Hello friends. We sit here in the end of May, a mere 90 something days from the start of college football season - so close, but so far. With TCU baseball heading toward another uncharacteristically early end to their season, though, college football is all we’ve got for the rest of the summer.
Last week, I introduced an expected points matrix, examining how game states and expected points can measure performance, culminating in some useful Game State Graphs to diagnose a team’s issues on offense and defense. I’ve got more on that, but this week, I’m setting that project aside to consider a more-TCU-centric question, namely one about TCU’s schedule.
I bought the Athlon preview magazine this week, and you should too - it’s a wonderful summer companion with the bulk of news, narratives, and team previews being a solid foundation for understanding the college football landscape come this fall. In reading the preview, though, I have early fixated on TCU’s schedule and how the location of each game might affect TCU’s potential ceiling this season.
This is far from a deep-preview of TCU’s schedule, that’ll come in due time. Instead, today, I’m going to look at how TCU football has responded to home-away mix and sequencing historically, and what that might tell us about their possible outcomes this fall.
@ Iowa State
@ Kansas State
@ Ok. State
@ Texas Tech
That seven game conference stretch to end the season looks daunting. At times in the past, TCU coach Gary Patterson discussed his philosophy of the season as being in three parts: out of conference play, conference games until the first bye, and then the home stretch. Using that framework, TCU has some serious work laid out for then in the home stretch. From October 26th to November 23, a five week span, TCU has games against three of the projected top four teams in the conference, two of those in Oklahoma, and a rivalry game against a rapidly improving Football Team from Waco. A physical Kansas State game to kick things off doesn’t do the Frogs any favors on that stretch, either.
The Big 12 embraces the nine-game conference schedule, lauding their round-robin as generating One True Champion. A necessary consequence of a nine game schedule is that the bulk of home and away games flips every other year, and teams oscillate between 5 and 4 home conference games yearly. Aside from an odd 2012 scheduling manipulation to facilitate A&M leaving the Big 12 and TCU joining, TCU’s groups of home and away have been:
Group A: Texas, Kansas, WVU, Baylor
Group B: Oklahoma, Texas Tech, Kansas State, Oklahoma State, Iowa State
That’s a pretty lopsided group, on a couple of levels. First, Group B has averaged 7.8 wins a year per team, whereas group A averages 6. Second, group B involves trips to three of the four weirdest places to play in the Big 12 (Lubbock, Stillwater, and Ames), and all in the same year. Third, Group B contains three of the four best teams in the Big 12 since TCU joined, plus the only team besides Baylor who considers TCU a rival (Tech). Home and road splits should matter, right?
Breaking Down the Schedule - History
To determine how schedule grouping affects TCU performance, I’m going to start broadly and get more granular in analyzing the yearly performance based on only home and away groupings. I want to start with non-parametric analysis - I’ll do some regression at the end of this post, but first, I just want the descriptives from the data.
First, I want to know how TCU fares in years when they have Group A at home vs Group B.
Group A Home Years: (Win-Loss, conf win-loss, conf home win-loss, conf away win-loss)
2013: 4-8, 2-7, 1-3, 1-4
2015: 11-2, 7-2, 4-0, 3-2
2017: 11-3, 7-2, 4-0, 3-2
Group B Home Years: (Conf win-loss, conf home win-loss, conf away win-loss)
2014: 12-1, 8-1, 5-0, 3-1
2016: 6-7, 4-5, 1-4, 3-1
2018: 7-6, 4-5, 3-2, 1-3
In the six years of the new schedule arrangement, TCU has averaged 2.67 wins against Group A, and ... 2.67 wins against Group B. Granted, Group B has more teams, so let’s do percentages: TCU’s win pct against Group A is 66.7% and 55.3% against Group B.
TCU’s home win pct overall is 66.7%, with three undefeated seasons at home (#protectthecarter). The Frogs’ away win pct overall is 51.8%. In years with Group A at home, TCU boasts a 75% home win pct. In years with Group B at home, that win pct is 60%. Looking at the away win percentages, in years TCU travels to Group A, they win 58.3% of games, whereas they win only 46% traveling to Group B.
What have we learned? Well, observationally, we see that three of TCU’s four best years came when TCU hosted Group A, which suggests that the number of road games matters less than the quality of opponent does. TCU football, like every team, struggles when they cannot win at home, and so hosting the “historically easier” secures those valuable home wins. Then, on the base of 3-4 conference wins against worse teams, TCU has the double benefit of 1) insurance against those road losses and 2) presumably some benefit to playing well at home, working out the kinks, improving chemistry, etc, which then translates to some out-of-conference wins.
A Simple Model for Predicting Team Wins
Wins this season are a function of, among many other things, wins last season, which group TCU has at home, and opponent group quality (S&P margin). I ran a quick regression on that data to see what we can learn about TCU’s ceiling this year.
Some quick observations while collecting the data - all three of TCU’s 10 plus wins seasons have come when they’ve hosted a group with an average S&P margin less than 10. The Frogs have had high quality teams, of course, but they’ve also taken advantage of weak home slates to climb atop the Big 12 standings. In 2018, if you exclude Oklahoma, TCU faced a home slate with an average 7.49 S&P margin (for comparison, that’s equivalent to playing Vanderbilt for every non-OU home game last year). Even in a down year, and even dropping a few games they shouldn’t have with injury and off-field issues, the Frogs still clawed their way to bowl eligibility on the back of crucial home wins vs mediocre teams Iowa State, Oklahoma State, and Kansas State.
The model doesn’t come close to anything causal, but what it does tell us is that there is a negative correlation of about 30% between the Average S&P Margin of a season’s home group and the wins that season. That confirms our observational theory, that the quality of the home games, rather than the number, determines in part TCU’s win totals.
Let’s take that simple model and look forward: What can we expect as a baseline expectation for TCU’s wins this fall? (Again, disclaimer: this model is not predictive, we don’t have enough data, and the error term is definitely correlated with the outcome variable. I’m not here to teach a stats lesson, though; it’s the offseason and I’m just here to fill the void with college football content.)
TCU comes up on a season in which they host Group A: Texas, Kansas, West Virginia, and Baylor. I’ll use the 2019 S&P+ Way-Too-Early Projections as the expected S&P+ margin of this group, and let’s see what the model says TCU’s ceiling should be in terms of wins.
TCU faces an average conference opponent at home with a 3. 525 S&P+ margin. That would be the lowest-quality conference slate in TCU’s Big 12 History, edging out 2015’s 3.7 and 2013’s 5.075 ( that 2013 team, you’ll remember, went 4-8 on the back of a couple of close games - they had 7 Second Order Wins and ranked 31st in S&P+).
The model projects Season Wins to equal 10.465 - .3019625*(HomeS&P+Avg.), which comes out to 9.4 wins. That seems pretty consistent with TCU’s ceiling this season, and facing an opponent with an average of 3.525 S&P+ would be the best case scenario. Let’s manipulate that a bit - for example, Baylor is rapidly improving and Texas is rising even faster. Let’s bump each of their margins up, see how it reflects the average and therefore the projected win total.
Adding 5 to both Texas and Baylor’s margins brings the average to 6.025, lowering the expected win total to 8.64 wins.
Adding 10 to Texas and 5 to Baylor brings the average to 7.275, lowering the expected win total to 8.26 wins.
Worst case scenario, let’s say Texas, Baylor, West Virginia and Kansas all play a their best level from the last five years. (Note: I’m just saying Kansas’s best case scenario is a 0 margin, because they’ve been negative forever.) The average is then: 12.1, lowering the expected win total to 6.8 wins.
The Big 12’s nine-game conference schedule presents a unique scheduling quirk, wherein each team flips between four and five conference home games each year. Since TCU’s membership in the Big 12 began, the Frogs have had clearly differential outcomes against their two home and away groups. On years when the Frogs get the lower-quality group at home, they buy themselves bowl and Big 12 standings insurance in the form of easy home wins. Much of TCU’s success in the Big 12 is tied to taking care of business at home against inferior opponents. Those consistent home wins have spillover effects in terms of team confidence and performance quality. This season, TCU hosts the historically lower quality group, and a simple model projects their ceiling at 9 wins. The quality of the groups, though, varies from year to year, and both Texas and Baylor are improving rapidly over the last few years. Adjusting potential quality outcomes for Texas and Baylor demonstrates that even slight variation in home opponent quality (+ or - 5 points), can lower TCU’s potential win ceiling by up to three wins.