Friday, May 19, 2017

NCAA FOOTBALL 2017: METRICS LOOKAHEAD PART I

Here is Part I of a multiple entry blog looking at various metrics from last year & how they can assist with developing initial thoughts and projections for the 2017 season.

Here is a matrix that shows 2016 record, 2016 final SBPI rating & ranking along with OFF/DEF returning starters for 2017 (OFF numbers highlighted green represents QB returning) along with turnover margin (TOM) in total.  If you want to see offense & defense raw unit rankings please reference entry from April 11th using toolbar on right side of the page.  These are the first set of variables we will use to look towards the 2017 season.

RECORD
SOS ADJ #'s
2017 RS
WINS
LOSS
Team
RATING
RANK
OFF
DEF
TOT
TOM
14
1
Alabama
307.4
1
6
5
11
10
11
2
Ohio State
296.3
2
8
7
15
15
14
1
Clemson
296.1
3
5
7
12
(1)
12
2
Washington
275.1
4
7
6
13
18
10
3
Florida State
266.8
5
7
9
16
3
10
3
Michigan
261.2
6
4
1
5
7
8
4
LSU
255.6
7
6
5
11
0
11
3
Wisconsin
254.0
8
8
7
15
12
10
4
Virginia Tech
244.8
9
5
7
12
(1)
10
4
Colorado

241.7
10

9
3
12

6
10
3
USC
241.4
11
5
7
12
0
9
4
Louisville
237.9
12
5
7
12
(7)
8
5
Auburn
230.1
13
8
7
15
3
9
4
Miami (Florida)
219.1
14
7
8
15
9
7
6
North Carolina State
216.2
15
9
8
17
2
11
2
Oklahoma
215.5
16
9
7
16
0
8
5
Pittsburgh
212.7
17
6
4
10
1
10
3
Appalachian State
211.9
18
7
6
13
8
11
3
Penn State
211.4
19
9
7
16
1
10
4
Temple
209.6
20
6
4
10
6
13
1
Western Michigan
207.0
21
5
8
13
18
9
4
Florida
206.2
22
9
5
14
2
11
3
San Diego State
203.8
23
5
6
11
14
10
3
West Virginia
203.3
24
5
3
8
4
8
5
North Carolina

201.7
25

5
7
12

(2)
8
5
Georgia
201.5
26
7
10
17
8
9
4
Houston
200.9
27
8
7
15
(7)
9
4
Utah
197.9
28
5
6
11
6
11
3
Western Kentucky
195.0
29
4
6
10
2
10
3
Stanford
194.2
30
8
8
16
2
9
4
Nebraska
190.1
31
4
6
10
5
9
4
Tennessee
190.0
32
7
7
14
(2)
7
6
Baylor
189.4
33
7
7
14
(5)
8
5
Washington State
189.4
34
7
9
16
6
10
3
Troy
188.4
35
8
7
15
10
9
4
BYU
187.5
36
6
6
12
12
10
3
Tulsa
187.0
37
7
6
13
0
9
4
Kansas State
186.4
38
8
6
14
13
7
6
Arkansas
186.3
39
7
6
13
(4)
10
3
Oklahoma State
185.1
40
7
5
12
11
8
5
Texas A&M
184.1
41
5
7
12
3
4
8
Missouri
183.6
42
10
5
15
(3)
9
4
Toledo
179.3
43
5
7
12
(4)
10
3
Boise State
177.1
44
5
4
9
(9)
8
5
Iowa
176.8
45
7
8
15
6
4
8
Notre Dame
175.9
46
8
7
15
(4)
10
3
Air Force
175.4
47
6
1
7
6
11
2
South Florida
174.3
48
7
9
16
9
7
6
Colorado State
173.8
49
6
9
15
(1)
8
5
Army
172.9
50
9
7
16
(3)
6
7
Indiana
171.6
51
6
9
15
(6)
9
4
Minnesota
171.3
52
8
6
14
8
9
4
Georgia Tech
171.1
53
8
8
16
4
7
6
Northwestern
170.7
54
8
8
16
9
6
7
TCU
167.8
55
10
7
17
(4)
4
8
UCLA
167.1
56
9
6
15
(2)
5
7
Texas
166.8
57
7
10
17
(3)
7
6
Southern Mississippi
166.1
58
6
6
12
(17)
3
9
Michigan State
163.5
59
4
5
9
(5)
5
7
Mississippi
162.4
60
5
6
11
(3)
8
5
Memphis
161.4
61
9
6
15
8
7
6
Boston College
159.7
62
8
7
15
7
8
6
Wyoming
159.5
63
6
8
14
3
10
3
Old Dominion
159.4
64
8
6
14
13
8
5
Arkansas State
158.0
65
5
5
10
5
4
8
Oregon
156.7
66
8
9
17
(3)
4
8
Duke
155.9
67
7
7
14
(4)
6
7
Miami (Ohio)
155.2
68
8
8
16
1
9
5
Louisiana Tech
152.8
69
5
6
11
1
6
7
Mississippi State
152.0
70
7
6
13
7
8
6
Ohio
148.6
71
7
6
13
1
7
6
Kentucky
148.4
72
8
9
17
(7)
5
7
California
146.1
73
6
8
14
3
4
8
Tulane
143.9
74
8
8
16
9
7
6
Wake Forest
143.9
75
9
6
15
8
6
7
UCF
143.4
76
9
4
13
1
8
5
Middle Tennessee
142.2
77
6
6
12
(2)
5
7
Texas Tech
140.9
78
8
6
14
(4)
5
7
Georgia Southern
139.4
79
5
5
10
1
6
7
UTSA
139.3
80
7
7
14
3
5
7
Northern Illinois
138.7
81
5
7
12
(1)
5
7
SMU
138.7
82
9
5
14
2
9
4
New Mexico
138.4
83
7
3
10
(1)
4
8
Oregon State
137.8
84
7
8
15
1
3
9
Iowa State
136.8
85
6
6
12
(3)
7
6
Eastern Michigan
134.3
86
8
6
14
1
6
7
Maryland
134.0
87
7
7
14
(7)
9
5
Navy
133.4
88
5
8
13
2
3
9
East Carolina
132.5
89
5
6
11
(16)
6
7
South Carolina
128.5
90
10
6
16
7
4
8
Syracuse
128.0
91
9
11
20
(1)
6
7
Louisiana-Lafayette
127.1
92
6
7
13
1
6
7
South Alabama
124.4
93
4
6
10
(2)
6
7
Vanderbilt
124.0
94
9
7
16
4
3
9
Utah State
123.3
95
5
4
9
(5)
5
7
Nevada
123.1
96
5
9
14
4
4
8
Ball State
120.7
97
8
4
12
(10)
4
8
UNLV
120.4
98
9
4
13
2
9
4
Idaho
120.1
99
5
5
10
11
6
7
Central Michigan
119.7
100
8
6
14
(6)
3
9
Illinois
119.1
101
5
6
11
(2)
4
8
San Jose State
117.9
102
7
8
15
(1)
3
9
Georgia State
117.0
103
8
6
14
(5)
4
8
Cincinnati
112.1
104
5
5
10
1
5
7
Arizona State
111.2
105
7
8
15
(4)
3
9
Kent State
109.7
106
7
6
13
11
4
8
UTEP
102.2
107
5
6
11
(5)
3
9
Arizona
100.0
108
7
7
14
(7)
7
7
Hawai'i
99.5
109
8
6
14
(8)
2
10
Virginia
98.9
110
6
8
14
(9)
4
8
Charlotte
98.5
111
6
6
12
8
3
9
Purdue
98.2
112
5
8
13
(17)
3
9
Marshall
98.0
113
8
7
15
4
3
9
New Mexico State
96.5
114
6
9
15
1
5
8
North Texas
94.4
115
6
5
11
1
2
10
Kansas
94.1
116
8
4
12
(14)
5
7
Akron
92.9
117
9
6
15
(8)
3
9
Rice

90.0
118

8
8
16

(7)
2
10
Rutgers
89.5
119
5
8
13
(5)
2
10
Massachusetts
87.6
120
6
8
14
(10)
4
8
Louisiana-Monroe
86.8
121
7
9
16
(11)
4
8
Florida International
86.2
122
7
8
15
(9)
4
8
Bowling Green
85.6
123
6
7
13
(16)
2
10
Buffalo
85.4
124
6
9
15
(6)
3
9
Connecticut
84.9
125
7
7
14
(8)
3
9
Florida Atlantic
78.9
126
9
9
18
(5)
1
11
Fresno State
72.2
127
10
6
16
(9)
2
10
Texas State
38.6
128
7
7
14
(14)


Here is a small grid showing average returning starters by group heading into 2016 season along with heading into the 2017 season:

OFF
DEF
TOT
2017
6.84
6.59
13.43
2017 QB's
88
2016
6.85
6.53
13.38
2016 QB's
87

So we can see not much change in any of these figures at all YOY, which, for the most part, is expected.

How can we use each of these metrics to get a head start on our 2017 opinions for each team?  By examining the relationships between (keep in mind this write-up looks at each of these metrics in a silo):

·      2016 record & 2016 SBPI rating: the rating column (for example Alabama at 307.4) is a very strong predictor of team record, with the two showing a 2016 correlation of 83% across the entire 128 teams competing in FBS (in prior seasons that correlation is typically closer to 90%).  Using just this relationship the teams we want to target to be bullish on are those who have a high SBPI rating and single digit wins (such as LSU, Louisville, Auburn, Miami, NC State, Pittsburgh, Florida & North Carolina amongst Top 25 teams) – the reason is those teams played good football last year but likely caught a few tough breaks late in games, or played a very tough schedule (either opponents which is represented naturally in the SBPI rating or their home vs. away splits) that led to fewer wins than expected.  Teams we want to be bearish on are those who fall further down the SBPI ratings but won more games than expected (such as Hawaii, Idaho, Navy, New Mexico, Middle Tennessee, Louisiana Tech & Old Dominion) – these teams successful record vs. true statistical performance was driven by the opposite effect mentioned above; probably many late game good breaks & easier schedule as measured by either opponents or H/A split.
·      2016 SBPI rating & 2017 RS: the higher the number of returning starters the more likely a team is, at a minimum, to replicate last year’s performance or increase their SBPI rating which in turn would lead to a greater chance of winning more games.  If we focus on the average RS figure above of 13.43 teams with more returning players than that number should be as successful as they were last year, especially teams that return their QB.  However keep in mind there are always exclusions to any broad application of any statistics – such as bigger name teams that consistently recruit well (Alabama, Ohio State, Clemson and Florida State to name a few) will have a much easier time turning over new starters vs. middle to low pack Power 5 conference teams.  In the Group of 5 returning starters is typically a very solid place to start when projecting future success.
·      2016 record vs. 2016 TOM: as you can see from the figures above teams that perform well in TOM are typically very successful with their record while teams who struggle there do not usually post solid records.  In addition, although there are outliers to any high level application of statistical theories, teams will tend to aggregate towards the mean the more time that passes.  Examining this relationship on last year’s data shows the 4 CFB Playoff teams ranked #1-4 in my SBPI; however, we can see Clemson was a (1) in TOM while the other three teams were double digit favorable in this metric.  If we look closer at the worst teams from last year we can also see this metric as very predictive as the bottom 13 teams were all negative in TOM with an AVERAGE of (9.4) – just shy of (1) per game.  Those teams not only played poorly but also took terrible care of the football, or did not force enough turnovers, which is obviously the recipe for bad records.  There are many subset variables you can use to project turnovers committed and forced which will naturally drive the TOM figure.


That was a high level look at a few metrics I will use to set initial baselines on each team heading into the 2017 season.

In the coming weeks I will be using these metrics & posting conference breakdowns where I share my initial projections for each team.  In those entries there will be additional details on each team such as offense & defense figures from 2016 and how returning starters could potentially impact those.  Stay tuned – it should be great information – and who doesn’t enjoy reading college football analysis on the summer!

Thanks again for reading, please feel free to:
Email me directly: boss@thesportsboss.com
Visit my website: www.thesportsboss.com
Follow me on Twitter: @SportsBoss



COPYRIGHT: THE SPORTSBOSS, 2017