Category Archives: Statistics

Using Advanced Stats to Project the Playoff Top 4

UPDATED 10/29

Using Bill Connelly’s conference win projections at Football Study Hall (Pac-12, Big 12, Big Ten, SEC, ACC), I have projected all of the remaining games for ranked unbeaten, one-loss teams (except for Marshall; if you want to go to the Playoff, all of your non-conference games can’t be against the MAC), and a select group of two-loss teams who are projected to win out. For the ACC teams, Notre Dame and ECU, I did my best to project their remaining games using F/+ ratings.

Scroll all the way to the bottom if you just want to see the top four.

Using the game projections, I found conference and overall win-loss records for the regular season as well as projected conference champions. Finally, using that information, I chose the most likely top-four teams who will be competing in the college football playoff.

Once the final records had been projected, to select the final four teams, I made a few assumptions beforehand. First of all, that undefeated Power 5 conference champions will be invited to the playoff. Secondly, the only possible two-loss teams to be invited would have to be from the SEC West or a conference champion. Third, that there is no way the committee will chose three teams from the same conference (even the SEC). Finally, that if two one-loss conference champions are available, the committee will look at the “quality” of those teams’ wins and loss.

#1 Mississippi St

Current Record: 6-0
Current Conference Record: 3-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Kentucky W 7-0 4-0
Arkansas W 8-0 5-0
@ Alabama L 9-1 5-1
Vanderbilt W 10-1 6-1
@ Ole Miss L 10-2 6-2

#2 Florida St

Current Record: 7-0
Current Conference Record: 4-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Louisville Toss-Up 8-0 or 7-1 5-0 or 4-1
Virginia W 9-0 or 8-1 6-0 or 5-1
@ Miami W 10-0 or 9-1 7-0 or 6-1
Boston College W 11-0 or 10-1 8-0 or 7-1
Florida W 12-0 or 11-1 8-0 or 7-1
ACC Championship W 13-0 or 12-1 9-0 or 8-1

#3 Ole Miss

Current Record: 7-0
Current Conference Record: 4-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ LSU W 8-0 5-0
Auburn W 9-0 6-0
@ Arkansas W 11-0 7-0
Mississippi St W 12-0 8-0
SEC Championship W 13-0 9-0*

*SEC Champs

#4 Alabama

Current Record: 6-1
Current Conference Record: 3-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Tennessee W 7-1 4-1
@ LSU W 8-1 5-1
Mississippi St W 9-1 6-1
Auburn W 11-1 7-1

#5 Auburn

Current Record: 5-1
Current Conference Record: 2-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
South Carolina W 6-1 3-1
@ Ole Miss L 6-2 3-2
Texas A&M W 7-2 4-2
@ Georgia L 7-3 4-3
@ Alabama L 8-4 4-4

#6 Oregon

Current Record: 6-1
Current Conference Record: 3-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Cal W 7-1 4-1
Stanford W 8-1 5-1
@ Utah W 9-1 6-1
Colorado W 10-1 7-1
@ Oregon St W 11-1 8-1
Pac-12 Championship W 12-1 9-1*

*Pac-12 Champs

#7 Notre Dame

Current Record: 6-1

Remaining FBS Schedule

Opponent W/L Overall Record
@ Navy W 7-1
@ Arizona St L 7-2
Northwestern W 8-2
Louisville W 9-2
@ USC L 9-3

#8 Michigan St

Current Record: 6-1
Current Conference Record: 3-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Michigan W 7-1 4-0
Ohio St L 7-2 4-1
@ Maryland W 8-2 5-1
Rutgers W 9-2 6-1
@ Penn St W 10-2 7-1

#9 Georgia

Current Record: 6-1
Current Conference Record: 4-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Florida W 7-1 5-1
@ Kentucky W 8-1 6-1
Auburn W 9-1 7-1
Georgia Tech W 11-1 7-1
SEC Championship L 11-2 7-2

#10 TCU

Current Record: 5-1
Current Conference Record: 2-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Texas Tech W 6-1 3-1
@ West Virginia W 7-1 4-1
K-State W 8-1 5-1
@ Kansas W 9-1 6-1
@ Texas W 10-1 7-1
Iowa St W 11-1 8-1*

*Big 12 Champs

#11 Kansas St

Current Record: 5-1
Current Conference Record: 3-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Texas W 6-1 4-0
Oklahoma St W 7-1 5-0
@ TCU L 7-2 5-1
@ West Virginia L 7-3 5-2
Kansas W 8-3 6-2
@ Baylor L 8-4 6-3

#12 Baylor

Current Record: 6-1
Current Conference Record: 3-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Kansas W 7-1 4-1
@ Oklahoma L 7-2 4-2
Oklahoma St W 8-2 5-2
Texas Tech W 9-2 6-2
K-State W 10-2 7-2

#13 Ohio St

Current Record: 5-1
Current Conference Record: 2-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Penn St W 6-1 3-0
Illinois W 7-1 4-0
@ Michigan St W 8-1 5-0
@ Minnesota W 9-1 6-0
Indiana W 10-1 7-0
Michigan W 11-1 8-0
Big Ten Championship W 12-1 9-0*

*Big Ten Champs

#14 Arizona St

Current Record: 5-1
Current Conference Record: 3-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Washington W 6-1 4-1
Utah W 7-1 5-1
Notre Dame W 8-1 5-1
@ Oregon St W 9-1 6-1
Washington St W 10-1 7-1
@ Arizona W 11-1 8-1
Pac-12 Championship L 11-2 8-2

#15 Arizona

Current Record: 5-1
Current Conference Record: 2-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Washington St W 6-1 3-1
@ UCLA L 6-2 3-2
Colorado W 7-2 4-2
Washington W 8-2 5-2
@ Utah L 8-3 5-3
Arizona St L 8-4 5-4

#16 Nebraska

Current Record: 6-1
Current Conference Record: 2-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Rutgers W 7-1 3-1
Purdue W 8-1 4-1
@ Wisconsin W 9-1 5-1
Minnesota W 10-1 6-1
@ Iowa W 11-1 7-1
Big Ten Championship L 11-2 7-2

#17 Oklahoma

Current Record: 5-2
Current Conference Record: 2-2

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Iowa St W 6-2 3-2
Baylor W 7-2 4-2
@ Texas Tech W 8-2 5-2
Kansas W 9-2 6-2
Oklahoma St W 10-2 7-2

#18 East Carolina

Current Record: 6-1
Current Conference Record: 3-0

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Temple W 7-1 4-0
@ Cincinnati W 8-1 5-0
Tulane W 9-1 6-0
@ Tulsa W 10-1 7-0
UCF W 11-1 8-0*

*American Champs

#19 Utah

Current Record: 5-1
Current Conference Record: 2-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
USC Toss-Up 6-1 or 5-2 3-1 or 2-2
@ Arizona St L 6-2 or 5-3 3-2 or 2-3
Oregon L 6-3 or 5-4 3-3 or 2-4
@ Stanford L 6-4 or 5-5 3-4 or 2-5
Arizona W 7-4 or 6-5 4-4 or 3-5
@ Colorado W 8-4 or 7-5 5-4 or 4-5

#21 Clemson

Current Record: 5-2
Current Conference Record: 4-1

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
Syracuse W 6-2 5-1
@ Wake Forest W 7-2 6-1
@ Georgia Tech W 8-2 7-1
Georgia St W 9-2 7-1
South Carolina W 10-2 7-1

#25 UCLA

Current Record: 5-2
Current Conference Record: 2-2

Remaining FBS Schedule

Opponent W/L Overall Record Conference Record
@ Colorado W 6-2 3-2
Arizona W 7-2 4-2
Washington W 8-2 5-2
USC W 9-2 6-2
Stanford W 10-2 7-2

Projected Records

TEAM PROJ CONF REC PROJ FINAL REC
Ole Miss 9-0 13-0*
Florida St 9-0 or 8-1 13-0 or 12-1*
Oregon 9-1 12-1*
Ohio State 9-0 12-1*
TCU 8-1 11-1*
East Carolina 8-0 11-1*
Alabama 7-1 11-1
Arizona State 8-2 11-2
Georgia 7-2 11-2
Nebraska 7-2 11-2
Michigan State 7-1 10-2
Clemson 7-1 10-2
Oklahoma 7-2 10-2
Baylor 7-2 10-2
UCLA 7-2 10-2
Mississippi St 6-2 10-2
Notre Dame 9-3
Kansas State 6-3 8-4
Arizona 5-4 8-4
Auburn 4-4 8-4
Utah 5-4 or 4-5 8-4 or 7-5

*Conference Champions

Now that we have projected the remaining games, I think we can safely put Ole Miss and Florida St into the playoff as undefeated SEC and ACC conference champions, respectively. Also, I think we can safely eliminate teams with two or more losses from consideration.

Since there are no other relevant unbeaten teams, using my rules for playoff eligibility, the following teams make up the possible pool for the two remaining spots.

Team Reason
Florida St ACC Champ
Oregon Pac-12 Champ
Ohio St Big Ten Champ
TCU Big 12 Champ
East Carolina American Champ
Alabama Only Loss to SEC Champ who is already in
Arizona St Pac-12 Runners-Up
Georgia SEC Runners-Up
Nebraska Big 12 Runners-Up

Other than those teams, I think you’d have a hard time making an argument for anyone else (even Mississippi St). With five one-loss teams (of which four are conference champions and the fifth being from the strongest division in the country), I can’t see how a two-loss team gets in.

That leaves us with Florida St, Oregon, Ohio St, TCU, East Carolina, and Alabama. It’s time to look at quality of wins and losses.

Florida’s only potential loss may come at the hands of Louisville on the road. Their wins against Oklahoma St and NC State were suspect and the close win against Clemson came without their Heisman-trophy winning QB. However, I think the human element of the selection committee gets them in based on past credentials. I don’t think the committee will be able to look past the fact that they’re defending champions and may have only lost once in two years.

Oregon’s loss to an un-ranked Arizona and barely squeaking by Washington St make them an unlikely pick out of these five.

Despite hanging half a hundred on four teams in a row, Ohio St‘s loss to Virginia Tech is looking worse and worse every week. And save Michigan St, their schedule is super weak.

TCU makes a good case for being chosen. They have strong wins over Oklahoma and Oklahoma St and their loss came at the hands of a quality Baylor team who they dominated for three of four quarters.

East Carolina. You almost lost to UConn. Sorry, you’re out.

While Alabama hasn’t been who most thought they would be, a huge win over Texas A&M and their only loss being a six-point deficit to arguably the best team in the nation. I think they’re in.

That’s it. There you have your advanced-stats-projected top-four college football playoff teams.

TOP 4

  1. Ole Miss
  2. Florida St
  3. Alabama
  4. TCU
Advertisements

Scoring Offenses and Defenses Ranked Better 2.0 – Pac-12 Edition

As I have thought more about how to better rank offenses and defenses other than just total yards or average points per game, I began to look at percentage of opponents average points scored and given up. I have two issues with my original method of calculating the expected points percentages: 1. the points per game number being used included defensive and special teams scores and 2. it did not take into account the number of plays used to score those points.

Regarding the second problem, if both the blue team and the red team average 45 points per game and their opponents give up an average of 30 points, they both will have an expected offensive scoring percentage of 150%. However, if the blue team runs an average of 85 plays per game and the red team delivers the same results while only running an average of 75 plays per game, we can now see that the red team has a more productive (i.e. better) offense.

Therefore, I have made two major changes to the calculations. First, only offensive points will be figured into the calculations (no defensive or special teams scores). That should have already been done; I just failed to realize it until now. Secondly, the figure will now be an expected percentage of points per play.

By figuring in the number of plays run in addition to the points scored, we’ll get a better idea of actual offensive production regardless of the style of offense. By nature, the fast-paced offenses being played today are going to run more plays and have more opportunities to score points (e.g. Arizona St. ranks 9th in the nation running average of 84 plays per game) while the more power-oriented, ground-and-pound offenses will not run as many plays, but can still be just as effective (e.g. Stanford ranks 115th out of 125 in number of plays run per game with 65).

Considering plays per game also adjusts for blowout wins. Many times, in very lopsided games, the winning team will score most of its points in the first half, then run a significantly fewer number of plays in the second half as they run the ball more often attempting to keep from running up the score.

Without further ado:

Percentage of Expected Points Scored Per Play (SCORING OFFENSE)

TEAM % OF EXP PTS/PLAY OLD METHOD CHANGE IN RK
1 Arizona St 150% 162%
2 Oregon 148% 149%
3 Stanford 123% 108% +6
4 UCLA 113% 115% -1
5 Utah 111% 110% +2
6 Oregon St 110% 112% -2
7 USC 108% 95% +4
8 Washington 104% 112% -3
9 Arizona 102% 110% -1
10 Washington St 92% 112% -4
11 Cal 81% 97% -1
12 Colorado 79% 76%

Percentage of Expected Points Given Up Per Play (SCORING DEFENSE)

TEAM % OF EXP PTS/PLAY OLD METHOD CHANGE IN RK
1 Stanford 60% 58%
2 USC 66% 61%
3 Oregon 71% 76%
4 Washington 74% 76%
5 Utah 78% 78%
6 UCLA 79% 81%
7 Arizona 94% 94%
8 Washington St 100% 103% +2
9 Arizona St 102% 99%
10 Oregon St 107% 99% -2
11 Colorado 121% 121%
12 Cal 122% 122%

I’m assuming that the reason there wasn’t as much change in the defensive numbers is that teams are facing different styles of offense and varying play counts which, by this point in the season, begin to balance each other out.

Combined Difference

TEAM % OF EXP PTS/PLAY OLD METHOD CHANGE IN RK
1 Oregon 77% 73%
2 Stanford 63% 50% +1
3 Arizona St 47% 63% -1
4 USC 42% 34% +2
5 UCLA 35% 35%
6 Utah 33% 32% +1
7 Washington 30% 37% -3
8 Arizona 8% 16%
9 Oregon St 3% 12%
10 Washington St -8% -5% -2
11 Cal -41% -25%
12 Colorado -42% -45%

Ranking Scoring Offenses and Defenses Better – ACC Edition

Below are the expected points scored and given up percentages for ACC teams through 11/7/13.

For the full explanation, please read the original post.

SCORING OFFENSE

TEAM % OF EXPECTED POINTS FOR
1 Florida St 182%
2 Clemson 142%
3 Miami 138%
4 Georgia Tech 115%
5 Duke 112%
6 Boston College 111%
7 UNC 105%
8 Pitt 96%
9 NC St 93%
10 Maryland 89%
11 Syracuse 86%
12 Virginia Tech 83%
13 Virginia 76%
14 Wake Forest 66%

SCORING DEFENSE

TEAM % OF EXPECTED POINTS AGAINST
1 Florida St 51%
2 Virginia Tech 59%
3 Clemson 80%
4 Wake Forest 83%
5 Georgia Tech 83%
6 Miami 87%
7 Pitt 91%
8 NC St 96%
9 UNC 98%
10 Boston College 102%
11 Syracuse 106%
12 Duke 107%
13 Virginia 109%
14 Maryland 110%

COMBINED DIFFERENCE

TEAM COMBINED DIFFERENCE
1 Florida St 131%
2 Clemson 61%
3 Miami 50%
4 Georgia Tech 31%
5 Virginia Tech 24%
6 Boston College 9%
7 UNC 7%
8 Duke 5%
9 Pitt 4%
10 NC St -3%
11 Wake Forest -17%
12 Syracuse -21%
13 Maryland -22%
14 Virginia -33%

CONFERENCE AVERAGES

Avg Off Exp Pts % Avg Def Exp Pts % Avg Diff
Conference 107% 90% 16%
Atlantic Division 110% 90% 20%
Coastal Division 104% 91% 13%

Ranking Scoring Offenses and Defenses Better – SEC Edition

Below are the expected points scored and given up percentages for SEC teams through 11/6/13. The offensive numbers represent the percentage of points scored on average compared to the average number of points a team’s opponents have given up. The defensive numbers represent the percentage of points given up compared to the average number of points scored by a team’s opponent. It’s an attempt to look not just at the number of points a team averages on offense and defense, but to also take into account the quality of the opponent those points were scored against or given up to.

The combined difference combines both percentages for an overall team scoring statistic.

For the full explanations, please read the original post.

SCORING OFFENSE

TEAM % OF EXPECTED POINTS FOR
1 Texas A&M 160%
2 Georgia 143%
3 LSU 141%
4 Alabama 136%
5 Missouri 132%
6 Ole Miss 123%
7 Auburn 116%
8 South Carolina 114%
9 Vanderbilt 106%
10 Tennessee 102%
11 Mississippi St 98%
12 Kentucky 98%
13 Florida 76%
14 Arkansas 72%

SCORING DEFENSE

TEAM % OF EXPECTED POINTS AGAINST
1 Alabama 37%
2 Florida 58%
3 Missouri 73%
4 Auburn 76%
5 South Carolina 79%
6 Ole Miss 79%
7 Mississippi St 85%
8 LSU 85%
9 Tennessee 95%
10 Texas A&M 98%
11 Georgia 99%
12 Arkansas 103%
13 Kentucky 104%
14 Vanderbilt 108%

COMBINED DIFFERENCE

TEAM COMBINED DIFFERENCE
1 Alabama 99%
2 Texas A&M 62%
3 Missouri 59%
4 LSU 56%
5 Georgia 44%
6 Ole Miss 44%
7 Auburn 40%
8 South Carolina 35%
9 Florida 19%
10 Mississippi St 14%
11 Tennessee 7%
12 Vanderbilt -2%
13 Kentucky -6%
14 Arkansas -31%

Updated Scoring Offenses and Defenses by Percentage of Expected Points (Pac-12)

I went back through the data and removed all games against DI-AA teams and I feel like that paints a more accurate picture. Here are the updated results. For the explanations of what this is, please read the original post.

SCORING OFFENSE

TEAM % OF EXPECTED POINTS FOR
1 Arizona St 170%
2 Oregon 156%
3 Utah 118%
4 Arizona 114%
5 Oregon St 114%
6 UCLA 112%
7 Washington 112%
8 Stanford 111%
9 Washington St 98%
10 USC 91%
11 Colorado 83%
12 Cal 83%

SCORING DEFENSE

TEAM % OF EXPECTED POINTS AGAINST
1 Stanford 60%
2 USC 62%
3 Oregon 73%
4 UCLA 74%
5 Washington 75%
6 Utah 86%
7 Arizona 88%
8 Oregon St 95%
9 Arizona St 102%
10 Washington St 107%
11 Colorado 108%
12 Cal 124%

COMBINED DIFFERENCE

TEAM COMBINED DIFFERENCE
1 Oregon 83%
2 Arizona St 69%
3 Stanford 51%
4 UCLA 38%
5 Washington 37%
6 Utah 32%
7 USC 28%
8 Arizona 26%
9 Oregon St 19%
10 Washington St -9%
11 Colorado -25%
12 Cal -41%

Ranking Scoring Offenses and Defenses Better (Pac-12 Edition)

These rankings were inspired by Paul Dalen‘s take on ranking total offenses and defenses. Instead of looking at total yards generated per game offensively or total yards given up defensively (statistics which do not account for the strength or weakness of opponents), he compared a team’s yards gained or given up to the team’s opponents’ average number of yards given up or gained. This number resulted in a percentage of expected yards gained or given up.

For example, if the blue team averages 400 yards per game and the red team holds them to 350 yards, the red team gave up 87.5% of expected yardage. Apply this concept to the season as a whole and you have a better way to rank total offense and defense by now accounting for the quality of the opponent.

I decided to take this same principle and apply it to points for and points against. Unfortunately, I didn’t have the time to compare all 125 teams, so I chose to just look at the Pac-12.

Here are the average points for and against for the Pac-12 as of week 9 games.

TEAM AVG PTS FOR TEAM AVG PTS AGAINST
1 Oregon 55.6 1 Oregon 16.9
2 Arizona St 45.4 2 USC 19.3
3 Oregon St 40.1 3 Stanford 19.4
4 Arizona 36.3 4 Arizona 19.9
5 UCLA 36.1 5 UCLA 22.4
6 Washington 34.5 6 Washington 23.6
7 Stanford 32.6 7 Utah 25.8
8 Utah 31.1 8 Arizona St 26.7
9 Washington St 30.9 9 Oregon St 27.4
10 Colorado 26.9 10 Washington St 29.9
11 USC 24.6 11 Colorado 37.1
12 Cal 22.9 12 Cal 44

Based on the above data, if you were an Oregon State fan, let’s say, you might say that Oregon State has the third-best scoring offense in the conference. I would argue that you are wrong. The numbers above do not account for quality of opponent at all. Colorado, for example, has played two I-AA teams while Stanford hasn’t played any.

Following Paul Dalen’s example, we’ll take the points scored and given up and compare them to teams’ opponents’ average points given up and scored for an expected points percentage. For example, if the green team scores 35 points against the yellow team and the yellow team is giving up on average 28 points a game, the green team’s expected points percentage would be 125% (they scored 125% of the number of points they were expected to score based on their opponent’s average). Apply that concept to the whole season and these percentages will give a better representation of scoring offense and defense.

EXPECTED POINTS FOR % (SCORING OFFENSE)

TEAM % OF EXPECTED POINTS FOR
1 Arizona St 192%
2 Oregon 172%
3 Washington 130%
4 Arizona 128%
5 Oregon St 127%
6 Utah 127%
7 Stanford 121%
8 UCLA 119%
9 Washington St 116%
10 Colorado 113%
11 Cal 97%
12 USC 95%

The above table shows that, on average, Arizona State is scoring 192% of expected points (i.e. 192% of their opponents’ average points given up per game) while USC is scoring only 95% of expected points (making opposing defenses look better than they have all season). I would now say to the Oregon State fan that his team has the fifth-best scoring offense in the conference.

If you just look at points per game, Arizona is fourth in the conference while Washington is sixth. What that doesn’t factor in is the fact that Washington has played the top two teams in the conference, while Arizona has not. By looking at percentage of expected points, we get a better comparison with Washington now at third and Arizona behind them.

EXPECTED POINTS AGAINST % (SCORING DEFENSE)

TEAM % OF EXPECTED POINTS AGAINST
1 Stanford 57%
2 Oregon 59%
3 USC 61%
4 Washington 67%
5 UCLA 69%
6 Arizona 71%
7 Utah 82%
8 Arizona St 87%
9 Oregon St 94%
10 Washington St 96%
11 Colorado 96%
12 Cal 116%

On average, Stanford on defense is holding teams to 57% of their average points scored per game while Cal is giving up 116% of teams’ average PPG.

With these two expected points percentages (offense and defense), we can now add the differences from 100% (e.g. ASU offense: 192% – 100% = 92%, ASU defense: 100% – 87% = 13%, ASU overall: 92% + 13% = 105%) for an interesting way to get an overall ranking of the teams in the conference.

COMBINED DIFFERENCES (EXPECTED POINTS FOR + EXPECTED POINTS AGAINST)

TEAM COMBINED DIFFERENCE
1 Oregon 113%
2 Arizona St 105%
3 Stanford 64%
4 Washington 64%
5 Arizona 57%
6 UCLA 50%
7 Utah 45%
8 USC 35%
9 Oregon St 33%
10 Washington St 20%
11 Colorado 17%
12 Cal -19%