Writers of Pro Football Prospectus 2008

Most Recent FO Features

AllenKee13.jpg

» Week 11 Quick Reads

Most of the headlines covering this weekend's Bills-Chargers game focused on Nathan Peterman and his very bad day. Few realized that at the same time, Keenan Allen was having a career day for Los Angeles.

16 Oct 2015

VN: A Different Team Every Week

by Bill Connelly

It is said that simple ratings systems tend to be the best, and I tend to agree. The more variables you enter into the mix, the more you're opening yourself up to volatility, unnecessary complexity, etc. This makes perfect sense, and it is the reason why I feel that, for instance, Sports Reference's Simple Rating System (SRS, with "simple" right there in the title!) is fun and useful. I use it a lot for comparing eras, for instance.

Naturally, then, my own system of ratings, S&P+, is just saturated in variables. It is the opposite of simple in every possible way. When I redrew it last winter, I actually added some more variables. I can't help myself.

Now, I did this in part because the tweaks helped S&P+ from a predictive standpoint. It's at least a little better now. But it's unstable and unwieldy, and it can take a while to begin to understand certain teams. It produces pretty strange results, especially in the first half of the season. And I love it ever so much.

When I started tinkering with college football numbers in the first place, it because I wanted toys to play with. Sure, I wanted to create the Greatest College Football Ratings Ever, but that was always secondary to the simple fact that I wanted to figure out ways for stats to help supplement my analysis. When I began scouring Google for "college football advanced stats" in 2007, if I had found that others had already detailed and measured college football's play-by-play universe, I may never have attempted it myself. But they hadn't, so I did.

From a pure predictive standpoint, S&P+ is fine. It's not amazing, but it's solid, and when added to Brian Fremeau's FEI, it gets even better. I would even call S&P+ and F/+ above average to good in that regard. But you can probably find better, simpler measurements elsewhere for those purposes.

What S&P+ offers that nobody else does, however, is detail, the same stuff you get on the pro side of Football Outsiders. Rushing and passing and standard downs and passing downs and first quarter and second down and garbage time S&P+ data. Adjusted line yards and sack rates and opportunity rates and situational breakouts of those, too. Personality stats. Havoc rates. Individual rushing and receiving stats. My desire to share this data with readers is what prompted the creation of the Football Study Hall stat profile pages. These are, I hope, excellent tools for analysis. And hopefully S&P+ is just good enough at the predictive side that you can take the analysis tools seriously.

One of my favorite new tools in the S&P+ universe is the per-game percentile data. Each team's performance is distilled into a single percentile figure, and over the course of a season we can use this to learn not only who's good and bad, but who's consistent and who most embodies Lou Holtz's famous "you coach a different team every week" truism.

Let's take a look at precisely that through six weeks of the 2015 season. Below is a table with a team's average percentile performance, along with its median, standard deviation, average week-to-week change (absolute value), maximum, and minimum.

S&P+ Rk Team AVERAGE MEDIAN STD. DEV AVG. CHANGE MAX MIN
1 Clemson 93.1% 96.1% 7.7% 9.3% 98.7% 79.6%
2 Alabama 90.7% 97.0% 13.7% 15.4% 98.4% 63.6%
3 Michigan 89.7% 93.9% 13.2% 8.5% 98.1% 63.1%
4 Florida State 89.8% 90.3% 4.7% 7.3% 94.4% 82.3%
5 Florida 84.1% 83.3% 10.3% 16.2% 99.1% 71.8%
6 USC 81.3% 93.5% 20.4% 23.6% 97.6% 50.6%
7 Notre Dame 83.4% 85.4% 12.1% 18.0% 95.8% 63.7%
8 West Virginia 76.0% 87.7% 30.7% 35.0% 99.4% 23.9%
9 Oklahoma 78.2% 84.2% 23.1% 26.2% 97.2% 40.9%
10 Stanford 79.6% 85.7% 18.2% 19.1% 96.5% 50.2%
11 LSU 80.5% 77.7% 8.4% 7.5% 93.4% 72.4%
12 Northwestern 75.9% 82.9% 25.2% 26.8% 96.8% 29.5%
13 Utah 77.1% 77.8% 17.7% 25.9% 97.6% 49.4%
14 Duke 77.8% 81.9% 15.5% 21.4% 97.1% 57.9%
15 Texas A&M 78.8% 85.7% 13.3% 23.0% 92.7% 62.1%
16 Ole Miss 75.4% 89.3% 32.3% 33.6% 94.8% 11.3%
17 TCU 76.9% 74.6% 14.1% 23.1% 96.3% 58.4%
18 Louisville 73.0% 85.2% 23.7% 34.7% 98.7% 46.0%
19 Western Kentucky 75.2% 80.6% 20.9% 17.0% 94.1% 49.8%
20 Arkansas 76.2% 77.4% 17.0% 22.4% 95.9% 54.1%
21 Toledo 75.2% 85.9% 20.4% 12.2% 95.0% 52.7%
22 Baylor 77.9% 77.2% 8.9% 9.7% 92.3% 70.0%
23 Washington 75.3% 87.4% 20.4% 28.7% 90.3% 44.1%
24 UCLA 76.3% 81.5% 14.6% 18.8% 92.6% 54.6%
25 Iowa 76.2% 77.9% 13.2% 15.9% 91.5% 52.4%
S&P+ Rk Team AVERAGE MEDIAN STD. DEV AVG. CHANGE MAX MIN
26 Kansas State 74.8% 82.7% 16.4% 11.7% 93.1% 54.0%
27 Texas Tech 74.8% 78.9% 16.9% 22.2% 92.4% 50.4%
28 Ohio State 76.6% 74.4% 7.4% 6.1% 86.4% 67.9%
29 Illinois 67.2% 64.5% 25.8% 24.5% 99.3% 34.7%
30 Tennessee 72.8% 76.2% 21.7% 31.1% 91.8% 34.2%
31 Penn State 71.7% 76.1% 20.1% 23.9% 96.0% 43.6%
32 BYU 72.1% 82.5% 26.5% 31.0% 91.8% 22.0%
33 Navy 73.5% 81.6% 19.9% 13.1% 85.4% 38.0%
34 Oklahoma State 73.0% 72.1% 10.7% 15.2% 91.8% 62.8%
35 Minnesota 68.6% 71.6% 26.9% 37.6% 95.1% 23.3%
36 Wisconsin 69.6% 77.3% 25.0% 21.0% 92.5% 28.1%
37 Arizona State 68.2% 73.2% 25.6% 34.8% 93.9% 32.2%
38 Pittsburgh 70.2% 78.6% 16.9% 25.1% 87.7% 51.0%
39 Temple 69.7% 75.2% 19.0% 22.5% 89.3% 39.0%
40 Michigan State 70.4% 66.9% 11.2% 13.0% 84.3% 58.7%
41 Georgia 67.3% 79.1% 29.8% 23.5% 93.1% 19.2%
42 Louisiana Tech 68.3% 68.0% 18.9% 24.0% 91.9% 43.4%
43 Georgia Tech 62.9% 57.7% 28.3% 23.7% 97.2% 32.9%
44 Boise State 66.1% 68.6% 25.7% 32.2% 94.8% 28.5%
45 Virginia Tech 63.5% 65.3% 29.8% 37.5% 95.6% 30.9%
46 Nebraska 66.2% 66.9% 19.6% 32.5% 92.6% 43.1%
47 Miami-FL 65.3% 74.9% 23.4% 12.4% 87.7% 38.2%
48 Indiana 65.3% 66.6% 14.0% 11.0% 78.9% 50.2%
49 California 63.4% 64.6% 22.2% 26.3% 87.4% 33.6%
50 Boston College 61.8% 65.6% 22.9% 33.3% 94.9% 27.2%
S&P+ Rk Team AVERAGE MEDIAN STD. DEV AVG. CHANGE MAX MIN
51 NC State 60.4% 61.0% 29.1% 31.4% 94.4% 19.8%
52 Ohio 63.3% 63.9% 13.3% 17.2% 77.3% 41.2%
53 Mississippi State 60.3% 66.9% 26.6% 41.7% 87.3% 21.0%
54 Vanderbilt 60.8% 70.9% 17.9% 29.7% 74.3% 33.2%
55 Bowling Green 60.1% 68.0% 21.5% 20.4% 80.7% 21.0%
56 Maryland 56.1% 49.0% 30.0% 38.2% 93.6% 22.2%
57 Auburn 58.5% 58.5% 27.5% 32.1% 88.1% 20.2%
58 Iowa State 58.1% 68.8% 26.3% 34.6% 83.1% 20.5%
59 Memphis 57.9% 64.9% 20.1% 19.9% 83.9% 34.6%
60 Houston 57.1% 61.2% 20.6% 23.1% 85.4% 31.0%
61 Texas 56.8% 68.3% 34.1% 44.0% 89.2% 12.7%
62 East Carolina 56.5% 57.5% 18.8% 24.4% 82.6% 33.0%
63 Northern Illinois 56.8% 55.4% 16.7% 15.7% 76.4% 36.0%
64 Oregon 57.3% 64.5% 27.4% 45.2% 82.9% 8.9%
65 Utah State 54.2% 48.0% 30.4% 27.6% 89.6% 23.7%
66 Tulsa 55.1% 54.6% 17.6% 21.5% 83.1% 38.7%
67 Air Force 55.0% 53.5% 21.7% 21.7% 85.1% 25.8%
68 Virginia 55.4% 61.4% 15.9% 19.5% 71.1% 38.4%
69 Marshall 53.6% 56.0% 28.9% 35.2% 81.7% 20.3%
70 Arizona 53.5% 56.1% 30.6% 38.0% 85.0% 10.7%
71 Massachusetts 51.7% 45.8% 29.6% 47.2% 88.9% 20.7%
72 Syracuse 52.9% 57.6% 28.2% 23.7% 86.5% 12.3%
73 South Carolina 53.3% 57.9% 23.2% 28.6% 75.4% 13.4%
74 North Carolina 52.3% 52.2% 23.2% 29.1% 78.5% 25.0%
75 Appalachian State 53.5% 57.4% 28.6% 45.8% 78.6% 9.0%
S&P+ Rk Team AVERAGE MEDIAN STD. DEV AVG. CHANGE MAX MIN
76 Missouri 52.1% 57.3% 17.6% 24.2% 71.9% 24.5%
77 Kentucky 51.1% 38.4% 21.5% 31.2% 76.5% 32.6%
78 Akron 51.7% 55.5% 27.5% 26.7% 80.1% 11.7%
79 Wake Forest 50.2% 48.1% 13.2% 15.5% 69.5% 31.9%
80 Georgia Southern 54.7% 66.5% 30.9% 29.9% 77.9% 1.9%
81 Middle Tennessee 44.2% 37.1% 31.1% 37.0% 98.4% 14.3%
82 South Florida 48.2% 36.3% 32.8% 36.2% 87.0% 9.9%
83 Purdue 48.5% 53.9% 25.9% 29.7% 82.7% 11.1%
84 Washington State 47.3% 55.1% 17.7% 14.6% 60.0% 18.0%
85 Oregon State 46.5% 30.3% 29.7% 39.1% 83.1% 17.2%
86 UNLV 43.8% 34.3% 29.4% 29.9% 94.8% 15.0%
87 Cincinnati 45.8% 51.9% 17.2% 19.0% 62.7% 19.3%
88 Nevada 44.9% 43.1% 26.6% 30.0% 88.9% 13.5%
89 UTSA 45.7% 52.3% 19.8% 25.1% 66.9% 13.7%
90 Florida Atlantic 42.1% 41.0% 18.4% 25.1% 63.9% 18.8%
91 Central Michigan 41.7% 33.9% 20.3% 31.8% 70.7% 19.9%
92 Ball State 41.3% 47.3% 15.1% 9.1% 56.0% 22.4%
93 San Jose State 41.4% 41.9% 24.2% 35.1% 76.0% 11.7%
94 Colorado State 40.3% 37.0% 27.8% 14.9% 85.6% 11.2%
95 Army 41.5% 45.3% 18.4% 13.8% 56.5% 5.8%
96 Arkansas State 41.3% 38.1% 31.8% 47.4% 84.0% 6.7%
97 UL-Lafayette 40.2% 22.0% 30.2% 42.5% 74.6% 14.5%
98 Buffalo 35.7% 29.0% 14.5% 10.4% 61.6% 27.7%
99 Western Michigan 36.9% 28.7% 24.4% 31.9% 67.7% 12.9%
100 SMU 35.2% 33.3% 15.5% 18.4% 63.6% 16.6%
S&P+ Rk Team AVERAGE MEDIAN STD. DEV AVG. CHANGE MAX MIN
101 San Diego State 36.0% 36.9% 20.4% 20.8% 56.6% 14.9%
102 Connecticut 35.2% 33.7% 20.7% 17.6% 70.2% 11.8%
103 Rutgers 37.3% 37.0% 23.9% 31.4% 66.4% 4.5%
104 Eastern Michigan 34.1% 29.9% 18.6% 20.7% 69.9% 14.7%
105 New Mexico State 34.9% 41.4% 22.7% 28.2% 62.9% 8.2%
106 Florida International 33.5% 20.6% 29.1% 31.8% 83.7% 7.4%
107 Kent State 39.8% 36.4% 38.8% 58.6% 88.3% 1.1%
108 Old Dominion 31.9% 27.5% 21.2% 12.3% 56.3% 11.8%
109 Hawaii 31.6% 23.9% 25.1% 31.1% 73.6% 4.5%
110 Idaho 28.3% 29.5% 13.0% 20.1% 46.2% 10.8%
111 Southern Miss 28.8% 26.6% 18.7% 25.8% 62.0% 8.5%
112 Tulane 32.5% 17.6% 31.1% 36.5% 73.0% 4.0%
113 Texas State 29.8% 14.7% 28.8% 34.4% 77.9% 8.2%
114 Troy 29.7% 23.0% 25.0% 29.3% 68.7% 8.9%
115 Miami-OH 29.1% 15.6% 28.0% 36.6% 65.5% 6.9%
116 Colorado 29.3% 18.9% 27.4% 35.9% 68.2% 5.4%
117 Fresno State 27.3% 20.4% 22.5% 29.4% 56.6% 5.6%
118 Wyoming 23.9% 24.4% 9.8% 5.6% 36.2% 12.3%
119 South Alabama 29.7% 46.0% 23.5% 42.8% 47.9% 2.7%
120 Kansas 22.2% 23.8% 5.7% 6.2% 26.7% 12.4%
121 Georgia State 22.9% 26.0% 11.3% 16.7% 34.9% 9.4%
122 Rice 25.9% 21.0% 23.0% 29.9% 58.7% 5.2%
123 New Mexico 26.4% 11.5% 32.3% 31.2% 85.8% 3.0%
124 UL-Monroe 25.6% 10.4% 29.5% 41.1% 74.9% 5.1%
125 Charlotte 27.7% 15.8% 31.1% 36.8% 80.3% 0.3%
126 Central Florida 19.8% 14.4% 17.7% 14.5% 54.1% 6.3%
127 UTEP 12.8% 12.7% 10.8% 9.6% 25.9% 1.6%
128 North Texas 6.0% 6.3% 4.5% 4.5% 12.2% 0.2%

What do we learn from this? Well, don't bet much money on West Virginia or Ole Miss, first of all. But you've probably already figured that one out, especially as far as the Rebels are concerned.

What else?

  • For the second straight season, Florida State has been disturbingly consistent. The Seminoles never played elite football last year but were consistently in the 80 to 85 percent range, which, when combined with a nice combination of closers (i.e., a good quarterback and great kicker), was enough to remain undefeated until playing an elite Oregon team in the Rose Bowl. This year, surprisingly enough, the Seminoles have actually established a higher level of play despite the loss of Jameis Winston and so many of last year's difference-makers.
  • Ring the alarm bells: Ohio State has been consistent in the exact same way that Florida State was last year -- always a good team but never the elite team that we expected to see. The Buckeyes still have time, obviously, but this is a concern. Their max performance is actually the lowest of anybody in the top 30 ... and their minimum performance is one of the highest. That's how you survive a season without any major upsets and then get pummeled by a really good team down the line.
  • Seriously, Ole Miss' performance against Florida (11.3 percent) was just dreadful. The Gators were also very good in that shocking blowout win, but damn.
  • The ACC has been ridiculously unstable so far. Clemson and Florida State have been consistently good, but among teams in the top 60, Pitt, Virginia Tech, Boston College, and N.C. State have seen some of the craziest week-to-week changes.
  • Go ahead and figure out Texas and report back to me. I'm at a loss. Their percentile performance has gone up or down by an average of 44 percent on the percentile scale each week. It was 18 percent against Notre Dame and 89 percent against Rice, 13 percent against TCU and 85 percent against Oklahoma.
  • I put the more stable teams in green, but consistency and stability aren't always good things. Those haven't been particularly positive variables for North Texas, Wyoming, Kansas, or UTEP.

Posted by: Bill Connelly on 16 Oct 2015

0 comments