Adventures in Stats: 2014 Tennessee Football Opponent Rankings

SBNation's Bill Connelly (Football Study Hall) recently released an updated version of his projected team ratings for 2014. Connelly's rating system, called S&P+, is derived from play-by-play and drive data from all of the games of a particular football season. His projected ratings weight the data from the previous season by the number of returning starters and two-year recruiting ranking.

If you're naturally predisposed to doubt statistical modeling, consider the following caveat: while no rating system is perfect, Connelly's ratings of Tennessee's future opponents provide us useful insight into how those opponents performed in the past, and although we may disagree with the conclusions, allow us to have a discussion based on a common reference point.

So, with the caveats and disclaimers out of the way, let's take a look at the projections.

How Is S&P+ Derived?

First, however, it's useful to look at exactly what Connelly is attempting to measure. Like most of the advanced football statistics, S&P+ uses play-by-play and drive data, because plays provide substantially more data points (around 2,100 for the average football team) than the final score from each individual game (usually 12, minus the bowl game).¹ Play-by-play and drive success data is also filtered for garbage-time plays and possessions (based on point margin by quarter), to eliminate late game comebacks and extra scores which skew the final margin.²

There are three key components to the S&P+ (from the explanation on Football Outsiders):

• Success Rate: used to measure efficiency by determining whether every play of a given game was successful or not. The terms of success in college football: 50 percent of necessary yardage on first down, 70 percent on second down, and 100 percent on third and fourth down.

Success Rate is primarily a measure of consistency: whether or not an offense can remain on schedule, maintaining drives and churning out yardage on most plays. Other popular measures of efficiency, like yards per play, disguise whether or not an offense is consistently successful, or being bailed out by a handful of big plays.

• EqPts Per Play (PPP): An explosiveness measure derived from determining the point value of every yard line (based on the expected number of points an offense could expect to score from that yard line) and, therefore, every play of a given game.
  

The higher the Expected Points Per Play, the more an offense is scoring relative to how other teams score from that same position on the field. PPP is a useful measure in distinguishing quick strike offenses from grind-it-out offenses, even if both have similar total offensive production.

• Opponent adjustments: team S&P is compared to the expected output based on their opponents and opponent's opponents.
  

Success Rate and PPP combine to form S&P, a measure of how consistent and how explosive an offense or defense is, which is then adjusted to take into account the expected output.

Lastly, the plus part of S&P+ refers to a recently added (February 2013) Drive Efficiency stat, which attempts to measure average success at scoring or preventing points based on starting drive field position. Drive efficiency is the most opaque part of the formula, and Connelly hasn't provided (at least as far as I can find) much insight into how it figures into the overall rating.

Tennessee's Historical Ratings

Team	Year	S&P Total	S&P Rank	Offensive S&P	Off. Rank	Def. S&P	Def. Rank	Win-Loss	Adjusted W/L
Tennessee	2013	203.3	67	93.7	76	109.6	46	5-7	N/A
Tennessee	2012	213.4	50	118.9	19	93.4	81	5-7	9-3
Tennessee	2011	213.6	45	106.3	47	107.3	47	5-7	6-6
Tennessee	2010	208.4	51	105.0	47	103.4	49	6-7	5-8
Tennessee	2009	233.6	19	114.4	31	119.2	21	7-6	11-2

Over the last few years, Tennessee has fared about as well as might be expected in the S&P ratings, with 2012's offense and 2009's defense the high point for each unit. From a statistical perspective, Dooley's three teams performed better than their on-field results:

• 2011's much maligned offense actually performed no worse in non-garbage time than the previous year's offense;
• the loss to Kentucky looks even more inexplicable from a statistical perspective than it did at the time; and,
  
• 2012's defensive collapse overshadowed a top-twenty offensive performance and should result in permanently blackballing Sal Sunseri's chances for another defensive coordinator position. Following the loss of DC Justin Wilcox, the 2012 defense placed worse in every defensive category with essentially the same personnel.

Comparing Butch Jones and Derek Dooley, it's worth noting that the 2013 team managed to match the win-loss record of the previous two teams despite having by far the worst overall rating. Adding to the damning evidence against Sunseri is the rebound of last year's defense from 81 to 46, despite starting a true freshman cornerback (Cam Sutton) and a lightly regarded fifth year senior at linebacker (Dontavis Sapp) following the injury to Curt Maggitt.

Tennessee's 2014 Projection

Team	S&P Total	S&P Rank	Offensive S&P	Off. Rank	Def. S&P	Def. Rank
Tennessee	203.2	55	95.7	64	107.5	41

Connelly's projections use the previous year's ratings as a base, then adjusts for the two-year recruiting rank and number of returning offensive and defensive starters (each unit's starters are calculated separately). For Tennessee, Connelly projects improvement by both offensive and defensive units, with the improvement in offensive skill players offsetting the loss in the trenches and making the Vols one of the most improved teams in the nation. If you're looking for reasons to feel optimistic, well, consider that Connelly's starter data (if it's anything like Phil Steele's) doesn't count any of the following players:

• former starter Marcus Jackson (24 games/5 starts) returns at left guard, where he joins Mack Crowder (10/1) at center and Kyler Kerbyson (19/0) at right guard;
• part-time starter Marlin Lane (35/6) returns at running back; and,
• former starter Curt Maggitt (20/17) returns at linebacker.

2014 Schedule and Opponent Projections

Date	Team	Location	2013 S&P Rank	2013 W/L	S&P Total	Projected 2014 S&P Rank	Off. S&P	Off Rk	Def. S&P	Def. Rk
8/31	Utah State	Knoxville	26	9-5	204.4	49	84.5	100	119.9	16
9/06	Arkansas State	Knoxville	91	8-5	183.3	84	87.1	95	96.2	85
9/13	Oklahoma	Norman, OK	24	11-2	230.5	17	110.8	23	119.6	17
9/27	Georgia	Athens, GA	17	8-5	234.7	10	120.8	9	113.9	26
10/04	Florida	Knoxville	49	4-8	219.9	29	92	79	127.9	6
10/11	Chattanooga	Knoxville	NA	8-4	NA	NA	NA	NA	NA	NA
10/18	Ole Miss	Oxford, MS	29	8-5	222.6	24	100.7	50	121.9	13
10/25	Alabama	Knoxville	2	12-2	263.5	2	125.2	6	138.4	2
11/1	South Carolina	Columbia, SC	12	11-2	237.4	7	119.6	10	117.9	19
11/15	Kentucky	Knoxville	93	2-10	191.7	76	94.9	67	96.9	80
11/22	Missouri	Knoxville	14	12-2	223.9	22	108.6	28	115.3	24
11/29	Vanderbilt	Nashville, TN	64	9-4	198.3	61	94.5	69	103.8	54

Tennessee plays three opponents projected to be in the top ten: #2 Alabama, #7 South Carolina, and #10 Georgia.  Two of the three games come on the road, with both Georgia and South Carolina likely to be on upset alert after close games against the Vols in 2013.

More encouragingly, Tennessee also plays four opponents with projected rankings in the bottom half of Division 1: #84 Arkansas State, Chattanooga (FCS schools are not ranked), #76 Kentucky, and #61 Vanderbilt.³ Three of the four games are at home, and the fourth is a virtual home game in Nashville. Both Arkansas State and Vanderbilt are dealing with coaching changes spurred by head coaches leaving for better opportunities, so it's not certain that the new coaches are better than the ones they're replacing. Tennessee could and should win all four of those games, even if the rest of the season falls apart.

That leaves five opponents ranked between 17 and 29, and Utah State, another team most Tennessee fans are counting on beating. Unfortunately, Utah State has a significantly better defense than anyone outside the Mountain West realizes, and the return of Chucky Keeton at quarterback means that the offensive ranking is almost certainly too low. Whoever wins the starting quarterback position for the Vols will need to be extremely careful with the football against an opportunistic Utah State defense, or the 2014 season could start with a demoralizing loss.

The remaining four opponents, #17 Oklahoma, #22 Missouri, #24 Ole Miss, and #29 Florida, are exactly the type of opponents Butch Jones will have to defeat to return Tennessee to the promised land.

• S&P+ rates Oklahoma significantly lower than the preseason #1 hype they're getting based on a bowl game victory (red flag), but the Sooners are still a balanced, perennial top 20 team playing at home, and Bob Stoops hasn't forgotten how to coach.
• Missouri has to replace a number of difference-makers from last year's team, including the starting quarterback (James Franklin), both defensive ends (Kony Ealy and Michael Sam), and all three starting wide receivers. The other other Tigers also have to visit Neyland Stadium, where they were last seen narrowly pulling out a victory against a Derek Dooley-coached team.
• Ole Miss has the potential to be the toughest and most improved team on the schedule. Hugh Freeze has always been able to manufacture offense out of thin air, but improved recruiting means that he can pair whatever offensive wizardry he can conjure with a legitimate top fifteen defense... and the game is in Oxford, Mississippi, a setting that drips with Southern Gothic blood magic and voodoo chicanery. Beware the freakish creatures from the Island of Dr. Bo!
• Florida is the biggest wild-card on the schedule, following up an 11-2 season in which they finished #5 in the S&P+ with a 4-8 season in which they finished #49. What isn't likely to change anytime soon is the ferocious Gator defense, which ranked #4 in 2012 and remained a quite respectable #15 in 2013 despite the offense falling apart like a day old piece of poundcake. As long as Tennessee doesn't have to start Nathan Peterman, the Vols should feel good about their chances in this game playing at home, against a Kurt Roper offense still trying to learn a new offensive system. If the Vols can't beat Florida this year, it might be time to retire them as a rival.
  

^{1. For more on how the number of plays can vary team-to-team based on pace, and how Tennessee is likely to play this year, I encourage you to read Will's highly informative article from earlier this week: 10 Questions for 2014: Pace.↩}

^{2. For example, South Carolina-Florida in 2010 looks much different from the box score than it does from the final score.↩}

^{3. Lest you think that this is comparable to other cupcake schedules around the country, consider Florida's non-Florida State non-conference schedule (so we're removing the top two teams from each schedule: Oklahoma and Florida State). The Vols play #41 Utah State (9-5), #95 Arkansas State (8-5), and #140 Chattanooga (8-4), while Florida plays #144 Eastern Kentucky (6-6), #190 Idaho (1-11), and #208 Eastern Michigan (2-10)(all rankings from Sagarin’s 2013 final standings, since he ranks FCS teams). Tennessee plays three bowl eligible teams with a combined record of 25-14; Florida plays a bucket of hot garbage with a combined record of  9-27. Tennessee’s worst opponent is better than all three of Florida’s non-Florida State opponents.}↩