## 0 for 3

Posted by Doug on January 19, 2007

The quarterback Petyon Manning is most often compared to is Dan Marino. This Sunday, Manning is hoping to avoid joining a club whose membership includes Marino and another future Hall of Famer: Brett Favre. Oh yeah, Bernie Kosar too.

Marino's, Favre's, and Kosar's teams all lost three playoff games to the same team led by the same quarterback. Favre's Packers lost three times in four years to a Cowboy team led by Troy Aikman (a quarterback who is often compared to Tom Brady). Marino's Dolphins lost to Jim Kelly and the Bills three times between 1990 and 1995. Kosar's Browns lost three championship games in four years against John Elway-quarterbacked Denver teams.

My individual playoff game database only goes back to 1975, so this is an incomplete list, but here is a list of pairs of quarterbacks who have faced each other more than once (since 1975) in the playoffs. In each case, the quarterbacks are listed alphabetically and the games are listed home-team-last.

Jim Kelly vs Dan Marino (3)

1990 d: mia 34, buf 44

1992 c: buf 29, mia 10

1995 w: mia 22, buf 37Brett Favre vs Steve Young (3)

1995 d: gnb 27, sfo 17

1997 c: gnb 23, sfo 10

1998 w: gnb 27, sfo 30John Elway vs Bernie Kosar (3)

1986 c: den 23, cle 20

1987 c: cle 33, den 38

1989 c: cle 21, den 37Troy Aikman vs Brett Favre (3)

1993 d: gnb 17, dal 27

1994 d: gnb 9, dal 35

1995 c: gnb 27, dal 38Troy Aikman vs Steve Young (3)

1992 c: dal 30, sfo 20

1993 c: sfo 21, dal 38

1994 c: dal 28, sfo 38Tom Brady vs Peyton Manning (2)

2003 c: ind 14, nwe 24

2004 d: ind 3, nwe 20Joe Montana vs Phil Simms (2)

1984 d: nyg 10, sfo 21

1985 w: sfo 3, nyg 17Peyton Manning vs Jake Plummer (2)

2003 w: den 10, ind 41

2004 w: den 24, ind 49Drew Bledsoe vs Kordell Stewart (2)

1997 d: nwe 6, pit 7

2001 c: nwe 24, pit 17Jeff Kemp vs Phil Simms (2)

1984 w: nyg 16, ram 13

1986 d: sfo 3, nyg 49Donovan McNabb vs Michael Vick (2)

2002 d: atl 6, phi 20

2004 c: atl 10, phi 27Terry Bradshaw vs Roger Staubach (2)

1975 s: pit 21, dal 17

1978 s: pit 35, dal 31Jeff Hostetler vs Jim Kelly (2)

1990 s: nyg 20, buf 19

1993 d: rai 23, buf 29Steve Bartkowski vs Danny White (2)

1978 d: atl 20, dal 27

1980 d: dal 30, atl 27Dan Marino vs Joe Montana (2)

1984 s: sfo 38, mia 16

1994 w: kan 17, mia 27John Elway vs Warren Moon (2)

1987 d: hou 10, den 34

1991 d: hou 24, den 26Troy Aikman vs Randall Cunningham (2)

1992 d: phi 10, dal 34

1995 d: phi 11, dal 30Peyton Manning vs Steve McNair (2)

1999 d: ten 19, ind 16

2006 d: ind 15, bal 6Pat Haden vs Fran Tarkenton (2)

1976 c: ram 13, min 24

1978 d: min 10, ram 34Terry Bradshaw vs Ken Stabler (2)

1975 c: oak 10, pit 16

1976 c: pit 7, oak 24Danny White vs Doug Williams (2)

1981 d: tam 0, dal 38

1982 w: tam 17, dal 30Pat Haden vs Roger Staubach (2)

1976 d: ram 14, dal 12

1978 c: dal 28, ram 0Stan Humphries vs Dan Marino (2)

1992 d: sdg 0, mia 31

1994 d: mia 21, sdg 22Trent Green vs Peyton Manning (2)

2003 d: ind 38, kan 31

2006 w: kan 8, ind 23Dave Krieg vs Dan Marino (2)

1983 d: sea 27, mia 20

1984 d: sea 10, mia 31Terry Bradshaw vs Dan Pastorini (2)

1978 c: hou 5, pit 34

1979 c: hou 13, pit 27Jim McMahon vs Phil Simms (2)

1985 d: nyg 0, chi 21

1993 w: min 10, nyg 17Brad Johnson vs Donovan McNabb (2)

2001 w: tam 9, phi 31

2002 c: tam 27, phi 10Joe Montana vs Wade Wilson (2)

1987 d: min 36, sfo 24

1988 d: min 9, sfo 34Vince Ferragamo vs Danny White (2)

1980 w: ram 13, dal 34

1983 w: ram 24, dal 17Mark Brunell vs John Elway (2)

1996 d: jax 30, den 27

1997 w: jax 17, den 42

My coaching database is much more complete than the playoff game database, but isn't 100% complete. I may be missing some guys from pre-1950, but here is a list of all pairs of coaches who have faced each other three or more times. A quick scan reveals that the company Tony Dungy is hoping to avoid consists of George Allen (who lost three times to Bud Grant), Dick Nolan (who lost in three straight years to Tom Landry), Don Shula (0-3 against Marv Levy), Marty Schottenheimer (0-3 against Don Shula), and Chuck Knox (0-3 against Bud Grant). Is anyone else surprised that Chuck Noll and Don Shula only faced each other three times in the playoffs?

Bud Grant vs Tom Landry (5)

1968 p: dal 17, min 13

1971 d: dal 20, min 12

1973 c: min 27, dal 10

1975 d: dal 17, min 14

1977 c: min 6, dal 23John Madden vs Chuck Noll (5)

1972 d: oak 7, pit 13

1973 d: pit 14, oak 33

1974 c: pit 24, oak 13

1975 c: oak 10, pit 16

1976 c: pit 7, oak 24George Halas vs Steve Owen (4)

1933 c: nyg 21, chi 23

1934 c: nyg 30, chi 13

1941 c: nyg 9, chi 37

1946 c: chi 24, nyg 14George Allen vs Bud Grant (3)

1969 d: ram 20, min 23

1973 d: was 20, min 27

1976 d: was 20, min 35Tom Landry vs Dick Nolan (3)

1970 c: dal 17, sfo 10

1971 c: sfo 3, dal 14

1972 d: dal 30, sfo 28John Madden vs Don Shula (3)

1970 d: mia 14, oak 21

1973 c: oak 10, mia 27

1974 d: mia 26, oak 28Paul Brown vs Raymond Parker (3)

1952 c: det 17, cle 7

1953 c: cle 16, det 17

1954 c: det 10, cle 56Tom Landry vs Ray Malavasi (3)

1978 c: dal 28, ram 0

1979 d: ram 21, dal 19

1980 w: ram 13, dal 34Chuck Noll vs Don Shula (3)

1972 c: mia 21, pit 17

1979 d: mia 14, pit 34

1984 c: pit 28, mia 45Bill Belichick vs Bill Cowher (3)

1994 d: cle 9, pit 29

2001 c: nwe 24, pit 17

2004 c: nwe 41, pit 27Marv Levy vs Don Shula (3)

1990 d: mia 34, buf 44

1992 c: buf 29, mia 10

1995 w: mia 22, buf 37Curly Lambeau vs Steve Owen (3)

1938 c: gnb 17, nyg 23

1939 c: nyg 0, gnb 27

1944 c: gnb 14, nyg 7Bill Parcells vs Bill Walsh (3)

1984 d: nyg 10, sfo 21

1985 w: sfo 3, nyg 17

1986 d: sfo 3, nyg 49Marty Schottenheimer vs Don Shula (3)

1985 d: cle 21, mia 24

1990 w: kan 16, mia 17

1994 w: kan 17, mia 27

Bud Grant vs Chuck Knox (3)

1974 c: ram 10, min 14

1976 c: ram 13, min 24

1977 d: min 14, ram 7Blanton Collier vs Tom Landry (3)

1967 d: cle 14, dal 52

1968 d: dal 20, cle 31

1969 d: cle 38, dal 14Mike Ditka vs Joe Gibbs (3)

1984 d: chi 23, was 19

1986 d: was 27, chi 13

1987 d: was 21, chi 17Chuck Knox vs Tom Landry (3)

1973 d: ram 16, dal 27

1975 c: dal 37, ram 7

1976 d: ram 14, dal 12

Another excellent set of figures you'll find nowhere else. Good work, Doug ðŸ™‚

===

On a different note, I have defined a Super Bowl "epoch" as a set of Super Bowls marked by the following characteristics:

An epoch is officially underway when teams from the previously losing conference win the Super Bowl 2 years in a row. When the other conference wins just 1 time, it is treated as an exception. If it happened twice in a row, a new epoch would be underway.

Here's what we have had so far in SBs I - XL (a = AFL/AFC, N = NFL/NFC):

Epoch 1: NN (NFL 2-0)

Epoch 2: aaaNaaaaaNaaa (AFL/AFC 11-2)

Epoch 3: NNaNNNNNNNNNNNNN (NFC 15-1)

Epoch 4: aaNaaNaaa (current) (AFC 7-2)

=

I am a little surprised about the relatively few of these epochs that have occurred in 40 years, and about how long they usually last. From this, it seems that Super Bowls tend to run in long streaks, and that conferences tend to get an upper hand for long periods, even though the specific teams are different from year to year. Two questions:

1. What causes this?

2. Mathematically, is this different than what should be expected?

Also, it's interesting there were no ambiguous periods, such as aNaN or NaNa.

Anybody got any thoughts?

Part of the cause of an epoch is single strong teams. New England is responsible for 3 of the AFC's current 7-game epoch, and Denver is Responsible for 2 of them.

But I have to think that more than anything, these long/few epochs are mathematical anomolies due to small sample size.

From 1990-1993 the Buffalo Bills were probably the best NFL team over that 4-year span. The NFC just happened to have a better representative in 3 of those years, and got lucky in the 4th year. So the fact that the NFC had an epoch during the period that an AFC team was the best team has to be an unusual occurance.

I dont think it is any more than a coincidence but it is really interesting that there have not been any NaNa or aNaN series. Especially in the NFL where there is so much "parity."

Interesting observation, Bill M. Someone definitely should check whether it's statistically significant. Also does it hold true for other measures of conference strength, e.g. interconference games. If it does, I'd guess it results from competition breeding a better product, as there's less on-field competition between the conferences (the strong tend to stay strong). And what about MLB prior to interleague play? NCAA conferences?

I'm don't think there's much about the Bill M epoch in th statistical literature :). But there is a thing (a few things, actually) called a "runs test" that tells you if a list of 0s and 1s is streakier than random chance would dictate. If you look at the As and Ns above, you'll see seven runs of As (and also seven runs of Ns). Is that too few? How many runs of As would be expected if this were a fair coin?

I really don't know much about statistics, but I have a few books on my shelf. this one says that, under 50/50 randomness, we should expect 10.25 runs of As in a string of 40, and that the standard deviation should be about 1.6. So our 7 runs are almost exactly two standard deviations below expected.

But I don't know if the distribution of runs is approximately normal, or what. So I don't know how unlikely it is to be 2 standard deviations from the mean.

The R statistical package, which by the way is a really excellent piece of free software, has a runs test built in. I was unable to determine exactly what was going on behind the scenes, but I typed in the super bowl sequence and it gave me back a p-value of about .02. In other words, if you flipped a fair coin 40 times, you'd get something this streaky (in some sense not exactly the same as Bill M's) only 2% of the time.

I don't know how to test the Bill M epoch streakiness except by simulation. Run a billion trials of 40 fair coin flips, count the number of epochs in each one, and see how rare it is to only have four epochs. I'll do that if I get a chance in the next few days.