40
THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Embed Size (px)

Citation preview

Page 1: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

THEORY OF WINNING

Coaching, recruiting and spending in college football

2010 Alabama Mr. Football Coty Blanchard

Page 2: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Table of Contents

Introduction How to predict a win Data sources Initial Model Out of sample prediction Practical applications Next steps

Jason Campbell, Auburn University

Page 3: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Authors Introduction

McDonald “Mac” Mirabile Manager of Strategic & Financial Analysis at

WWF Undergraduate and graduate thesis on the

predictors of a successful transition from college to NFL

Prior academic publications on topics such as biases in college football polls, the NFL Rookie Cap, the Wonderlic Test, and the Peer Effect in the NFL draft

Mark Witte Assistant Professor at College of Charleston Generally awesome guy

Page 4: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Topic Introduction

The importance of winning in college Shapes alumni support, attendance Influences quality of recruiting Self-enforcing cycle

Page 5: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

How to predict a win

Vegas point spread, totals, and money line theoretically capture all available information under the efficient market hypothesis (EMH)

Existing literature consistently enforces EMH, though there are some published examples of deviations and profitable strategies within wagering markets

Within the framework of this paper, we will assume EMH holds within college football wagering markets and will measure the success of our developed models relative to the baseline Vegas model

Page 6: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Predicting Wins with the Vegas Line

Bubble chart illustrates the home team’s winning percent by the Vegas Line, with the size of the bubble based on the number of observations

Page 7: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Predicting Wins with the Vegas Line

Bar chart of home team’s winning percentage by the Vegas line

Page 8: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

The Vegas Line model

Home Win (0,1) = b1*Line + error

This model within our data explains 29% of the variation in wins (Pseudo R2).

The line coefficient is 0.1091, with a standard error of 0.00437, and an Odds Ratio of 1.115

Interpretation: for each additional point a team is favored, their odds of winning increase by 11.5%

Non-linear model shows similar results

Page 9: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Improving the Vegas Line model

Can it be done, or does the Vegas line incorporate all publically available information?

To test this, we added several variables: Home, Away win and losing streaks Home, Away AP Rankings, Top 25 matchups Dummy variables for conference games, neutral

field matchups, and night games Distance between schools, stadium size, rivalry

information Conference dummy variables

Page 10: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Improving the Vegas Line model

Effect DF Wald Pr > ChiSqLine 1 285.4691 <.0001ETP 1 1.1213 0.2896HWS 1 0.522 0.47HLS 1 0.8024 0.3704AWS 1 1.8483 0.174ALS 1 0.1004 0.7513Hrank 1 0.7195 0.3963Arank 1 0.1588 0.6903HNR 1 1.591 0.2072ANR 1 1.5452 0.2138TrueT25 1 0.2535 0.6146ConfGame 1 0.003 0.9566Neutral 1 0.001 0.9743Nightgame 1 0.3414 0.559Stadium 1 1.078 0.2992Distance 1 0.0145 0.9042Rivalry 2 2.0766 0.3541Conf 12 11.5154 0.4853

• Table on left shows these additional variables and a their corresponding Wald Chi2 statistics• The Vegas line successfully incorporates all available information. • Adding more explanatory variables does not improve the model’s fit. • None of the added variables are statistically significant as their importance is already captured in the Line variable.

Page 11: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Data Sources

To develop a model of winning without utilizing the Vegas line, the authors gathered data on the following topics: Game-specific factors Institutional factors/history Team player composition/recruiting Team coach factors/history

We will discuss the collection and organization of this data next

Page 12: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Game-specific Factors

Matchup data comes from Covers.com Data includes game location, time, day,

conference information Each matchup (home vs away) is one

observation in the dataset There are about 500 games per season

Page 13: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Institutional Factors & History Historical team performance comes from

CFBDatawarehouse.com University football team expenditure and

student body size data come from the Equity in Athletics website

Each of these variables is reported for a particular year (e.g., Michigan’s historical team performance through 2007 and their team expenditure data for the 2008 season would all be used as predictors for the 2008 season matchups)

Page 14: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Team player composition and recruiting Class recruiting data comes from Rivals.com,

Scouts.com, and Prepstar.com Recruiting classes in 2005 (RS-Senior), 2006

(Senior / RS-Junior), 2007 (Junior, RS-Sophomore), 2008 (Sophomore, RS-Freshman), an 2009 (Freshman) are used as predictors for the 2009 season matchups.

Due to the NFL draft, transfers, and general attrition, these variables are imperfect measures of the talent comprising a team in a particular season

Page 15: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Team coach factors and history Historical coach performance comes

from CFBDatawarehouse.com Coach biographical information comes

from various university athletics department websites

Each of these variables is reported for a particular year (e.g., Michigan’s coach’s historical performance through 2007 would be used as a predictor for the 2008 season matchups)

Page 16: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Summary Statistics of Model variables

Variable Mean Std Dev Minimum Maximum N N Misshome_win 0.6 0.49 0 1 3418 0Stadium 55.458 22.246 16 107.501 3418 0

Student_Body_H 17.883 8.017 0.002 43.026 3304 114cum_winpct_adf 0.008 0.11 -0.392 0.408 3418 0

total_expense_all_football_ldf 0.109 0.563 -1.611 2.386 3204 214class_rank_scouts_l4_adf -5.843 31.682 -106.25 100.75 3340 78

first_year_HC_H 0.072 0.258 0 1 3418 0first_year_HC_A 0.083 0.275 0 1 3418 0coach_age_adf 0.239 11.764 -46 48 3418 0

coach_experience_adf 0.447 12.456 -46 54 3417 1seasons_coach_adf 0.323 11.091 -41 42 3418 0cum_winpct_coach_adf 0.017 0.246 -0.826 0.84 3418 0

nfl_years_adf -0.028 3.369 -16 16 3417 1Home_Coach_Minority 0.047 0.213 0 1 3418 0Away_Coach_Minority 0.048 0.213 0 1 3418 0

Page 17: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Initial Model

Matchup-specific variables:• Stadium Size• Home team student size

School-specific variables:• Cumulative Team Win Pct Diff • Log Diff of Total Team expenditures

Team-specific variables (Difference home – away):• Scouts.com weighted average class ranking

Coach-specific variables (Difference home – away) :• First year head coach Home team dummy • First year head coach Away team dummy • Coach age• Coach experience (assistant + HC)• Head coach seasons• Lifetime Coach Win Pct Diff• Years as NFL player• Home team’s head coach minority dummy• Away team’s head coach minority dummy

Stadium 1.004 1 1.008Student_Body_H 1.005 0.99 1.017cum_winpct_adf 1.714 0.63 4.647

total_expense_all_fo 2.529 2.03 3.153class_rank_scouts_l4 0.992 0.99 0.996

first_year_HC_H 0.754 0.54 1.047first_year_HC_A 1.291 0.93 1.784coach_age_adf 0.984 0.97 0.996

coach_experience_adf 1.002 0.99 1.013seasons_coach_adf 1.007 1 1.019

cum_winpct_coach_adf 6.806 4.43 10.45nfl_years_adf 0.961 0.93 0.99

Home_Coach_Minority 0.581 0.38 0.892Away_Coach_Minority 1.866 1.16 2.997

Odds Ratio EstimatesEffect Estimate 95% Wald

Confidence

N: 2,948R-Square: .215

Page 18: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Initial Model - InterpretationsMatchup-specific variables:• Stadium Size – for every additional 10,000 seats, the home team is 4% more likely to win • (also considered game time, location, rivalry variables)

School-specific variables:• Log Diff of Total Team expenditures – the odds ratio of the % difference (home/away) in team spending of 2.5 suggests that a team spending 100% more (twice as much) is 150% more likely to win, (Alternative, equivalent interpretation: odds of winning increase 15% for each 10% increase in excess of your opponent’s expenditures)

Team-specific variables (all Difference home – away) :• Scouts.com average class ranking – for each unit increase in average class ranking between the home and away, the home team is 1% more likely to win

Coach-specific variables (all Difference home – away) :• First year head coach dummy variables – marginally significant and coefficients in the direction one would expect

• Diff in HC’s ages – for each additional year in age difference b/w the Home and Away team’s coach, the home team is 1% less likely to win

• Diff in HC’s cumulative Win % – for each 1% difference in lifetime win percentage between the home team’s HC and the away team’s HC, the home team is about 6% more likely to win

• Years as NFL player – for each additional year of NFL playing experience between the home team’s HC and the away team’s HC, the home team is about 4% less likely to win

• Home team Head Coach Minority – minority coaches are 42% less likely to win than non-minority coaches at home• Away team Head Coach Minority – home teams are 87% more likely to win when playing against a minority coach

Page 19: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Out of Sample prediction

Analysis Variable : Vegas_Model_Correct

sample Correct Incorrect

% Correct

In 1,450 541 72.8%

Out 612 233 72.4%

Analysis Variable : Our_Model_Correct

sample Correct Incorrect

% Correct

In 1,420 648 68.7%

Out 603 278 68.4%

Both models have comparable in and out of sample performance

Page 20: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Out of Sample by Line

Vegas line does a better job predicting everything except games where the line is between -2 and +2

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

01) underdog 28+

02) underdog 21 -27

03) underdog 14 -20

04) underdog 7 -13

05) underdog 3 -6

06) underdog .5 -2

12) favorite 0 -2

07) favorite 3 -6

08) favorite 7 -13

09) favorite 14 -20

10) favorite 21 -27

11) favorite 28+

Vegas Model Our Model Home Win

Page 21: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 Season (SEC results)

Data from 2004-2008 used to develop the model Data from 2009 used in an out-of-sample validation

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTAlabama 11 1 92% 12 0 100% 11 1 92%Arkansas 8 4 67% 7 5 58% 11 1 92%Auburn 7 5 58% 7 5 58% 10 2 83%Florida 12 0 100% 11 1 92% 11 1 92%Georgia 11 1 92% 7 5 58% 8 4 67%Kentucky 4 8 33% 6 6 50% 8 4 67%LSU 10 3 77% 9 4 69% 10 3 77%Mississippi 5 6 45% 7 4 64% 7 4 64%Mississippi St. 2 9 18% 4 7 36% 7 4 64%South Carolina 6 5 55% 5 6 45% 10 1 91%Tennessee 8 4 67% 6 6 50% 8 4 67%Vanderbilt 3 7 30% 1 9 10% 8 2 80%

Model Actual Evaluation

Note: Non Div1A opponents not scored/modeled

Page 22: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Practical Applications

Predict 2010 season results – conference standings, national champion, before a single game has been played

Page 23: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Next steps

What can be added to the model? New sources of data (attendance,

compensation/bonus – impute missing values based on relative rank of team within conference?)

Additional data cleanup (game time, more years 2001-2003)

Different estimation methodologies

Page 24: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

BACKUP/OLD SLIDES BEGIN HERE

Page 25: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Who is hiring minority coaches?

The coach is more likely to be young (see coach_age), belong to a historically crappy program (Cum_WinPCT_School_H) as well as belong to a recently crappy program (MA5_Win_PCT_School_H) of relatively newer schools (School_Seasons_H) and larger schools (Stadium).

Standard WaldError Chi-Square

Intercept 8.4961 2.1216 16.0358 <.0001Stadium 0.0311 0.0116 7.1846 0.0074

School_Seasons_Home -0.014 0.00771 3.2851 0.0699Cum_WinPCT_School_Home -6.9228 3.0907 5.0172 0.0251MA5_Win_PCT_School_Home -2.1436 1.3669 2.4592 0.1168

Coach_Age_H -0.1657 0.0537 9.526 0.002Coach_Experience_H 0.0479 0.0497 0.9289 0.3352

Analysis of Maximum Likelihood EstimatesParameter Estimate Pr > ChiSq

Page 26: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Predicting recruiting classes

GLM estimation of dependent variable: Scouts class ranking

Previous year and 5-year MA Win % impact recruiting

Previous classes are also good predictors of current year’s class ranking

Conference impacts recruiting

Parameter Estimate Std Error Chi2 Pr > Chi2Intercept 43.38 4.49 34.58 52.17 93.4 <.0001

Prev_WinPCT_School -9.69 3.34 -16.23 -3.15 8.43 0.0037MA5_Win_PCT_School -15.52 4.81 -24.96 -6.08 10.4 0.0013

Class_Rank_Scouts_Lag1 0.27 0.04 0.20 0.34 55.8 <.0001Class_Rank_Scouts_Lag2 0.18 0.04 0.10 0.26 20.6 <.0001Class_Rank_Scouts_Lag3 0.10 0.04 0.02 0.18 6.23 0.0126Class_Rank_Scouts_Lag4 0.13 0.04 0.05 0.20 10.1 0.0015

ACC -17.79 2.96 -23.58 -11.99 36.2 <.0001Big 12 -16.18 2.84 -21.75 -10.61 32.4 <.0001

Big East -15.02 2.84 -20.59 -9.45 27.9 <.0001Big Ten -16.26 2.85 -21.84 -10.67 32.5 <.0001

Conf USA -7.57 2.38 -12.23 -2.90 10.1 0.0015I-A Ind -12.09 3.73 -19.41 -4.77 10.5 0.0012Indep -11.87 8.44 -28.41 4.68 1.98 0.1597MAC 0.17 2.30 -4.34 4.68 0.01 0.9412MWC -3.67 2.54 -8.66 1.32 2.08 0.149

Pac-10 -18.79 3.05 -24.78 -12.81 37.9 <.0001SEC -21.78 3.05 -27.75 -15.80 51.1 <.0001

Sun Belt -3.33 2.68 -8.59 1.93 1.54 0.2149WAC 0.00 0.00 0.00 0.00 . .

Wald 95%

Alabama (2010) = 43.4 – (9.7*1) – (15.5*.77) + (.27*2) + (.18*1) + (.1*22) + (.13*18) – 21.8 = 3 (Actual rank 4)Auburn (2010) = 43.4 – (9.7*.615) – (15.5*.66) + (.27*16) + (.18*18) + (.1*6) + (.13*9) – 21.8 = 15 (Actual rank 5)Vanderbilt (2010) = 43.4 – (9.7*.167) – (15.5*.38) + (.27*72) + (.18*74) + (.1*87) + (.13*61) – 21.8 = 63 (Actual rank 61)

Page 27: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 out of sample (A-F)

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTAkron 4 6 40% 2 8 20% 8 2 80%Alabama 11 1 92% 12 0 100% 11 1 92%Arizona 4 8 33% 7 5 58% 7 5 58%Arizona St. 8 3 73% 3 8 27% 6 5 55%Arkansas 8 4 67% 7 5 58% 11 1 92%Arkansas State 3 5 38% 1 7 13% 6 2 75%Auburn 7 5 58% 7 5 58% 10 2 83%BYU 10 2 83% 10 2 83% 10 2 83%Ball State 4 5 44% 2 7 22% 5 4 56%Baylor 2 9 18% 3 8 27% 8 3 73%Boise St. 10 3 77% 13 0 100% 10 3 77%Boston College 10 2 83% 7 5 58% 9 3 75%Bowling Green 5 6 45% 5 6 45% 7 4 64%California 9 3 75% 7 5 58% 6 6 50%Central Michigan 7 3 70% 8 2 80% 9 1 90%Cincinnati 10 2 83% 11 1 92% 11 1 92%Clemson 10 3 77% 8 5 62% 11 2 85%Colorado 9 2 82% 3 8 27% 5 6 45%Colorado State 6 4 60% 2 8 20% 4 6 40%Connecticut 5 7 42% 7 5 58% 10 2 83%Duke 2 7 22% 3 6 33% 6 3 67%East Carolina 6 6 50% 7 5 58% 11 1 92%Eastern Michigan 1 9 10% 0 10 0% 9 1 90%Florida 12 0 100% 11 1 92% 11 1 92%Florida State 9 3 75% 6 6 50% 9 3 75%Fresno State 7 5 58% 7 5 58% 8 4 67%

Model Actual Evaluation

Page 28: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 out of sample (G-M)

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTGeorgia 11 1 92% 7 5 58% 8 4 67%Georgia Tech 11 2 85% 10 3 77% 10 3 77%Hawaii 7 4 64% 4 7 36% 6 5 55%Houston 2 9 18% 9 2 82% 4 7 36%Idaho 1 12 8% 8 5 62% 6 7 46%Illinois 3 8 27% 2 9 18% 8 3 73%Indiana 3 8 27% 3 8 27% 11 0 100%Iowa 8 4 67% 11 1 92% 9 3 75%Iowa St. 1 10 9% 4 7 36% 8 3 73%Kansas 4 8 33% 4 8 33% 12 0 100%Kansas State 4 6 40% 4 6 40% 8 2 80%Kent State 0 10 0% 4 6 40% 6 4 60%Kentucky 4 8 33% 6 6 50% 8 4 67%LSU 10 3 77% 9 4 69% 10 3 77%Louisiana Tech 4 6 40% 3 7 30% 7 3 70%Louisville 4 7 36% 3 8 27% 10 1 91%Marshall 3 9 25% 6 6 50% 9 3 75%Maryland 3 8 27% 1 10 9% 7 4 64%Memphis 5 6 45% 1 10 9% 7 4 64%Miami-Florida 4 8 33% 8 4 67% 8 4 67%Miami-Ohio 1 9 10% 0 10 0% 9 1 90%Michigan 8 3 73% 4 7 36% 7 4 64%Michigan State 5 7 42% 5 7 42% 8 4 67%Middle Tennessee 4 6 40% 7 3 70% 7 3 70%Minnesota 1 10 9% 5 6 45% 7 4 64%Mississippi 5 6 45% 7 4 64% 7 4 64%

Model Actual Evaluation

Page 29: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 out of sample (M-S)

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTMississippi St. 2 9 18% 4 7 36% 7 4 64%Missouri 7 4 64% 7 4 64% 9 2 82%NC State 3 7 30% 3 7 30% 8 2 80%Nebraska 12 2 86% 10 4 71% 12 2 86%Nevada 7 6 54% 8 5 62% 12 1 92%New Mexico 1 10 9% 1 10 9% 9 2 82%New Mexico State 0 12 0% 2 10 17% 10 2 83%North Carolina 6 5 55% 6 5 55% 7 4 64%North Texas 1 7 13% 1 7 13% 6 2 75%Northern Illinois 7 4 64% 6 5 55% 8 3 73%Northwestern 5 7 42% 6 6 50% 7 5 58%Notre Dame 8 3 73% 6 5 55% 7 4 64%Ohio 5 6 45% 7 4 64% 5 6 45%Ohio State 11 0 100% 9 2 82% 9 2 82%Oklahoma 10 2 83% 7 5 58% 9 3 75%Oklahoma State 6 6 50% 8 4 67% 8 4 67%Oregon 8 5 62% 10 3 77% 7 6 54%Oregon St. 5 7 42% 7 5 58% 8 4 67%Penn State 9 3 75% 10 2 83% 9 3 75%Pittsburgh 4 6 40% 7 3 70% 7 3 70%Purdue 6 5 55% 4 7 36% 5 6 45%Rice 5 6 45% 2 9 18% 8 3 73%Rutgers 9 0 100% 5 4 56% 5 4 56%SMU 8 3 73% 7 4 64% 8 3 73%San Diego State 5 5 50% 3 7 30% 6 4 60%San Jose St. 3 8 27% 1 10 9% 9 2 82%

Model Actual Evaluation

Page 30: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 out of sample (S-U)

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTSouth Carolina 6 5 55% 5 6 45% 10 1 91%South Florida 5 5 50% 5 5 50% 8 2 80%Southern Cal 12 1 92% 9 4 69% 8 5 62%Southern Miss 6 6 50% 6 6 50% 8 4 67%Stanford 2 11 15% 8 5 62% 7 6 54%Syracuse 2 9 18% 3 8 27% 8 3 73%TCU 9 2 82% 10 1 91% 8 3 73%Temple 4 4 50% 5 3 63% 7 1 88%Tennessee 8 4 67% 6 6 50% 8 4 67%Texas 13 1 93% 13 1 93% 14 0 100%Texas A&M 6 7 46% 6 7 46% 11 2 85%Texas Tech 8 4 67% 8 4 67% 8 4 67%Troy State 7 3 70% 6 4 60% 9 1 90%Tulane 0 10 0% 1 9 10% 9 1 90%Tulsa 10 1 91% 4 7 36% 5 6 45%UAB 0 11 0% 4 7 36% 7 4 64%UCF 5 6 45% 6 5 55% 10 1 91%UCLA 11 2 85% 7 6 54% 7 6 54%UL-Lafayette 2 6 25% 4 4 50% 6 2 75%UL-Monroe 0 8 0% 2 6 25% 6 2 75%UNLV 2 8 20% 4 6 40% 6 4 60%UTEP 5 6 45% 4 7 36% 8 3 73%Utah 8 4 67% 9 3 75% 11 1 92%Utah State 2 9 18% 3 8 27% 8 3 73%

Model Actual Evaluation

Page 31: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

2009 out of sample (V-W)

School Wins Losses Pct Wins Losses Pct Correct Incorrect PCTVanderbilt 3 7 30% 1 9 10% 8 2 80%Virginia 3 8 27% 3 8 27% 7 4 64%Virginia Tech 10 3 77% 10 3 77% 11 2 85%Wake Forest 3 7 30% 4 6 40% 7 3 70%Washington 2 10 17% 5 7 42% 9 3 75%Washington St. 3 9 25% 1 11 8% 10 2 83%West Virginia 6 6 50% 8 4 67% 10 2 83%Western Michigan 4 5 44% 2 7 22% 7 2 78%Wisconsin 11 1 92% 9 3 75% 10 2 83%Wyoming 1 9 10% 5 5 50% 6 4 60%

Model Actual Evaluation

Page 32: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Other considerations (backup slide) Off the field model .18 On the field model .26 Are the coefficients robust? Future problems: things that recruits like –

new stadiums, new weight rooms, facilities

Could we do a recruiting paper modeled on NCAA football recruiting info – coach history, academic prestige, location, tv time, etc

Page 33: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Out of Sample prediction (intercept)

Analysis Variable : Vegas_Model_Correct

sample Correct Incorrect

% Correct

In 837 315 72.7%

Out 384 132 74.4%

Analysis Variable : Our_Model_Correct

sample Correct Incorrect

% Correct

In 787 365 68.3%

Out 352 164 68.2%

Both models have comparable in and out of sample performance

Page 34: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Friday Meet with profs about research Present to a class Lunch Seminar presentation Dinner

Page 35: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Models

To begin, we will look at each of these data sources and its relationship to our outcome variable individually.

Because each of these data sources is described with dozens of potential variables, this initial modeling will inform our final set of models where data from all possible sources are considered in development.

All models are developed using a Logit function as our outcome variable, Home Win, is binary. We will discuss the resulting coefficients as Odds Ratios to aid interpretation.

Page 36: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Model 1: Game specific factors

Odds Ratio Estimates

Effect Point Estimate 90% Wald

Confidence Limits

Neutral 0.804 0.62 1.05

Nightgame 0.868 0.76 0.99

Stadium 1.018 1.01 1.02

Conference ACC vs WAC 0.573 0.4 0.83

Conference Big East vs WAC 0.741 0.49 1.12

Conference Big Ten vs WAC 0.444 0.3 0.66

Conference Big Twelve vs WAC 0.665 0.46 0.96

Conference CUSA vs WAC 0.576 0.4 0.84

Conference INDP vs WAC 0.554 0.36 0.86

Conference MAC vs WAC 0.789 0.55 1.13

Conference Mountain West vs WAC 0.766 0.52 1.12

Conference NC vs WAC 0.992 0.74 1.34

Conference Pac Ten vs WAC 0.483 0.33 0.7

Conference SEC vs WAC 0.347 0.24 0.51

Conference Sun Belt vs WAC 0.955 0.64 1.43

day_of_week Fri vs Wed 1.184 0.65 2.16

day_of_week Mon vs Wed 0.662 0.31 1.43

day_of_week Sat vs Wed 1.407 0.81 2.44

day_of_week Sun vs Wed 1.103 0.55 2.21

day_of_week Thu vs Wed 1.686 0.92 3.09

day_of_week Tue vs Wed 1.727 0.85 3.52

Distance 1 1 1

Page 37: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Model 1: Game specific factors Other considered variables

Distance b/w schools Rivalry game (major/minor/none)

Other variables to consider in the future: Game-time (need to clean some data)

Page 38: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Model 2: Institutional factors & history

Odds Ratio Estimates

Effect Point Estimate 95% Wald

Confidence Limits

Cum_Losses_School_H 1.005 1.003 1.007

MA5_Wins_School_H 1.039 1.03 1.047

TOTAL_EXPENSE_ALL_Fo 0.972 0.932 1.014

EFMaleCount_H 1 1 1

EFFemaleCount_H 1 1 1

Cum_Losses_School_A 0.996 0.994 0.998

MA5_Win_PCT_School_A 0.082 0.048 0.14

TOTAL_EXPENSE_ALL_Fo 1.028 0.986 1.072

school_seasons_ldf 0.53 0.363 0.775

cum_winpct_adf 185.1 20.63>999.999

total_expense_all_fo 3.366 2.149 5.273

school_seasons_31t75 1.912 1.269 2.882

school_seasons_31t75 0.711 0.491 1.031

school_seasons_m101_ 0.715 0.586 0.873

Page 39: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Model 2: Institutional factors & history

Other considered variables

Other variables to consider in the future:

Page 40: THEORY OF WINNING Coaching, recruiting and spending in college football 2010 Alabama Mr. Football Coty Blanchard

Model 3: Recruiting

Odds Ratio Estimates

Effect Point Estimate 95% Wald

Confidence Limits

Cum_Losses_School_H 1.005 1.003 1.007

MA5_Wins_School_H 1.039 1.03 1.047

TOTAL_EXPENSE_ALL_Fo 0.972 0.932 1.014

EFMaleCount_H 1 1 1

EFFemaleCount_H 1 1 1

Cum_Losses_School_A 0.996 0.994 0.998

MA5_Win_PCT_School_A 0.082 0.048 0.14

TOTAL_EXPENSE_ALL_Fo 1.028 0.986 1.072

school_seasons_ldf 0.53 0.363 0.775

cum_winpct_adf 185.1 20.63>999.999

total_expense_all_fo 3.366 2.149 5.273

school_seasons_31t75 1.912 1.269 2.882

school_seasons_31t75 0.711 0.491 1.031

school_seasons_m101_ 0.715 0.586 0.873