NCAA 1-A Week 11 Beatpaths

All right, we have a quick look now at the college rankings for just the Division 1-A teams. This time I’m using the BCS as a tiebreaker for when teams’ BeatPower rankings are identical. Here’s the graph, just to get a general visual idea (sorry, it’s big enough that I had to shrink it down to the point where you can’t really read the labels unless you’re high).

2005-11-Clean

Here’s the Beatpaths NCAA Top 25, from games through 11/5/05. You should totally take these with a grain of salt, because I’m not even sure I parsed the game data correctly, and I haven’t seriously looked at the algorithm to see if I missed anything obvious.

Rank Team Notes BCS BeatPower

1

USC (Beat Stanford) Four teams with maximum beatpower.

1

100.0

(75/75 – 0/75)

2

Texas (Beat Baylor) These four teams are basically in the order of the BCS as a tiebreaker

2

100.0

(73/73 – 0/73)

3

Alabama (Beat Mississippi St) Although at least for the 100.0 teams, you could order them by how many beatwins they have…

3

100.0

(54/54 – 0/54)

4

Florida St (Lost to NC State) And then Florida State would be #1. Their loss to NC State didn’t matter because they have an alternate beatpath route to them, courtesy of their win over Miami.

19

100.0

(84/84 – 0/84)

5

LSU (Bye) (no comment)

7

99.2

(61/62 – 0/62)

6

Texas-El Paso (Beat Tulsa) I don’t know much about UTEP. The graph likes them. I’m not sure what it’s missing. BCS doesn’t rank them.

98.9

(43/44 – 0/44)

7

Miami FL (Beat Virginia Tech) Definitely lower here than the mainstream polls.

4

98.8

(83/84 – 1/84)

8

Penn St (Beat Wisconsin) Caught in a couple of beatloops with teams they haven’t managed to overcome.

5

98.6

(69/71 – 0/71)

9

Oregon (Beat California) Go Ducks!

10

98.4

(60/61 – 1/61)

10

TCU (Beat Colorado St) (no comment)

17

98.1

(50/52 – 0/52)

11

Wisconsin (Lost to Penn St) (no comment)

16

98.0

(71/74 – 0/74)

12

Texas Tech (Beat Texas A&M) (no comment)

12

97.9

(46/47 – 1/47)

13

Virginia Tech (Lost to Miami FL) Ranked lower than BCS partly because Miami is also ranked lower than BCS

6

97.5

(79/81 – 2/81)

14

UCLA (Lost to Arizona) (no comment)

15

97.4

(54/57 – 0/57)

15

Florida (Beat Vanderbilt) (no comment)

13

97.2

(52/54 – 1/54)

16

Ohio State (Bye) Perhaps it’s possible the bye has hurt some teams, since they’re behind on gathering beatwins…?

8

97.1

(66/68 – 2/68)

17

Georgia (Bye) (no comment)

9

96.2

(50/52 – 2/52)

18

West Virginia (Bye) (no comment)

14

95.9

(70/73 – 3/73)

19

Stanford (Lost to USC) Unranked by BCS…

95.6

(54/57 – 2/57)

20

Notre Dame (Beat Tennessee) (no comment)

11

95.5

(51/55 – 1/55)

21

Fresno St (Beat San Jose St) (no comment)

22

95.3

(41/43 – 2/43)

22

Colorado (Beat Missouri) My Alma Mater, the Buffs. I founded a singing group there named “In The Buff”.

18

94.6

(53/56 – 3/56)

23

Iowa State (Beat Kansas St) unranked by BCS…

93.8

(42/48 – 0/48)

24

Louisville (Bye) (no comment)

25

93.7

(66/71 – 4/71)

25

Oklahoma (Bye) unranked by BCS…

93.6

(43/47 – 2/47)

Other BCS teams: We have Michigan (21) at 26, Minnesota (24) at 27, Georgia Tech (23) at 30, and Auburn (20) all the way down at 36.

Who’s last at 118th? Rice.

Technorati Tags: , ,

8 Responses to NCAA 1-A Week 11 Beatpaths

  1. ThunderThumbs says:

    Looks like Florida State will definitely sink now. Its previous two losses weren’t enough to make up for its victory over Miami, but I can’t imagine that their Clemson loss won’t be.

  2. Pat says:

    The deal with UTEP is easy: they have the easiest schedule in college football, by far. Well, close to: take a look at their strength-of-schedule calculation on http://www.colleyrankings.com – they’re 117/119.

    When you’ve got almost the easiest schedule in all of college football, you can’t afford to lose to Memphis. In terms of beatpaths, that loss only hurts a little, because they quickly beatloop it away since the rest of their conference is so weak.

    Even from the graph, you can tell that UTEP’s ranking is weak. They can gain a maximum of 11 beatpaths from their own conference, and their entire conference has won a total of 8 games out-of-conference (!). Of those 8 games, 2 teams have zero wins. TCU managed to beatloop its SMU loss, which means all of UTEP’s beatpaths are probably coming from its one win against New Mexico – which, since it was out of conference, can’t easily be beatlooped away.

    I think if you probably computed the average BeatPower ranking for the teams UTEP’s got a beatpath to, you’d find it’s really, really low.

  3. ThunderThumbs says:

    The average BeatPower of direct beatwins could be an interesting tiebreaker. I think part of that is already figured into the system – because if you beat a poor team, then their beatpath length isn’t apt to be very long (or including a large number of teams) in the first place, which would keep an (n-1)/n calculation from being as high as another team with a larger n.

    But it does kind of tie into another thing I’ve been pondering. Sometimes someone’s placement near the top is pretty rickety, as we saw with Carolina this week in the NFL. Graphically it means that their beatpath is linear for many nodes, with no reinforcement, and doesn’t branch out until further down the graph. I’m wanting to reward stability of a beatpath a bit more.

  4. Pat says:

    Yah. In a well-connected system, that rickety ranking isn’t likely to stick around for a long time, because eventually you’re going to have to play someone connected to that, and you’ll either win and reinforce the ranking, or lose and go plummeting down. But, college football isn’t well connected, and so one out-of-conference win can give you an entirely huge set of beatpaths, which will *never* be challenged.

    As for beating a bunch of poor teams, while it does factor into the system in a league where losses happen more than once a season, I think here it doesn’t work as well. Look at UTEP: clearly beating weak teams doesn’t help them – 43 beatwins over 44 teams is the lowest number of beatwins of anyone around them. By far. But the fact that they don’t have any beatlosses (as their loss happened in conference, so it’s easy to beatloop) pushes them way up.

    Iowa State is another example, with three conference losses, which they rapidly beatloop in the weak Big 12. If those losses had been out of conference instead, they’d be much lower.

  5. ThunderThumbs says:

    I’m not as concerned about the difficulty of forming beatloops. I think it’s moderated by the possibility that out-of-conference losses can just be contained in much longer beatloops if it truly is ambiguous. The extra conference activity just gives more data about whether those portions of the beatloop chains should be kept or thrown out. I think.

    What seems more pressing, though, is finding a graceful way to calculate in the strength of beatwins, beatlosses, and beatloops (for a given team), hopefully without introducing any “alpha” business.

  6. Pat says:

    I don’t think it’s that big of a deal either, though the chance of an out-of-conference loss being contained in a beatloop is much, much lower than an in-conference loss. That’s the problem. Take a look at Northwestern’s loss to Arizona State. There was simply no way for a beatloop to be formed simply because the Big 10 and the Pac-10’s schedules are so disjoint (Big 10 primarily played the MAC, and the Pac-10 didn’t play the MAC at all).

    But I agree that adding in strength of beatwins is way more important, at least for college football. I don’t think it’s a problem in the NFL because you can’t really develop disconnected “islands” of weak teams, whereas UTEP playing in Conference-USA and playing a WAC and MWC team is pretty much the definition of an island of weak teams.

    In terms of the graph, UTEP is probably the lowest team on the graph without a beatloss. “King of the midgets”.

  7. Becephalus says:

    I enjoy your graphs nice work. Defintely a way of looking at teams many people have mentally toyed with but never had the skill/desire to follow through with. My only complaint is that the Colts are not ranked number 7 (wink wink).

  8. ThunderThumbs says:

    Funny. 🙂 Actually, while experimenting with strength of beatwins/beatlosses, Indianapolis very much does sink in the rankings. I just haven’t been able to narrow down to a calculation that is simple enough that it seems obvious and elegant.

Leave a Reply

Your email address will not be published. Required fields are marked *