Thursday, April 29, 2021

2020 Yards Per Play: Pac-12

This week we head out west for our penultimate Power Five YPP review.   

Here are the 2020 Pac-12 standings.
So we know what each team achieved, but how did they perform? To answer that, here are the Yards Per Play (YPP), Yards Per Play Allowed (YPA) and Net Yards Per Play (Net) numbers for each Pac-12 team. This includes conference play only, with the championship game not included. The teams are sorted by division by Net YPP with conference rank in parentheses.
College football teams play either eight or nine conference games (typically fewer in 2020). Consequently, their record in such a small sample may not be indicative of their quality of play. A few fortuitous bounces here or there can be the difference between another ho-hum campaign or a special season. Randomness and other factors outside of our perception play a role in determining the standings. It would be fantastic if college football teams played 100 or even 1000 games. Then we could have a better idea about which teams were really the best. Alas, players would miss too much class time, their bodies would be battered beyond recognition, and I would never leave the couch. As it is, we have to make do with the handful of games teams do play. In those games, we can learn a lot from a team’s YPP. Since 2005, I have collected YPP data for every conference. I use conference games only because teams play such divergent non-conference schedules and the teams within a conference tend to be of similar quality. By running a regression analysis between a team’s Net YPP (the difference between their Yards Per Play and Yards Per Play Allowed) and their conference winning percentage, we can see if Net YPP is a decent predictor of a team’s record. Spoiler alert. It is. For the statistically inclined, the correlation coefficient between a team’s Net YPP in conference play and their conference record is around .66. Since Net YPP is a solid predictor of a team’s conference record, we can use it to identify which teams had a significant disparity between their conference record as predicted by Net YPP and their actual conference record. I used a difference of .200 between predicted and actual winning percentage as the threshold for ‘significant’. Why .200? It is a little arbitrary, but .200 corresponds to a difference of 1.6 games over an eight game conference schedule and 1.8 games over a nine game one. Over or under-performing by more than a game and a half in a small sample seems significant to me. In the 2020 season, which teams in the Pac-12 met this threshold? Here are Pac-12 teams sorted by performance over what would be expected from their Net YPP numbers.
Using the .200 cutoff, three teams saw their actual record differ significantly from their expected record based on YPP. And if we do a little rounding, that number jumps to five. Southern Cal, Stanford, and Colorado all exceeded their expected records and each performed well in close games. The trio combined for a 9-1 record in one-score conference games and the lone defeat came when Stanford hosted Colorado. Meanwhile, Arizona State and UCLA significantly underperformed. The Sun Devils and Bruins were on the other end of the close game spectrum, posting a combined 1-6 record in one-score conference games. 

Can You Count to 70?
If Kevin Sumlin was a dead man walking prior to the Territorial Cup rivalry against Arizona State, Herm Edwards put him out of his misery (metaphorically of course, we wish Kevin Sumlin nothing but the best as he enjoys all his buyout loot) on December 11th. The Sun Devils scored the games first 42 points and led 63-7 late in the third quarter before showing some mercy and only winning 70-7. Thanks to College Football Reference's fantastic Game Finder, I can proudly inform you that  Arizona became just the twenty third Power Five/BCS conference team this century to give up 70 points to a conference opponent (in regulation). Compared to those other twenty two teams that had a steroid aided home run number put on them, how bad was this shellacking?  

Arizona became the fourth Power Five/BCS team to give up 70 points to a conference opponent that did not finish with a winning record (Arizona State finished 2-2 in the pandemic shortened 2020 season). Here are the other teams that managed this feat.
We're cheating a bit here as Arizona State was likely to finish with a winning record over a full season, but at 2-2, the math checks out. Still, its hard to argue against Rutgers having the worst showing among these four teams in allowing 80 points to Rich Rod's first West Virginia team that outside of Rutgers only managed to beat a pair of MAC schools. 

Arizona also became the fifth Power Five/BCS team to give up 70 points to a conference opponent at home. 
Those other teams that gave up 70 points at home were either involved in a competitive shootout (Texas Tech led Oklahoma State in the fourth quarter and was within three points with under five minutes to go) or lost to College Football Playoff contenders (Washington made the playoff in 2016 as did Clemson is 2020 and Michigan missed it by a few inches). Meanwhile, Arizona State finished 2-2. 

When the quality of opponent is combined with the venue, Arizona may have posted the worst conference defeat of any Power Five/BCS team this century. And if they didn't, they were a close second behind Rutgers. 

Thursday, April 22, 2021

2020 Adjusted Pythagorean Record: Mountain West

Last week we looked at how Mountain West teams fared in terms of yards per play. This week, we turn our attention to how the season played out in terms of the Adjusted Pythagorean Record, or APR. For an in-depth look at APR, click here. If you didn’t feel like clicking, here is the Reader’s Digest version. APR looks at how well a team scores and prevents touchdowns. Non-offensive touchdowns, field goals, extra points, and safeties are excluded. The ratio of offensive touchdowns to touchdowns allowed is converted into a winning percentage. Pretty simple actually.

Once again, here are the 2020 Mountain West standings.
And here are the APR standings with conference rank in offensive touchdowns, touchdowns allowed, and APR in parentheses. This includes conference games only with the championship game excluded. Since teams played a varied number of games (some played as many as eight or as few as four), the rankings are on a per game basis, not raw totals. 
Finally, Mountain West teams are sorted by the difference between their actual number of wins and their expected number of wins according to APR.
I used a game and a half as an arbitrary line of demarcation to determine if a team significantly over or under-performed relative to their APR. By that standard, no team saw their actual record differ significantly from their APR.

Divisional Dominance
The Mountain West scrapped its divisional format for the 2020 campaign and emulated the Big 12 by sending the two teams with the best record to the conference title game. For the fourth year in a row, one of those participants was Boise State. While the Broncos ultimately fell to San Jose State in the Mountain West Championship Game, they continued to dominate their divisional brethren. The Broncos managed to play five regular season conference games in 2020 and four of those came against teams that normally play in the Mountain Division of the Mountain West (Air Force, Colorado State, Utah State, and Wyoming). The only Mountain Division member that avoided the Broncos was New Mexico. The Broncos won all four of those games with relative ease and in the process ran their divisional win streak to 19 games. In 2016, the Broncos actually lost back to back games against Mountain Division teams, dropping a classic in Laramie to eventual division champs Wyoming and their regular season finale to Air Force. In the four seasons since, the Broncos have lost four times to Mountain West teams (with two coming in the conference title game), but all those losses have come against teams that play in the West Division (Fresno State twice, San Diego State, and San Jose State). How does this divisional win streak compare to Group of Five teams? Pretty favorably. 
The Broncos own the longest divisional win streak among Group of Five teams. and they will have a chance to make it an even twenty when they travel to Utah State on September 25th. 

Thursday, April 15, 2021

2020 Yards Per Play: Mountain West

This week, we follow Horace Greeley's advice and go west.   

Here are the 2020 Mountain West standings.
So we know what each team achieved, but how did they perform? To answer that, here are the Yards Per Play (YPP), Yards Per Play Allowed (YPA) and Net Yards Per Play (Net) numbers for each Mountain West team. This includes conference play only, with the championship game not included. The teams are sorted by Net YPP with conference rank in parentheses.
College football teams play either eight or nine conference games (typically fewer in 2020). Consequently, their record in such a small sample may not be indicative of their quality of play. A few fortuitous bounces here or there can be the difference between another ho-hum campaign or a special season. Randomness and other factors outside of our perception play a role in determining the standings. It would be fantastic if college football teams played 100 or even 1000 games. Then we could have a better idea about which teams were really the best. Alas, players would miss too much class time, their bodies would be battered beyond recognition, and I would never leave the couch. As it is, we have to make do with the handful of games teams do play. In those games, we can learn a lot from a team’s YPP. Since 2005, I have collected YPP data for every conference. I use conference games only because teams play such divergent non-conference schedules and the teams within a conference tend to be of similar quality. By running a regression analysis between a team’s Net YPP (the difference between their Yards Per Play and Yards Per Play Allowed) and their conference winning percentage, we can see if Net YPP is a decent predictor of a team’s record. Spoiler alert. It is. For the statistically inclined, the correlation coefficient between a team’s Net YPP in conference play and their conference record is around .66. Since Net YPP is a solid predictor of a team’s conference record, we can use it to identify which teams had a significant disparity between their conference record as predicted by Net YPP and their actual conference record. I used a difference of .200 between predicted and actual winning percentage as the threshold for ‘significant’. Why .200? It is a little arbitrary, but .200 corresponds to a difference of 1.6 games over an eight game conference schedule and 1.8 games over a nine game one. Over or under-performing by more than a game and a half in a small sample seems significant to me. In the 2020 season, which teams in the Mountain West met this threshold? Here are Mountain West teams sorted by performance over what would be expected from their Net YPP numbers.
Colorado State and Wyoming were the only Mountain West teams to see their actual record differ significantly from their expected record based on YPP. Both the Rams and Cowboys underachieved. Colorado State had meltdowns on special teams, allowing five such touchdowns in their last two games. This turned a probable loss against Boise State into a blowout and a toss-up game with San Diego State into a loss. For Wyoming, the culprit was close game luck. The Cowboys finished 0-3 in one-score games, losing in overtime to Nevada, by one point to New Mexico, and in a blizzard to Boise State. 

Services Academies Against the Spread
Conventional wisdom holds that service academies (Air Force, Army, and Navy) perform well against the spread (ATS) as underdogs, particularly as large underdogs, and not so well as favorites (especially large favorites). The logic behind that assumption is solid. Service academies run unique offenses that are difficult to prepare for and those offenses tend to depress the number of possessions in a game which leads to increased variance. Theoretically, variance would reward underdogs as less variance would lead to a more 'true' result. Is this conventional wisdom accurate or just another well known fact that is disproven by data? All three service academies have long-tenured coaches so we have a relatively large sample of them in different ATS roles. Let's dive into the data and see what it tells us.

Since this is the Mountain West recap, Air Force comes first alphabetically, and Troy Calhoun is the longest-tenured head coach at a service academy, we'll begin with the Falcons. Here is how the Falcons have performed ATS against FBS opponents in all non-bowl games as a favorite and underdog under Calhoun. 
Conventional wisdom jumps to an early lead. The Falcons have covered just 44% of the time as a favorite under Calhoun, but over 58% of the time as an underdog. What about Army under Jeff Monken?
Monken has been at Army seven years, which represents only half of Calhoun's time at Air Force, but is still enough time to begin to develop a track record. Army's cover record as a favorite under Monken is nearly identical to Air Force's under Calhoun (44%). They have been less successful than Air Force as an underdog, but have still performed much better in that role versus as a favorite. Finally, let's check in on Navy under Ken Niumatalolo
It's eerie how similar the three service academies have performed as favorites. At 45% under Niumatalolo, Navy has done just a smidge better than Air Force and Army. As an underdog, they have covered more than 60% of the time. Cumulatively, the service academies have covered slightly more than 44% of the time as favorites and nearly 58% of the time as underdogs under their current head coaches.
So conventional wisdom seems to hold. However, we were interested in how these teams perform as big favorites or underdogs. Let's look at how the service academies have done as both double-digit favorites and double-digit underdogs. Once again, we'll start with Air Force.
The Falcons are worse as double-digit favorites (40%) than they were as favorites of any size (44%), but they are also worse as double-digit underdogs (55%) than they were as an underdog overall (58%). Does the trend hold for Army?
In a small sample (ten games), the Black Knights are better as a double-digit favorite (50%) than they were as a favorite of any size (44%). Their record as a double-digit underdog (50%), was basically indistinguishable from their record as an underdog of any size (51%). And finally, Navy. 
Like Army, the Midshipmen were slightly better as a double-digit favorite (48%) than they were as a favorite of any size (45%). And like Air Force, they were worse as a double-digit underdog (55%) than they were as an underdog overall (60%). Here is how the service academies stack up cumulatively. 
For the most part, the conventional wisdom holds. It's a long-term losing proposition to back a service academy as a favorite, be it double-digit or otherwise. Conversely, backing a service academy in the underdog role is a long-term winning strategy. However, at least under their current administrations, you'd be better off taking them as smaller underdogs. Don't be bashful just because they aren't catching double-digits.    

Thursday, April 08, 2021

2020 Adjusted Pythagorean Record: MAC

Last week we looked at how MAC teams fared in terms of yards per play. This week, we turn our attention to how the season played out in terms of the Adjusted Pythagorean Record, or APR. For an in-depth look at APR, click here. If you didn’t feel like clicking, here is the Reader’s Digest version. APR looks at how well a team scores and prevents touchdowns. Non-offensive touchdowns, field goals, extra points, and safeties are excluded. The ratio of offensive touchdowns to touchdowns allowed is converted into a winning percentage. Pretty simple actually.

Once again, here are the 2020 MAC standings.
And here are the APR standings with conference rank in offensive touchdowns, touchdowns allowed, and APR in parentheses. This includes conference games only with the championship game excluded. Since teams played a varied number of games (some played as many as six or as few as three), the rankings are on a per game basis, not raw totals. 
Finally, MAC teams are sorted by the difference between their actual number of wins and their expected number of wins according to APR.
I use a game and a half as a line of demarcation to determine whether a team significantly over or under-performed relative to their APR. By that standard, Ball State significantly overachieved and Northern Illinois significantly underachieved. Ball State also overachieved relative to their YPP and we went over some reasons for that last week. Northern Illinois underachieved thanks to a poor record in one-score games. While the Huskies finished 0-6 overall, half of their losses came by a touchdown or less. 

Two Winless Teams
The 2020 college football season was unique to say the least. For many teams, the season did not start on time, and every team dealt with testing, postponements, and canceled games. The quirky schedule was bound to produce some odd results, and it did for the MAC. In 2020, a pair of MAC teams finished conference play without a win. Bowling Green and Northern Illinois combined to go 0-11 in MAC play. Northern Illinois lost by an average of about two touchdowns per game while Bowling Green was abysmal, losing by about five touchdowns per game. Dating back to the 2019 season, the Huskies and Falcons have combined to lose fourteen consecutive MAC games. But I'm not here to mock and gawk at all the losing that has befallen them the past two seasons. I know misery loves company, so I wanted to see how often two teams in a given conference finished winless. I decided to limit my analysis to the BCS and CFB Playoff eras (since 1998). In order for a conference to have two teams finish winless (assuming each team plays a full schedule), it needs to have more teams than conference games. And it doesn't hurt to separate those teams into divisions. While college football divisions existed before the BCS, this date serves as a nice arbitrary divide and gives us nearly a quarter century worth of data to examine. Without further adieu, here are the other conferences to produce multiple winless teams since 1998. 
The WAC expanded to sixteen teams in 1996, but the super-conference was not long for the world. 1998 was the last year of the expanded WAC and the conference continued to hemorrhage teams for the next decade or so until it finally dissolved after the 2012 campaign. In 1998, Hawaii and UNLV not only finished 0-8 in WAC play, they both failed to win any non-conference games, combining for an 0-23 overall record. Their struggles did lead to high-profile coaching changes that were at least moderately successful. Hawaii hired June Jones and was bowl eligible in seven of the next nine seasons, culminating with a Sugar Bowl appearance in 2007. UNLV hired legendary Southern Cal coach John Robinson, and while he left Sin City with a losing record, he guided the Rebels to their third ever bowl appearance in 2000. 
In 1999, the MAC pulled off a unique hat trick. Ball State and Buffalo both finished winless in MAC play and overall. In addition, Marshall, behind the pinpoint accuracy of Chad Pennington, finished unbeaten and ranked tenth in the final AP Poll. Ball State was in the midst of a twenty-one game losing streak that they would not break until October of the following season. Meanwhile, Buffalo at least had the excuse of playing their first season at the FBS level since dropping football after the 1970 season. 
Our first power conference appearance on the list. Mississippi State and Vanderbilt combined for an 0-16 SEC record, but did manage to win five non-conference games. However, Connecticut, Furman, Jacksonville State, Memphis, and Troy were not exactly formidable opponents in 2002.
The MAC is back for another appearance in 2004. Western Michigan did manage to win a game outside the league, trouncing Tennessee-Martin (runteldat) in their opener. Before bolting for Conference USA, UCF closed out their brief MAC tenure by losing all eleven of their games under former Notre Dame coach George O'Leary. The Knights would extend the losing streak to thirteen in 2005 before rallying to win eight of their final nine regular season games and play in the inaugural Conference USA Championship Game.  
The SEC returns for their second appearance on this list. The 2012 campaign was the first year of a fourteen team SEC and marked the final season for Gene Chizik at Auburn and Joker Phillips at Kentucky. However, both schools would make solid hires in Gus Malzahn and Mark Stoops respectively. 
We finally have another power conference that is not the SEC! NC State, in their first season under future Tennessee coach Dave Doeren finished winless in league play and would push their ACC losing streak to twelve games before a victory against Syracuse in November of 2014 broke it. Meanwhile, Virginia somehow beat a solid BYU team to open the season before dropping all eight of their conference games (with seven coming by double digits). 
The 2013 Mountain West saw a good head coach finish without a conference win (Troy Calhoun) and a bad head coach do the same (Norm Chow). Air Force rebounded and won ten games the next season, while Hawaii would finish the Norm Chow era with a 4-25 conference record.
2013 also gave us one final appearance of the SEC on this list. Bret Bielema (Arkansas) and Mark Stoops (Kentucky) both finished winless in conference play in their first season in charge. Bielema got his team to a bowl game faster, winning the Texas bowl the next season, but Stoops had more sustainability. He is still coaching the Wildcats while Arkansas has already hired two head coaches since firing Bielema. 
After 2013, big time losing took a break until 2019. Old Dominion and UTEP combined to lose all their conference games in 2019, but they both won tight games against FCS opponents in their openers to avoid the ignominy of losing all their games. 
Look who decided to join the MAC with a reduced schedule winless campaign in 2020. FIU and UTEP combined for an 0-7 conference record against an abbreviated schedule. By the letter, they meet the requirements, but that 0-7 mark is not even a full season's worth of conference games in normal times. Of course, technically, three Conference USA teams finished winless in league play in 2020 as Old Dominion elected to not play a fall schedule.

There you have it. Those are the conferences that produced multiple winless teams since 1998. Excluding the outlier that was 2020, it happened nine times between 1998 and 2019, which makes it more common than I thought. 

Thursday, April 01, 2021

2020 Yards Per Play: MAC

This week, we examine the MAC, a conference that started late but managed to have two teams finished ranked in the AP Poll for the first time since 2003.   

Here are the 2020 MAC standings.
So we know what each team achieved, but how did they perform? To answer that, here are the Yards Per Play (YPP), Yards Per Play Allowed (YPA) and Net Yards Per Play (Net) numbers for each MAC team. This includes conference play only, with the championship game not included. The teams are sorted by division, by Net YPP with conference rank in parentheses.
College football teams play either eight or nine conference games (typically fewer in 2020). Consequently, their record in such a small sample may not be indicative of their quality of play. A few fortuitous bounces here or there can be the difference between another ho-hum campaign or a special season. Randomness and other factors outside of our perception play a role in determining the standings. It would be fantastic if college football teams played 100 or even 1000 games. Then we could have a better idea about which teams were really the best. Alas, players would miss too much class time, their bodies would be battered beyond recognition, and I would never leave the couch. As it is, we have to make do with the handful of games teams do play. In those games, we can learn a lot from a team’s YPP. Since 2005, I have collected YPP data for every conference. I use conference games only because teams play such divergent non-conference schedules and the teams within a conference tend to be of similar quality. By running a regression analysis between a team’s Net YPP (the difference between their Yards Per Play and Yards Per Play Allowed) and their conference winning percentage, we can see if Net YPP is a decent predictor of a team’s record. Spoiler alert. It is. For the statistically inclined, the correlation coefficient between a team’s Net YPP in conference play and their conference record is around .66. Since Net YPP is a solid predictor of a team’s conference record, we can use it to identify which teams had a significant disparity between their conference record as predicted by Net YPP and their actual conference record. I used a difference of .200 between predicted and actual winning percentage as the threshold for ‘significant’. Why .200? It is a little arbitrary, but .200 corresponds to a difference of 1.6 games over an eight game conference schedule and 1.8 games over a nine game one. Over or under-performing by more than a game and a half in a small sample seems significant to me. In the 2020 season, which teams in the MAC met this threshold? Here are MAC teams sorted by performance over what would be expected from their Net YPP numbers.
Eventual MAC champ Ball State significantly exceeded their expected record as did Akron. For Ball State, it was all about close game performance. The Cardinals finished 4-1 in one-score conference games, winning all their tight games after dropping their opener by a touchdown to Miami. Meanwhile, Akron was not especially lucky, they just happened to be really bad and benefited from beating a team that was statically a little better than they were. That team happened to be Bowling Green. While the Falcons significantly under-achieved relative to their YPP numbers, they were still quite bad; just not catastrophic. In five games, Bowling Green forced just a single turnover, while committing eleven, for a MAC worst turnover margin of -10.  

What's the Deal with the MAC Championship Game?
In the 2020 MAC Championship Game, an undefeated Buffalo team entered as a nearly two touchdown favorite against an upstart outfit from Muncie, Indiana that had not finished with a winning record since 2013. The upstarts traded scores with Buffalo for a quarter and a change before a pair of touchdowns with under two minutes to go in the half (including a scoop and score) gave them some breathing room. In the second half, Buffalo put together some nice drives, but capped those drives with a touchdown just once and Ball State was able to escape Detroit with their first MAC title since 1996. The upset was not out of character for the MAC Championship Game. The last three games have been won outright by underdogs, and underdogs have covered the past five games. Overall, the MAC Championship Game has not gone well for the betting favorite, particularly Against the Spread (ATS).
In nearly a quarter century of action, the betting favorite has won the game outright just around 58% of the time. That isn't great, especially considering some of the larger spreads in the history of the game. However, their performance ATS has been ghastly. They have covered just 26% of the time! To put that number in perspective, if you had blindly placed the same monetary bet on the underdog plus the points in each MAC Championship Game, your rate of return would be over 41%!

It certainly appears the MAC Championship Game is a destination where favorites go to die, or at least kill the profits of their investors. However, before we make any bold proclamations, we need to compare the MAC to other Group of Five (G5) conferences. The MAC was the second non-power conference to hold a championship game (after the WAC in 1996), but does have the longest running championship game among the G5 since the WAC's championship tilt lasted just three seasons. With the Sun Belt adding a title game in 2018, all the current G5 conferences have a championship game. Here is how all those other title games have played out for favoites, both straight up and ATS.
Compared to the MAC, favorites in the other G5 conferences have fared much better in winning the title outright. They have won about 71% of the time (versus about 58% for MAC favorites). However, like MAC favorites, they do have a losing record ATS, although their winning percentage (44%) is much higher than those of MAC favorites (26%). Why is this? Why do MAC favorites struggle in their title games as compared to other G5 favorites? I think the biggest culprit is the venue.

In the early days of the MAC Championship Game, when Marshall was the dominant program, it was played at campus sites. The team with the best conference record got to host the game for the first seven seasons of its existence. However, since 2004, the game has been played on a neutral site in Detroit, Michigan. That makes the MAC an outlier among the G5. Every other conference rewards their best regular season team by allowing them to host the title game (the 2020 Mountain West Championship Game was played in Las Vegas because public health restrictions in California prevented San Jose State from hosting the game). When you compare MAC favorites to other G5 favorites and consider venue, the MAC results don't seem so bizarre.
The MAC doesn't have a very large sample of home favorites, but the fact their home favorites never lost in four tries compares favorably to the 21-6 straight up record of the other G5 home favorites. Their ATS record, while .333 is just one game away from .500 which would equal the ATS record of the other G5 favorites. 
When it comes to road favorites, MAC favorites have actually done better than other G5 favorites in a very limited (and not statistically valid) sample, both straight up and ATS.
The neutral field is where MAC favorites have struggled. MAC favorites have played four times as many games on a neutral field as the other G5 combined. In addition, most of those other G5 neutral site title games came last century. If the MAC rewarded its best teams with homefield advantage, these upsets would slowly go the way of the WAC
What's the deal with the MAC?
I for one, knew they would be back.