ArtsAutosBooksBusinessEducationEntertainmentFamilyFashionFoodGamesGenderHealthHolidaysHomeHubPagesPersonal FinancePetsPoliticsReligionSportsTechnologyTravel

Ridley's Rankings, Week 9: Shepherding in Win Probability

Updated on October 28, 2014

Looking over the current college football landscape, it was pretty clear there weren't many teams, if any, that had a viable shot to make the playoffs that I hadn't covered. One could argue about the Arizona's, K-State or Nebraska, but that argument would have likely fallen on deaf ears.

The truth is, there are really only about 10 to 12 (maybe 15) teams that are currently in play to make the playoffs. Because the top of the rankings are loaded with SEC West teams that will continue to cannibalize each other, we know that there is a team or two not in the top four that will end up making the playoffs. The question is who?

The Bulldogs are the early favorite to be the top seed in the playoffs.
The Bulldogs are the early favorite to be the top seed in the playoffs.

To the Calculators!

To figure out who was the most likely to make it to the playoff (or rather, have a record to put them in contention), I developed an algorithm to predict win probability. While this algorithm is far from perfect, it would have correctly predicted over 69 percent of the games played this year; a decent number considering that no algorithm would be able to predict an unranked Mississippi State team going into Death Valley against an eighth-ranked LSU team and coming out victorious. Or at least not one that would produce reliable results.

For those of you wondering how I came across this algorithm, I performed multiple regression on a series of variables relating to wins for every FBS game played since 2004. This produced statistically-relevant data points (home versus away, conference, ranking,etc) and how significant they were to producing wins. These points were put together in a formula which was applied to each team in their respective game. The results of the two teams were then compared to each other, with the higher total being the predicted winner.

I'll use this information to cover the top 10 teams in each week's rankings and breakdown their winning probability of each game and their projected final record. With these projected final records, I will deliver my best guess of who will make the playoffs.

(Quick note: My win probability does not work for FCS teams. The extra data required to incorporate them in there was not worth the time. Especially since only three ranked teams have fallen to FCS teams since the beginning of 2004, none after Week 2. So for the sake of this exercise, we will assume a 99% win probability.)

The Rankings

Rank
Team
Total
My Rank
1
Mississippi State
100.00
1
2
Auburn
89.48
3
3
Florida State
88.13
4
4
Ole Miss
87.91
2
5
Alabama
87.05
5
6
Oregon
84.53
7
7
TCU
80.68
8
8
Michigan State
80.22
10
9
Georgia
78.75
11
10
Notre Dame
78.74
15
11
LSU
75.71
9
12
Kansas State
75.41
13
13
Ohio State
73.76
14
14
Baylor
72.86
18
15
Arizona
72.63
16
16
Nebraska
71.78
12
17
Marshall
71.72
6
18
Arizona State
70.76
17
19
Utah
67.82
21
20
West Virginia
66.97
19
21
Clemson
65.19
20
22
Oklahoma
63.74
24
23
Duke
61.47
23
24
UCLA
60.75
22
25
East Carolina
59.59
31
Opponent
Rank
Win Probability
Arkansas
N/A
75.47%
Tennessee-Martin
N/A
99.00%
@Alabama
3
62.64%
Vanderbilt
N/A
76.91%
@Mississippi
7
63.97%
Projected Final Record
 
12-0

1. Mississippi State (7-0) (100.00) [1]

Despite games against Alabama and Mississippi on the road, the Bulldogs will go into their remaining four games as favorites. Over the time span of my study, teams in away games that were ranked higher than their opponent (when the opponent is also ranked) won nearly 66% of their games, with the top-ranked team being 15-5 against ranked teams on the road. If the Dawgs can avoid being the sixth team to lose in those situations, they will find themselves as the top seed going into the playoffs.

Opponent
Rank
Win Probability
@Ole Miss
7
45.98%
Texas A&M
N/A
77.36%
@Georgia
9
60.20%
Samford
N/A
99.00%
@Alabama
3
26.91%
Projected Final Record
 
9-3

2. Auburn (6-1) (89.48) [3]

Like the top-ranked Bulldogs, the Tigers face a brutal schedule down the stretch. Unlike Hail State, they won't be favored to get through it unscathed. The numbers predict Auburn will fall to both Ole Miss and Alabama on the road. The game against the Rebels is too close to state with any certainty but it appears they will have little hope in the Iron Bowl. Of course, stranger things could happen. When Mississippi State won in Death Valley, they had just an 11% chance of winning. Of course, a large portion of this was attributed to the Dawgs being vastly underrated.

Opponent
Rank
Win Probability
@Louisville
N/A
66.79%
Virginia
N/A
75.37%
@Miami(FL)
N/A
66.30%
Boston College
N/A
76.42%
Florida
N/A
70.34%
Projected Final Record
 
12-0

3. Florida State (7-0) (88.13) [4]

God bless the ACC! That has to be the thought traveling across the Florida State campus this season. By being in the ACC, the Seminoles get to boast about being in a coveted "Power Five" conference while also lacking any top-tier opponents. In fact, were it not for their out-of-conference game against Notre Dame, Jimbo Fisher's team wouldn't have played a single team in the top 20 at the time of the game. At 7-0, they have four conference games (none against ranked opponents) and a bout with Florida remaining, meaning the folks in Tallahassee can likely start planning their New Years' trip now.

Opponent
Rank
Win Probability
Auburn
4
54.02%
Presbyterian
N/A
99.00%
@Arkansas
N/A
63.70%
Mississippi State
1
36.03%
Projected Final Record
 
10-2

4. Ole Miss (7-1) (87.91) [2]

The remainder of the season will not be easy for the Rebels. They will be favored in three of their four remaining games, but two of those have a win probability lower than what they would have had going into the LSU match-up. Ole Miss does have Presbyterian along with a bye coming up to help right the ship before facing Arkansas and Mississippi State, but they must get through Auburn first. If they can't do that, it won't matter what they do in the rest of their games.

Opponent
Rank
Win Probability
@LSU
16
65.75%
Mississippi State
1
37.36%
Western Carolina
N/A
99.00%
Auburn
4
73.09%
Projected Final Record
 
10-2

5. Alabama (7-1) (87.05) [5]

The run of the SEC West continues with Alabama being the fourth team in the top five. The Tide roll into the five spot (pardon the pun) with a 7-1 record but are a mere 1-1 against ranked teams. But that will all change as their schedule gets considerably harder down the stretch. They first take a trip to Death Valley ("where opponents dreams come to die") following their bye, then play host to Mississippi State, Western Carolina (sure to be a barn-burner) and Auburn in the Iron Bowl. Alabama will be favored in all but the Hail State game, but that loss alone will likely eliminate Nick Saban's team from contention.

Opponent
Rank
Win Probability
Stanford
N/A
77.53%
@Utah
18
67.98%
Colorado
N/A
76.56%
@Oregon State
N/A
69.10%
Projected Final Record
 
11-1

6. Oregon (7-1) (84.53) [7]

Oregon's path to the playoffs is among the cleanest of the one-loss teams. They must defeat rival Stanford -- something they haven't done in the last two years -- as well as take care of business against the Utes in Utah, but their path to the final four is completely attainable. With the Ducks being a 67% favorite in each of their games, their biggest problem, as it has been for the past couple seasons, is not stepping on their own feet (like they did against Arizona). If this talented, but sometimes unfocused group can do that, they will find themselves two wins away from a national title.

Opponent
Rank
Win Probability
@West Virginia
20
66.83%
Kansas State
11
88.02%
@Kansas
N/A
67.81%
@Texas
N/A
67.68%
Iowa State
N/A
77.35%
Projected Final Record
 
11-1

7. TCU (6-1) (80.68) [8]

In evaluating TCU, we come across the first flaw with my algorithm. Without getting too deep into the specifics, the algorithm has a variety of negative and positive variables that combine to spit out a number. When two teams have similar answers to these variables, such as win percentage, ranking, etc, the variables are essentially negated and only the remaining variables are accounted for. This can create an uncharacteristically lopsided win probability as we see with TCU and Kansas State. These teams are ranked right next to each other, have the same record and are in the same conference, so variables like home or away, offensive performance and defensive performance are amplified. Because TCU is the home team, ranked slightly higher and have been much better offensively, the winning probability was skewed. TCU would still be projected winner, but certainly would not be considered 88 percent favorites.

Before TCU even gets to K-State, they must first overcome the Mountaineers, who have been playing lights out lately. If they can survive those two match-ups against ranked opponents, they should find themselves in the clear with lowly Kansas, Texas and Iowa State remaining.

Opponent
Ranking
Win Probability
Ohio State
13
75.08%
@Maryland
N/A
68.66%
Rutgers
N/A
79.35%
@Penn State
N/A
68.16%
Projected Final Record
 
11-1

8. Michigan State (7-1) (80.22) [10]

Much like Oregon, who delivered the Spartans their lone loss on the season, Michigan State will be decently favored in all of their remaining games. The Ohio State probability line will probably look a little off. With Sparty being both higher-ranked and the home team, they have history on their side. Home teams have won nearly 73% of games as the higher ranked team (when both are ranked) and that number climbs to almost 80% when you only consider teams ranked eighth (as MSU is) or higher.

The rest of the games on Michigan State's schedule should present little challenge. Maryland, Penn State and Rutgers all have winning records at the moment, but they are a combined 0-4 against ranked teams this season. The bigger question is whether it will be enough to enter the dance.

Opponent
Rank
Win Probability
Florida
N/A
71.83%
@Kentucky
N/A
67.61%
Auburn
4
39.80%
Charleston Southern
N/A
99.00%
Georgia Tech
N/A
84.33%
Projected Final Record
 
10-2

9. Georgia (6-1) (78.75) [11]

Georgia is the sole member of the top 25 in what is a pitiful SEC East. For every overachieving team in the SEC West, there's an equally disappointing team in the East. Luckily for the Bulldogs, they have seemed to resist the plague that is inflicting the rest of their division. They're now 6-1 and have only Auburn remaining as a true challenge. Should they win out, they will secure a spot in the SEC title game with their eyes on the playoffs. If they stumble, say to Auburn, they will have to see Missouri trip up as well. This is paramount as a two-loss Georgia team that has "SEC Champions" behind their name will get into the playoffs (sorry, that's the way it is); a two-loss Georgia team that doesn't even make it to it's own conference championship will wrap up their season in 2014.

Opponent
Rank
Win Probability
@Navy
N/A
65.57%
@Arizona State
15
59.40%
Northwestern
N/A
71.79%
Louisville
N/A
75.62%
@USC
N/A
63.80%
Projected Final Record
 
11-1

10. Notre Dame (6-1) (78.74) [15]

The biggest thing going for Notre Dame right now is their reputation. The Irish haven't been particularly thrilling or mind-blowing, but they sit sixth in the AP thanks to the national respect their name (or brand) carries. Their biggest win so far is a three-point victory over Stanford, who is no longer ranked. Their best game, as far as proving national relevance, was the penalty-marred loss to the Seminoles two weeks ago. Other than that, their season and resume is pretty bland.

Because the name Notre Dame carries such weight in the college football arena, they will automatically be considered for the playoffs if their record is comparable. To do that, they will need to get by Arizona State and a tough USC team, both on the road. Should that happen, there's no telling how the committee will vote.

Who Makes the Playoffs?

That question will be answered in a follow-up post to hopefully be published tomorrow. Until then, feel free to speculate in the comments.

Comments

    0 of 8192 characters used
    Post Comment

    No comments yet.

    Click to Rate This Article