2013 - 3:20pm

I remember reading the "baseball has the most parity" argument awhile back, too. But quite a few years ago, like in 2007 or so. I think it was from King Kauffman, back when he still wrote sports columns for Salon.

The gist of his argument was that the eventual champion was usually different in MLB, as opposed to the NFL (and even more so the NBA) where the teams that won the championship were often from a small pool. His thought was that because the other sports playoffs had more teams (this was before MLB went to 10 teams), you saw more diversity in playoff teams in other sports. But the top teams were always there, and they tended to win more in the NFL than in MLB.

To be explicit--the Yankees and Red Sox were always in the playoffs in MLB, but they weren't winning all the time. Whereas the Colts, Patriots, and Steelers were in the playoffs all the time, and they won often. If MLB had been bringing in the next 4 teams into the playoffs, they would be changing a lot and you would think the league was more even, even though it wasn't.

It's interesting. The notion that the #9-#12 best teams tend to rotate more frequently, and that causes a perception of parity, has stuck in my mind. I'm not certain I believe it, but I often consider it. The other thing I think is that you don't tend to find longtime bastions of suck in the NFL (read: Royals). I think that contributes to the notion of parity, too--that in the NFL any team is just a year or two from grabbing that #6 seed.

Anyway, I'm rambling, but for my own benefit and curiousity I compiled the list of the last 13 champions* in the two sports:

MLB (9 different):
Red Sox x3
Giants x2
Cardinals x2
Yankees
Diamondbacks
Angels
Marlins
White Sox
Phillies

NFL (8 different):
Patriots x3
Ravens x2
Steelers x2
Giants x2
Buccaneers
Colts
Saints
Packers

*conveniently chosen end point alert

: end devil's advocate

83
by JIPanick :: Thu, 12/19/2013 - 3:36pm

"you don't tend to find longtime bastions of suck in the NFL"

Cleveland doesn't count?

84
by In_Belichick_We... :: Thu, 12/19/2013 - 3:48pm

I think Detroit should be listed before Cleveland.

98
by Duke :: Thu, 01/02/2014 - 4:02pm

Both of those teams have been in the playoffs in the last 12 years, and have had winning seasons in the last 7.

Flukes maybe, but some MLB teams don't even have flukes.

85
by TimK :: Thu, 12/19/2013 - 3:58pm

re Salary cap effect on top seed.

There does seem to be a case that almost all teams have some weaknesses since the salary cap. This seems to sometimes make a kind of scissors/paper/stone effect where some teams have weaknesses against others disproportionate to their 'overall' strength. Having to make more choices and trade-offs in assembling a roster likely leads to more of this than in the pre-salary cap era, and so there is a good chance in a 3 game playoff run of running into at least one team that has the kryptonite for the best team (even though they might be weaker overall, and have lost more games vs other opponents). There is no longer the option to simply throw money at the problem and fix your roster totally, something will always have to be done on the cheap.

Obviously the best teams in the long term waste as little value as possible, but the market for top players is such that you can only find bargains by playing a different scheme for so long.