Penn State's No. 1 hopes fade again

Penn State is my adoptive big time football school to cheer for - it's in the region, the coach is an Ivy League alum, decent academics, just about the only standard bearer of eastern football excellence except the occasional foray into the top 20 for a year at a time by BC, Pitt, Syracuse, now Rutgers. My hope was that:

- Penn State would finish the year unbeaten. Which didn't happen with the loss to Iowa Saturday.
- Penn State would make it to the BCS title game. Which might not have happened because there always seems to be two teams ahead of Penn State in the BCS polls.
- Give Joe Paterno a reason to leave without appearing to be pushed.

Has any other top twenty football team been shafted more in the polls over the coach's tenure than Penn State? Another reason to see a Cornell connection.

The Big Ten is a weak conference; the annual drilling of Ohio State in the title game has become boring. Even if Penn State ran the table this was the wrong year to think that the team was getting screwed.

I'm a PSU fan, too (My dad and uncle are grads). So the loss last night was tough.

But their strength of schedule is weak. I saw it's something like 23rd strongest of the BCS teams. This is mostly because the Big ten is weak, but they schedule some serious cupcakes before the Big Ten schedule starts, too. (Coastal Carolina? Temple? Please!)

As of November 11:

The Big XII is that much stronger than the rest of the NCAA? They must have an amazing OOC record.

And, wow, to see two Pac-10 (or any BCS conference) teams at the bottom of the chart is shocking.

jtwcornell91

some gigantic image nonsense

GAH! Must you hijack EVERY thread in which someone mentions a sporting result with your unwieldy graphical nerd-pictures? Especially since it throws off my browser scroll width, load time, and in this case, it covers up our championship graphic on the right side. Oddly enough, the google ad-box forces itself on top and some loser team is covered up by "1 rule to flat abs."

And what can we learn from this? If you stick to your guns and keep the nickname "Red Raiders" you become awesome at football. If you cave and change it to "Raiders" you never win anything, like Colgate.

(Apparently "Mean Green" does not hold the same magical powers.)

ugarte
The Big XII is that much stronger than the rest of the NCAA? They must have an amazing OOC record.

And, wow, to see two Pac-10 (or any BCS conference) teams at the bottom of the chart is shocking.

Not too shocking, Wazzou suuuuuuuuuuuuuuucks.

I kinda like them. They look like derigibles covered with graffiti.

It woudld be nice if replyers omitted the graphics so it only appears once.

ugarte
And, wow, to see two Pac-10 (or any BCS conference) teams at the bottom of the chart is shocking.

Not at all - you just have to remind yourself that the charts reflect math, not reality. (with no offense to JTW's methods)

I like the graphical nerd-pictures.

I never got around to doing this in December, but here's how the hockey-style KRACH with absolutist interpretations of perfections (aka the Jeffries prior in the Bayesian picture; search the archives for an explanation )

It's a bit of a mess because nine teams have "perfect" ratings relative to the pack, which also shows up in the Bradley-Terry table:

  # Team                BT    RRWP  W- L   W/L   SOS  RPI
  1 Oklahoma            infin .983 11- 1 11.00   inf .683
  2 Texas               infin .983 11- 1 11.00   inf .665
  3 Texas Tech          infin .983  9- 1 9.000   inf .647
  4 Utah                infin .979 11- 0 infin   inf .635
  5 Boise St            infin .970 11- 0 infin   inf .611
  6 Oklahoma St         infin .958  8- 3 2.667   inf .584
  7 TCU                 infin .958  9- 2 4.500   inf .597
  8 Brigham Young       infin .949  9- 2 4.500   inf .560
  9 Florida             7329. .892 11- 1 11.00 666.3 .666
 10 Alabama             5716. .883 11- 1 11.00 519.6 .621

Sorting this out is a good example of why the fictitious games (aka regularlizing prior) are a good idea. With those, the ratings of the top ten become:

  # Team                BT    RRWP  W- L   W/L   SOS  RPI
  1 Oklahoma            1954. .920 11- 1 6.000 325.6 .683
  2 Utah                1734. .912 11- 0 12.00 144.5 .635
  3 Texas               1516. .902 11- 1 6.000 252.6 .665
  4 Texas Tech          1352. .893  9- 1 5.000 270.3 .647
  5 Florida             1075. .872 11- 1 6.000 179.2 .666
  6 Boise St            1030. .868 11- 0 12.00 85.87 .611
  7 Alabama             857.5 .849 11- 1 6.000 142.9 .621
  8 Southern Cal        771.7 .837 11- 1 6.000 128.6 .606
  9 Penn State          720.1 .829 10- 1 5.500 130.9 .610
 10 TCU                 601.0 .807  9- 2 3.333 180.3 .597

(W/L now includes one fictitious win and loss against a team with a rating of 100.)

Afterwards, not only is Utah the only undefeated team, but Oklahoma and Texas Tech have both lost to teams in the pack, which sucks Texas and TCU down:

The top ten looks like

  # Team                BT    RRWP  W- L   W/L   SOS  RPI
  1 Utah                infin 1.000 12- 0 infin   inf .649
  2 Florida             23435 .964 12- 1 12.00 1953. .683
  3 Texas               16743 .956 12- 1 12.00 1395. .672
  4 Oklahoma            16373 .955 11- 2 5.500 2977. .672
  5 Alabama             12211 .946 11- 2 5.500 2220. .627
  6 TCU                 11029 .943 10- 2 5.000 2206. .614
  7 Southern Cal        6358. .921 12- 1 12.00 529.8 .627
  8 Texas Tech          6008. .918  9- 2 4.500 1335. .622
  9 Boise St            4983. .910 11- 1 11.00 453.0 .601
 10 Penn State          1976. .853 10- 2 5.000 395.3 .599

Of course, this doesn't tell you any more than that Utah, as the last undefeated team, should be #1 if you assume the game results tell you everything. But that's unsatisfying since it doesn't address the strength of schedule issue; A 1-0 Utah would look the same as a 12-0 Utah, USC's wins over Washington and Washington State mean literally nothing, etc. But if we add in the fictitious games (or equivalently find the maximum of the posterior using a regularizing prior we get what I think is the best objective ranking using only win-loss information from games against other FBS teams:

  # Team                BT    RRWP  W- L   W/L   SOS  RPI
  1 Utah                2592. .938 12- 0 13.00 199.4 .649
  2 Florida             1699. .911 12- 1 6.500 261.4 .683
  3 Texas               1530. .903 12- 1 6.500 235.4 .672
  4 Oklahoma            1348. .893 11- 2 4.000 336.9 .672
  5 Southern Cal        1121. .876 12- 1 6.500 172.5 .627
  6 Alabama             911.7 .856 11- 2 4.000 227.9 .627
  7 TCU                 784.8 .840 10- 2 3.667 214.0 .614
  8 Texas Tech          741.4 .833  9- 2 3.333 222.4 .622
  9 Boise St            670.6 .821 11- 1 6.000 111.8 .601
 10 Penn State          559.4 .798 10- 2 3.667 152.6 .599

That allows us to seed the NCAA I-A (or whatever you want to call it) tournament:

---

Round one:
#6 Alabama at #3 Texas
#5 Southern Cal at #4 Oklahoma
(Utah and Florida get byes)
Semifinals:
USC/OU winner at #1 Utah
Alabama/Texas winner at #2 Florida
National Championship game:
Semifinal winners at neutral site