Writers of Pro Football Prospectus 2008

Most Recent FO Features


» The Deep Ball Project

Guest columnist John Kinsley breaks down the tape of every deep pass in the NFL in 2017 and comes away with a shocking conclusion: even without Andrew Luck, the Colts had the best long-ball quarterback in the league.

17 Oct 2008

Varsity Numbers: A New Day or an Overreaction?

We're going to take a week off from the wonkiness of S&P, "+" Numbers, and Win Correlations to address an ongoing theme in college football in the wake of yet another Saturday shakeup: Have we really entered a new level of parity in college football, or has this all happened before? Is it a new age of evenness, or is it part of a cycle?

The last year and a half of college football have been amazingly unpredictable. This year has already seen an "unbeatable" No. 1 team (USC) lose to a team with a losing record (Oregon State) and two top 5 teams (Florida and Missouri) lose at home to decent-sized underdogs. The season has shaped up as a decent encore to last year's unprecedented unpredictability. After years where the top teams seemed on a crash course for the title game from beginning to end (Oklahoma and USC in 2004, Texas and USC in 2005, for example), it has been hard to keep up with the insanity.

But is this actually a sign of a "new age" of college football, or is this just part of another cycle? With the ESPN College Football Encyclopedia in hand, we can attack this question in a couple of different ways. First, with last year's upset specials and two-loss national champion in mind, we can look at how many losses teams near the top of the polls have had from year to year. Second, we can look at how many undefeated teams have survived, on average, seven weeks into the season. Until a deep pool of in-depth data is compiled, these are probably the best ways to look at this question.

Total Combined Losses in the AP Top 10 (End of Regular Season)*
1 2003 18
2 2007 17
1990 17
4 1967 16.5
5 1974 15.5
6 2002 15
1999 15
1984 15
9 1980 14.5
1965 14.5
* This list looks at every season since 1960. Before 1996, there existed the possibility of a tie. For the purposes of this list, a tie is considered 0.5 losses.

The 2007 regular season ended with one undefeated team (Hawaii), two one-loss teams (Ohio State, Kansas), and that is it. Heading into the bowl season, the AP Top 10 had combined for 17 losses, and as you see, only one year since 1960 had seen more than that. That year was also from this decade (2003). Is this part of an upward trend? Since 1960, the average number of losses for teams in the top 10 has been 12.2; in the last nine years, that average has climbed to 13.9. That average would be even higher if not for the outlier that was 2004, which produced a whopping five unbeaten teams (USC, Oklahoma, Auburn, Utah, and Boise State) in the regular season. So we may have something here.

Undefeated Teams in the AP Poll after Week Seven
1 1977 1
2 1978 4
3 1991 5
1976 5
1972 5
6 2007 6
1981 6
8 10-way tie (including 2006, 2005, and 2000) 7
18 2008 9

This table suggests that the number of remaining undefeated teams in 2008, despite the interesting upsets, is not much different than in previous years or decades, nor was the number in 2007. (Of course, upsets can happen at any point in the season -- we could just be getting started on another upset binge.) But the strength of teams' non-conference schedules were much tougher in the 1970s (represented well in the table), so that might explain the lack of undefeateds in that time period.

What has made the last couple of seasons so interesting is, in part, what has seemed like a large number of unranked teams knocking off those in the top 5. For instance, a few weeks ago USC became the first No. 1 team in 18 years to lose to a team with a losing record. Last year, as the No. 2 team in the country, they lost at home to 1-3 Stanford. Is that something more unique to more recent seasons?

So far in 2008, this level of upset has happened twice:

  • Oregon State 27, No. 1 USC 21
  • Ole Miss 31, No. 4 Florida 30

Last year, it happened a ridiculous 13 times:

  • Appalachian State 34, No. 5 Michigan 32
  • Colorado 27, No. 3 Oklahoma 24
  • Auburn 20, No. 4 Florida 17
  • Stanford 24, No. 2 USC 23
  • Illinois 31, No. 5 Wisconsin 26
  • Oregon State 31, No. 2 California 28
  • Rutgers 30, No. 2 South Florida 27
  • Florida State 27, No. 2 Boston College 17
  • Illinois 28, No. 1 Ohio State 21
  • Arizona 34, No. 2 Oregon 24
  • Texas Tech 34, No. 4 Oklahoma 27
  • Arkansas 50, No. 1 LSU 48
  • Pittsburgh 13, No. 2 West Virginia 9

Thirteen is an outlier, without a doubt. How many times in the last 40 years has this sort of upset happened more than even five times in a season?

1974 (six times)

  • Wisconsin 21, No. 4 Nebraska 20
  • Purdue 31, No. 2 Notre Dame 20
  • Kansas 28, No. 5 Texas A&M 10
  • Missouri 21, No. 5 Nebraska 10
  • Michigan State 16, No. 1 Ohio State 13
  • SMU 18, No. 5 Texas A&M 14

1981 (seven times)

  • Wisconsin 21, No. 1 Michigan 14 and Georgia Tech 24, No. 2 Alabama 21 in the same weekend
  • Clemson 13, No. 4 Georgia 3
  • Arizona 13, No. 1 USC 10
  • Arkansas 42, No. 1 Texas 11
  • South Carolina 31, No. 3 North Carolina 13
  • Miami-FL 17, No. 1 Penn State 14
  • Washington 13, No. 3 USC 3

1982 (six times)

  • LSU 24, No. 4 Florida 13
  • Tennessee 35, No. 2 Alabama 28
  • Stanford 43, No. 2 Washington 31
  • Notre Dame 31, No. 1 Pittsburgh 16
  • Baylor 24, No. 5 Arkansas 17
  • Washington State 24, No. 5 Washington 20

1984 (seven times)

  • Miami-FL 20, No. 1 Auburn 18
  • BYU 20, No. 3 Pittsburgh 14
  • Washington 20, No. 3 Michigan 11
  • Syracuse 17, No. 1 Nebraska 9
  • Purdue 28, No. 2 Ohio State 23
  • Kansas 28, No. 2 Oklahoma 11
  • Navy 38, No. 2 South Carolina 21

1985 (seven times)

  • Baylor 20, No. 3 USC 13
  • Tennessee 38, No. 1 Auburn 20
  • Arizona 28, No. 3 SMU 6
  • Illinois 31, No. 5 Ohio State 28
  • Miami-FL 27, No. 3 Oklahoma 14
  • Texas 15, No. 4 Arkansas 13
  • Wisconsin 12, No. 3 Ohio State 7

1990 (seven times)

  • Oregon 32, No. 4 BYU
  • Stanford 36, No. 1 Notre Dame 31
  • Michigan State 28, No. 1 Michigan 27
  • Texas 14, No. 4 Oklahoma 13
  • Alabama 9, No. 3 Tennessee 6
  • UCLA 25, No. 2 Washington 22
  • Hawaii 59, No. 4 BYU 28

2002 (eight times)

  • Louisville 26, No. 4 Florida 20
  • Pittsburgh 28, No. 3 Virginia Tech 21
  • Boston College 14, No. 4 Notre Dame 7
  • Florida 20, No. 5 Georgia 13
  • Texas A&M 30, No. 1 Oklahoma 26
  • Texas Tech 42, No. 5 Texas 38
  • Washington 29, No. 3 Washington State 26
  • Oklahoma State 38, No. 3 Oklahoma 28

There was a tremendous power shakeup in the early '80s, but since then the strings of upsets have been few and far between.

In all, the evidence of a trend toward a new level of parity isn't tremendously strong. We haven't seen yet whether last year's upsets were an outlier or a bellwether. But there is an argument to be made for a trend toward parity. Thousands of words could be spent explaining why this may be happening, but I will offer a few thoughts and move on for now.

  • The spread offense has caught fire at the high school level, meaning even teams who don't get blue-chip recruits can find players who are polished and well suited to their style of play, allowing them to compete at a higher level.
  • Big-time football camps are almost as prevalent as those for basketball at this point, and more high schoolers are getting better training before they reach college, resulting in a deeper pool of high school talent. This adds heft to the impact of the 85-scholarship limit and the parity it was intended to cause.
  • Some of the country's national powers from the 1990s (or perhaps even the 1980s) have faded and fallen from grace, resulting in a major power shift. Previously great programs like Florida State, Nebraska and (until this season) Penn State have trended downward in recent years. Virginia Tech and Wake Forest now appear to be the class of a weak ACC, Penn State was a few steps below Ohio State and Michigan until this year's shakeup of power, and Nebraska has at least briefly ceded control of the Big 12 North, first to Colorado and then to Missouri and/or Kansas.

Meanwhile, emerging powers took at least a small step backwards recently. USC was fantastic from 2003-2005, and there's no questioning that the 2005 Texas team was unbelievable. But in the last three years, USC has not been able to avoid random setbacks, and Texas has returned to their pre-Vince Young level of good-but-not-unbeatable stature. Miami-FL was outstanding from 2000-2002, but Butch Davis jumped to the professional ranks, and the program tailed off a couple of seasons later. Oklahoma, too, has yet to quite reattain (and hold onto) the high level of play that it experienced from about 2002 to 2004.

And beyond all this, there is the cannibalism that takes place in the SEC on a yearly basis. The conference is competitive enough that a true dominant power cannot emerge from there for more than a year or two at a time. The same thing might be starting to happen in the Big 12 as well.

But I'll stop there -- otherwise this column could reach 10,000 words quickly. The bottom line is that upsets are exciting, and the lack of a true above-all-others powerhouse only increases the public outcry for a playoff. The powers that be have resisted that outcry so far, but it is only a matter of time, is it not?

Posted by: Bill Connelly on 17 Oct 2008

13 comments, Last at 20 Oct 2008, 9:27am by mrh


by Ken (not verified) :: Fri, 10/17/2008 - 2:14pm

I'd argue that a part of the reason for more losses is that schedules are bigger, all teams in the top 10 play 12 games and usually a few will play 13, in the 60's I think they played 10 games pretty often, so winning percetage is an important factor.

by mm (not verified) :: Fri, 10/17/2008 - 3:02pm

In addition to more games overall, as the above commenter said, there are more difficult games.

With the BCS bringing computer polls (and previously a strength of schedule component that has since been eliminated), most schools now schedule at least one good non-conference team to their schedule every year to give them a better chance at the BCS title game, as well as a better chance at an at large bid if they miss their conference title. Journalists have also become more aware of strength of schedule arguments in recent years (which is a big reason why LSU made it into the title game), but I think that's because the BCS forced them to think more deeply about ranking over the years.

Plus, many conferences went to 8 conference games with expansion. The number of 'easy' games are way down.

Lastly, players are bigger and faster at all levels now, and there are probably more injuries. With scholarship limits, each team has fewer backups. With less 'easy' games, those backups have fewer opportunities to play in blow-outs to get experience. So when they have to come in and play for an injured starter, we should expect a greater drop-off in play than in previous years.

by Pat (filler) (not verified) :: Fri, 10/17/2008 - 7:49pm

Previously great programs like Florida State, Nebraska and (until this season) Penn State have trended downward in recent years.

Penn State was "down" for 4 years - 2000, 2001, 2003, and 2004. Prior to that, they finished ranked #11, #17, #16, #7 - you get the picture, basically they were a perennial Top 25 team. In 2002 they finished ranked #16, 2005 they finished ranked #3, 2006 and 2007 they finished ranked #24 and #27(ish).

For a better look at it, here are Colley's rankings for Penn State at the end of the year, from 1998 to now: 14, 9, 69, 51, 18, 87, 76, 3, 25, 28. The "turnaround year" was 2005, not this year. Being ranked at the bottom of the Top 25 was a bit low for Penn State historically, but we're not talking anywhere near as bad as they were in the four years of hell. No one in the Big Ten wrote them off over the past two years.

by davis21wylie :: Sat, 10/18/2008 - 6:04pm

Here's one way to show that there's more parity than ever... I took every D-IA season since the modern era began in 1946 and calculated each team's pythagorean winning % using pythagenpat exponents. Then I ran the standard deviation of PW% for each season as a very broad measure of parity. Here are the bottom 20 seasons by standard deviation:

Year Stdev
2007 0.238068561
2004 0.245071326
1992 0.246136584
1984 0.250641477
2005 0.250953528
1999 0.253139789
2001 0.253909272
2002 0.256295550
1994 0.259195886
2006 0.260942159
1987 0.263416823
1995 0.263590607
2003 0.265070993
1986 0.265202655
1998 0.266722808
1993 0.266949135
1985 0.267136213
1990 0.269452404
2000 0.270735849
1979 0.271123870

I'm seeing a lot of post-2000 seasons in the top 10. I didn't even include 2008 in the study, but it's clear that the gap between the best and worst teams in the country is smaller than it's ever been before. People thought last year was an aberration, but in fact it's the new reality. The current BCS system would have "worked" for the most part 20 years ago, but now there are too many good-but-not-great teams out there (heck, it's reached a point where even a plus-one won't do the trick anymore). So, yes, all of this definitely points to a new age of evenness in college football.

by davis21wylie :: Sat, 10/18/2008 - 10:09pm

Then again, here's some more interesting data that might suggest the contrary...

I ran year-to-year correlations on PW% for each school since 1950 -- in other words, how does performance in one season correlate to performance the following season? I broke it down by decade:
1950s 0.481516840
1960s 0.564673397
1970s 0.592097378
1980s 0.622650291
1990s 0.638047643
2000s 0.658697335

In my mind, the concept of parity extends beyond the the fact that the best and worst teams are more evenly matched, but also to the idea that schools which aren't good can turn their programs around in a short period of time, provided they have competent recruiting and coaching. This data suggests that while the gap between the haves and have-nots is smaller than ever, there is also a visible trend toward more year-to-year consistency in performance among schools -- in other words, good programs and bad programs tend to stay that way now more than ever before, which would actually be an argument against the existence of greater parity.

by Pat (filler) (not verified) :: Sat, 10/18/2008 - 10:58pm

Parity always needs a scale. In-season parity and year-to-year parity are the two you're listing (scale less than a year, scale = 1 year), but the parity that most college football fans are thinking of is probably in-season parity. Whether or not multiyear parity is increasing (i.e. winning percentage on, say, a 5-year span) is probably more interesting, though.

by Pat (filler) (not verified) :: Sat, 10/18/2008 - 10:55pm

The only way the BCS "doesn't work" is if there are more than 2 undefeated teams out there, not multiple one-loss teams. "Good but not great" teams (i.e. teams that have lost) always had a way to get in - win that game.

by CuseFanInSoCal (not verified) :: Sun, 10/19/2008 - 1:16pm

Wrong. The BCS only works if
1) The teams with the two best won-loss records in college football are from major conferences
2) Both teams won their conference
3) No one else has the same won-loss record as the #2 team

If all of these conditions are not satisified, there will be significant contraversy over who plays in the BCS title game, though it will be downplayed somewhat if the teams in the BCS title game are #1 and #2 in both major polls, and forgotten by many as time passes. But that doesn't mean it didn't happen.

1998 - FSU, Wisconsin, Ohio State, and UCLA were all 10-1 (and as a Cuse fan, I'm required to maintain that without a phantom pass interference call, BCS champion Tennessee would have been 11-1); FSU's title game spot was somewhat arbitrary
1999 - Exactly two major-conference undefeated teams. No problems.
2000 - FSU, Miami, and Washington all had one loss; choosing FSU to lose to Oklahoma was again, somewhat arbitrary
2001 - Oregon and Maryland also had one loss, like BCS title game loser Nebraska, but they won their conference. Both Oregon and Big 12 champ Colorado were ranked above Nebraska in the polls. Choosing Nebraska wasn't just arbitrary, it was silly.
2002 - Like 1999, exactly two undefeated major-conference teams. Worked.
2003 - Another case of a Big 12 non-champion in the BCS title game, this produced the only split title (among the major polls) in the BCS era. Outside of SEC country, this is probably the most contraversial year, because USC was #1 in both major polls, but was not selected for the title game, and Oklahoma, #3 in both major polls and coming off a Big 12 title game loss, was selected.
2004 - Three BCS conference teams and a Utah team that would win its BCS bowl game handily all finished the regular season undefeated (so did one other team that I can't remember). Since USC and Oklahoma were #1 and #2 going in, no one argues this much outside of SEC country.
2005 - Third time the BCS lucked out with exactly two major conference unbeatens
2006 - Because Florida won the BCS title game handily, it's easy to forget that Michigan, Wisconsin, and Louisville were also one-loss major conference teams; it was Florida's selction that was contraversial, not Ohio State's.
2007 - Again, LSU's victory over Ohio State shouldn't wipe this year from contraversy. Ohio State was, again, the only easy pick. Kansas had a better record than LSU (although they avoided playing Oklahoma and Texas, and failed to make the Big 12 title game when they lost to Missouri). Hawaii was undefeated, albeit on a paper-thin schedule. Georgia, Oklahoma, USC, West Virginia, Missouri and Virginia Tech also had two losses in major conferences.

by War Eagle (not verified) :: Sun, 10/19/2008 - 7:28am

Sorry Pat, but that's nonsensical. What is the difference in picking between >2 undefeated and >2 one loss teams? Arbitrarily picking 2 of 3 one loss teams and saying to the other, "you should have won the game you lost" is hardly "working."

by Madison (not verified) :: Sun, 10/19/2008 - 2:32pm

Isn't one of the biggest ways of solving the problem demanding that teams are winners of their own division before they can play for a national championship? That seems somewhat uncontroversial as a proposition to me...

by War Eagle (not verified) :: Sun, 10/19/2008 - 3:22pm

A good system should have the following features:

1)Everyone should have the possibility of winning the championship before any games are played.

2)The criteria that determine the champion (and who plays for the championship) should be fair, objective, and easy to understand.

3)It is possible for only one team to have fulfilled above criteria at the end of the year.

The BCS fulfills one of these criteria. In other words, the only thing the BCS is good for is making sure that there is only one "champion." It, in practice, excludes many teams from the possibility of winning and chooses it's champion with a component that is subjective and an component that is objective but convoluted. It's stupid.

The fix would be easy too; that's the most ridiculous thing of all. Simply play a tournament of conference champions and all three criteria are met.

by CuseFanInSoCal (not verified) :: Sun, 10/19/2008 - 4:35pm

Well, you need at least one at-large spot or independents can't get in. Plus an 11-team tournament looks wierd. Also, until/unless the Big Ten adds a 12th team and a championship game or drops a team and plays a Pac 10 style schedule, a tie between teams that didn't play each other is a very real possibility.

So my preference is for a 16-team tournament.

11 conference champs
5 at large teams
At-large teams and seeding determined by a basketball-style selection committee
Divided into an East and West bracket
1st and 2nd rounds played at home site of higher seed
Semis played at regional sites
Final played at rotating location

by mrh (not verified) :: Mon, 10/20/2008 - 9:27am

One thing that affected parity and competitive balance was the gradual desegragation of college football. As teams began recruiting African-american players, power shifted around. The dominant Oklahoma teams of the '50s had their first balck player in '56, Alabama didn't until the early '70s. It is not surprising that teams showed more year-to-year consistency after this process completed, roughly in the mid-70s.