The Worst Winners
CLICK TO
REACT 
More and more, the disconnect between the regular season and postseason has us surprised when the Super Bowl is decided. Last February, the New York Giants beat an up-to-then undefeated New England Patriots team despite the fact that quarterback Eli Manning led the NFL in interceptions in the regular season. A late-season hot streak, and a dominant performance by their defensive line against Tom Brady and the "greatest offense ever", competed perhaps the most improbable Super Bowl run ever.
Until this year. Because if the Arizona Cardinals win Super Bowl XLIII, they'll be the worst team to do so by the Pythagorean measure. Our friends at Pro Football Reference are using a stat called "Expected Wins", which is based on point differential. Because it's not dependent on play-by-play, and therefore not subject to the availability of play-by-play records per year like DVOA is, Expected Wins can be used for any team in any season.
The stat is based on the Pythagorean Theorem, which was invented by Bill James about a generation ago (read more about the application to football here). James theorized that the true excellence of a team could be better estimated by runs scored and runs allowed then by wins and losses. While it's more conclusive to use this method in a sport that plays 162 regular-season games instead of 16 (sample-size alert...), the Pythagorean Theorem has proven to be accurate enough in football to give an interesting measure of past teams, especially as they match up in specific situations. And according to the Pythagorean Theorem, the Cards have about as much business being in the Super Bowl as Johnny Knoxville has at the Inaugural Ball.
The Cardinals had a positive point differential of 1. That put them at 8.0 Expected Wins, the lowest for any Super Bowl team in a non-strike, 16-game season. Actually, to put it in perspective, no Super Bowl team had 8.0 or fewer Expected Wins from 1966 through 1977, the era of the 14-game season. The 2006 New York Giants (who won) and the 2003 Carolina Panthers (who lost) share the former low mark with 8.6. The 1982 Washington Redskins and Miami Dolphins each entered the Super Bowl with 6.5 Expected Wins in nine-game, strike-shortened seasons. So, when Cardinals fans bemoan the fact that nobody expected their team to be here, they're going to have to cut the world a little slack.
I thought it might be interesting to look at postseason point differential, to see if the turnaround is always obvious. We're including the Super Bowl scores, since they are indicative of those playoff runs. Here are the five biggest upsets, based on these numbers:
Super Bowl XLII (-5.2 Wins)
New York Giants 17 (8.6 Wins), New England Patriots 14 (13.8 Wins)
Postseason Point Differential: New York Giants +20 (85-65); New England +17 (66-49)
Super Bowl XV (-3.0 Wins)
Oakland Raiders 27 (9.6 Wins), Philadelphia Eagles 10 (12.6 Wins)
Postseason Point Differential: Oakland Raiders +46 (102-56); Philadelphia Eagles +11 (61-50)
Super Bowl III (-2.8 Wins)
New York Jets 16 (10.1 Wins), Baltimore Colts 7 (12.9 Wins)
Postseason Point Differential: New York Jets +13 (43-30); Baltimore Colts +35 (65-30)
Super Bowl XLI (-2.8 Wins)
Indianapolis Colts 29 (9.6 Wins), Chicago Bears 17 (12.4 Wins)
Postseason Point Differential: Indianapolis Colts +40 (105-65); Chicago Bears +16 (83-67)
Super Bowl XXXVI (-2.2 Wins)
New England Patriots 20 (10.8 Wins), St. Louis Rams 17 (13.0 Wins)
Postseason Point Differential: New England Patriots +13 (60-47); St. Louis Rams +30 (91-61)
What's interesting here, of course, is that the last two Super Bowl winners are on this list. The 2006 Colts rebounded from the worst run defense in the league to run the table, and the G-Man pulled off quite a miracle last year. Postseason point differential isn't always an indicator, either, as the 1968 Colts can tell you. They decimated the National Football League, only to fall to the AFL Jets. The 2001 "Greatest Show on Turf" Rams had a similar experience when they were upended by a Patriots team that seemed to be easy pickings going in.
If the Cardinals pull off their own upset, and did it by the same score that the Giants put up on the Pats last year (17-14), they would be second on the upset list above. Ah, but look at that postseason point differential. How tempting!
Super Bowl XLIII (-3.8 Wins)
Arizona Cardinals 17 (8.0 Wins), Pittsburgh Steelers 14 (11.8 Wins)
Postseason Point Differential: Arizona Cardinals +36 (112-76); Pittsburgh Steelers +23 (75-54)
It's not out of the realm of possibility for the Cardinals to pull this off. And if they do, they'll be the third straight "What are you doing here?" team to do so, leaving analysts searching for answers as the gap between the two seasons continues to grow.
By
Doug Farrar
|
January 22, 2009; 1:37 PM ET
| Category:
Doug Farrar
,
Statistics
Save & Share:
Previous: DVOA Matchup: Conference Championship Edition |
Next: DVOA Matchup: Super Bowl Edition
Posted by: albinoflea | January 27, 2009 1:27 PM
Doug can crunch all the numbers he wants. The reality is, the team that's best by the end of the season is the winner. Who can outlast their "better" opponents and maintain their momentum until the bitter end? We'll see.
Posted by: emayar | January 24, 2009 12:43 PM
The computer doesn’t select who is going to play in the BCS. Its computer models and several polls. The results would be better if it was just the computer models.
By the way, Pythagoras came up with the Pythagorean Theorem. Probably Bill James just came up with an application of it. However any model probably oversimplifies reality. For example, the BCS models take into account the opponent’s ability. If you add this to the current prediction the Cardinals are even less likely to win.
On the other hand, all of these models assume the data is stationary (i.e., things do not change over time). Perhaps Ben Roethlisberger’s latest concussion (or other recent injuries) will be a factor. Then there is always a certain degree of randomness in sports- a missed tackle here, a ref who doesn’t see a hold there, or even a last minute 52 yrd field goal that could go either way.
Posted by: hal77 | January 24, 2009 11:14 AM
Shall we do it like college football and have the computer select who will play in the Super Bowl. GIVE ME A BREAK. The real issue is the intensity of play which takes a huge toll on teams throughout the season. That means the depth and experience of the backups is critical for teams at the end of the season.
Posted by: jnaway | January 23, 2009 6:00 PM
The comments to this entry are closed.
Doesn't add up...
For the Super Bowl XLIII numbers: Pittsburgh Steelers +23 (75-54)... if the 75-54 stands, this should be +21, not +23...
But when I do the math, (23-14 vs. SD, 35-24 vs. B'more, and a proposed 14-17 vs. AZ), I wind up with +17 (72-55)...
The math for AZ seems fine.