Tuesday, August 26, 2014

Super Bowl hangover: myth or reality

You’ve probably heard about the Super Bowl hangover. The myth is that teams playing in the big game ultimately flop the next season. There are a variety of reasons cited, the most common one being it's too expensive to hang on to players in the salary cap era. 

I decided to test the post-Super Bowl dropoff theory by using data from all the first 47 Super Bowls. We'll see how the Seattle Seahawks and Denver Broncos do this season.  

And off we go ... 

Excluding last year’s game, Super Bowl participants won at a 78.4% clip in the regular season (the champs did slightly better than the losers). The following season, those same teams averaged a 65.5% winning percentage. 

Over a 16 game schedule, the difference in percentages translates into an average of 2.06 more games won by the Super Bowl participant in the year of the run than the year after. I also found that the year preceding the Super Bowl run, teams had a 66.8% winning percentage, slightly above the year before. 

Of the 94 teams to make the Super Bowl, 29 (30.9%) did not make the playoffs the following season. Spaced out over the 16 game schedule, 34% had a winning percentage lost three more games the following season. 

There's definitely some evidence that a Super Bowl hangover occurs, but the majority of teams still have solid seasons. The median win-loss record for the teams the next season is 11-5. It's nearly impossible to sustain long-term success at this level unless a team has a Hall of Fame quarterback, like Tom Brady or Peyton Manning.

Hometown curse?
To this day, no team has won a Super Bowl in its own stadium, sparking speculation about another jinx - the hometown curse.

Teams that have hosted a Super Bowl won just 44.1% of their games in the regular season, with the median at 43.8% (7-9).

The San Francisco 49ers were the closest to win one at home, winning Super Bowl XIX at Stanford, maybe 45 minutes away from Candlestick. The Los Angeles Rams lost a Super Bowl in Pasadena, roughly half an hour away from the Coliseum. 

Most of the games have taken place in Florida (15), Southern California (10) and New Orleans (10). Florida is broken down into Miami (10), Tampa Bay (4) and Jacksonville (1). Southern California is as follows: Pasadena (5), San Diego (3) and Los Angeles (2).

Of the 51 hosts, just 16 teams were above .500 in the regular season and recorded double-digit wins nine times. Last season, 11 teams (34.3%) had double-digit wins in the regular season.

I know what you're thinking. Historically, the New Orleans Saints have stunk and their appearances are dragging down the rest of the hosts. Well, even excluding the Saints, the rest of the hosts won at a 46.8% clip. 

Note: I counted the Raiders and Rams as host teams when it was in Pasadena, when those teams were located in Los Angeles even though it wasn't their home stadium because of the close proximity between the two cities. I also counted the 49ers and Stanford's.  

Sunday, August 17, 2014

AP Preseason Poll: Historically the most overrated team is (spoiler)

The Associated Press Top 25 poll is out. For once, polls will not have a direct impact on who gets to play in the championship. Instead, a committee will meet to determine the best teams.

Now that the polls don't matter, I thought it would be fun to figure out the most overrated and underrated teams in the BCS era. I couldn't find receiving votes in the AP poll prior to 2003, so this will just focus on the last 11 seasons.

To compile the most overrated team in college football, I compared the preseason AP Top 25 poll with the final poll (after the bowls). The team with the greatest discrepancies in the polls are either overrated or underrated. To calculate this I took each team's percentage of possible votes for each preseason and postseason poll because there are differing amount of voters each year.

I've included all teams in current Power Five conferences, plus Notre Dame, BYU, Boise State, Cincinnati and Connecticut.

Let's get right to it.

If we were to conduct an AP Top 25 based on the preseason polls from the last 11 seasons, with 1600 maximum points (64 voters), it would look like this (2014 not included):

1. Oklahoma 1279
2. Ohio State 1164
3. Southern California 1150
4. LSU 1112
5. Texas 1098
6. Georgia 927
7. Florida 920
8. Virginia Tech 755
9. Florida State 747
10. Alabama 720
11. Michigan 706
12. Oregon 619
13. West Virginia 557
14. Auburn 546
15. Miami 539
16. Wisconsin 507
17. Boise State 421
18. Tennessee 376
19. Nebraska 373
20. Louisville 370
21. Clemson 370
22. TCU 360
23. California 334
24. Notre Dame 332
25. Iowa 324

Receiving Votes: Oklahoma State 320, South Carolina 313, Stanford 276, Texas A&M 273, Penn State 256, Arkansas 247, Kansas State 220, Missouri 202, Pittsburgh 181, Michigan State 143, Ole Miss 141, Arizona State 138, Georgia Tech 128, Virginia 125, Purdue 123, Texas Tech 119, Utah 97, Maryland 96, UCLA 87, BYU 86, NC State 80, Kansas 76, Washington 70, North Carolina 68, Rutgers 62, Oregon State 58, Illinois 53, Boston College 36, Wake Forest 24, Northwestern 20, Cincinnati 19, Minnesota 16, Baylor 10, Colorado 6, Connecticut 4.1, Arizona 3.6, Washington State 2.3, Vanderbilt 2.1, Iowa State 1.9, Mississippi State 0.5, Kentucky 0.1.

None of Top 10 is the least bit surprising, although I was a little at how low Oregon is. Indiana, Syracuse and Duke did not receive a single vote.

On the flip-side, if we were to conduct an AP Top 25 based on the postseason polls from the last 11 seasons, it would look like this (same scale):

1. Ohio State 1083
2. LSU 1040
3. Southern California 984
4. Oklahoma 953
5. Alabama 849
6. Texas 809
7. Oregon 757
8. Boise State 732
9. Georgia 710
10. Florida 673
11. Auburn 661
12. Virginia Tech 576
13. TCU 541
14. Florida State 493
15. Stanford 454
16. Michigan 436
17. Wisconsin 426
18. Louisville 414
19. West Virginia 410
20. Penn State 365
21. Utah 350
22. Iowa 344
23. Oklahoma State 341
24. South Carolina 338
25. Missouri 329

Receiving Votes: Clemson 305, Michigan State 300, Notre Dame 293, Miami 284, Arkansas 270, Tennessee 237, Kansas State 221, Cincinnati 216, BYU 214, Nebraska 208, Texas Tech 207, Texas A&M 205, Boston College 196, Ole Miss 182, California 173, Baylor 160, UCLA 134, Arizona State 119, Kansas 117, Oregon State 117, Georgia Tech 96, Washington State 95, Pittsburgh 87, Rutgers 84, Maryland 67, Mississippi State 58, Wake Forest 54, Northwestern 49, Purdue 48, Illinois 40, Minnesota 33, Vanderbilt 29, Virginia 24, Duke 18, NC State 13, Washington 11, Kentucky 5, Connecticut 2.7, Colorado 1.2, Arizona 1.0.

North Carolina, Iowa State, Syracuse, and Indiana did not receive a vote, with the latter two not receiving a vote in the preseason poll either. North Carolina actually received points in the preseason poll five seasons, but did not perform well enough to finish with any after the postseason. Iowa State only had 21 preseason points, so their lack of getting any after the season isn't a big dissappointment.

From these two polls, we can figure out the most overrated and underrated teams in FBS football.

Top 10 most overrated teams:
1. Oklahoma -327
2. Texas -288
3. Michigan -270
4. Miami -255
5. Florida State -254
6. Florida -248
7. Georgia -217
8. Virginia Tech -179
9. Southern California -165
10. Nebraska -165

These schools are strong brands. Half of these teams actually finished in the Top 10 in the postseason poll. Miami, Michigan, Florida State and Nebraska finished outside the Top 10, but all have historically great programs.

Top 10 most underrated teams:
1. Boise State +311
2. Utah +253
3. Cincinnati +197
4. TCU +180
5. Stanford +178
6. Boston College +160
7. Michigan State +157
8. Baylor +151
9. Oregon +137
10. Alabama +129

The top four teams from this list played in a non-major conference during at least part of the observed period. Utah and TCU currently play in big conferences, but they haven't found the same success. The main reason Cincinnati is so high on the list is because of its 2009 season, one which it entered the bowls undefeated. Stanford is the first team from this list that has spent the 11 years in a power conference.

Friday, August 15, 2014

Does the NFL preseason matter?

The NFL preseason is underway, which means football is back. It's a time full of overreactions and unnecessary panic attacks. For instance, ESPN's Ron Jaworksi said Arizona Cardinals third-string quarterback Logan Thomas played like a Pro Bowl quarterback

These games are important for a lot of players trying to make a team's final roster, but are they vital in determining a team's success in the ensuing regular season? A good preseason could be a sign that a team has significant roster depth. 

The most infamous overreaction occurred in 2008, when the Detroit Lions won all four of their preseason games, and ended up going winless in the regular season. 

I went back six years and found a team's preseason record and regular season record. The findings: 

Teams that won 100% of their preseason games: 39.8% regular season win percentage; 4/11 (36.4%) made postseason.
Teams that won 75% of their preseason games: 54% regular season win percentage; 22/51 (43.1%) made postseason.
Teams that won 50% of their preseason games: 54.1% regular season win percentage; 29/66 (43.9%) made postseason.
Teams that won 25% of their preseason games: 42.9% regular season win percentage; 12/39 (30.8%) made postseason.
Teams that won 0% of their preseason games: 47.5% regular season win percentage; 4/15 (26.7%) made postseason.

The numbers do show that teams that are 2-2 or 3-1 in the preseason do slightly better than their 1-3 or 0-4 counterparts, but just don't win too many games in the preseason. 

The average difference in regular and preseason win percentage is 23.1%, which would be the equivalent of roughly 3.7 regular season games. 

In addition, the correlation between preseason and regular season win percentage is just .085, although there are really only five options for a team in the preseason (excluding Hall of Fame game), while there are 17 possibilities in the regular season (excluding a rare tie). 

Here are all of the instances in graphical form, and keep in mind some data points have more than one instance represented. 

From 2008-13, comparing NFL preseason and regular season win percentage.