clock menu more-arrow no yes mobile

Filed under:

Oklahoma Football: How you win matters

New, comments

The record looked fine, but the Sooners can’t keep playing like they did in 2021.


A recent tweet from ESPN’s Chris “The Bear” Fallica caught my eye:

This stat often gets portrayed as a harbinger of doom for a team in the upcoming season. (For the record, I definitely look at a team’s record in one-score games from the previous season when assessing teams that may underperform or beat expectations in the coming year.) For instance, college football data guru Phil Steele writes every year in his season preview about what a good or bad record in one-score games in the prior season portends for the coming one.

Even though the article linked above from Steele was published in 2015, let’s assume not much has changed since then. Given that OU netted out at plus-five wins in games decided by seven points or fewer in 2021, roughly 80% of teams in that situation post an equal or worse overall record the following year.

Why do teams’ records in one-score games have such a strong correlation with the following season’s overall record? You could come up with a number of explanations that, unfortunately, we can’t test. And, of course, “correlation does not equal causation” applies here.

But one implication of the relationship between performance in one-score games and a team’s record the following season is that it reflects good or bad fortune one year and a reversal in the next one. The concept of second-order wins developed by Bill Connelly of ESPN puts a finer point on that premise by projecting what a team’s record should have been versus what it actually was. Historically speaking, teams whose second-order wins deviate significantly from their actual win totals in one season are strong candidates for regression/progression the next year.

For an example that hits close to home, OU went 11-2 in 2013, which was 2.6 actual wins above its second-order wins total of 8.4. The 2014 team put together an 8-5 record in one of the worst seasons of Bob Stoops’ coaching tenure in Norman.

So let’s reframe last season for OU around second-order wins, rather than one-score games. According to Connelly’s advanced stats, OU had a second-order win total of 9.8 last season. This suggests OU won 1.2 more games than it should have in ‘21. The difference essentially came from the Sooners’ victories over West Virginia and Texas – Connelly’s postgame win expectancy for OU was close to a coin flip in both contests. In contrast, Connelly’s numbers indicate the Sooners got what they deserved out of their performances in losses to Baylor (win expectancy of 9%) and Oklahoma State (24%).

Unlike OU’s shiny record in one-score games, posting 1.2 more actual wins than second-order wins doesn’t really suggest that the Sooners are prime candidates for regression in 2022. So some of the signals around OU in ‘22 vis á vis last season are, well, kinda weird. You could definitely make a case that OU will take a step back this year based on the aforementioned stats, and that’s before you factor in changes in personnel and the coaching staff. (It’s also worth noting that a final record of 10-3 or 11-3 would qualify as worse than last year’s mark of 11-2, and I doubt many OU fans would view those as disastrous outcomes in the Sooners’ first season under new coach Brent Venables.)

But let’s look at this from one more angle: the simple fact that OU played in seven one-score games in ‘21. CFB stats maven Parker Fleming pointed out recently that playing tight games doesn’t really carry over between seasons:

Fleming offered three potential explanations behind that phenomenon:

If we were to apply those hypotheses to the ‘21 Sooners team, we’d be left with three different explanations for the abundance of close games:

  • OU was a good team that underperformed.
  • OU was a bad team that overperformed.
  • OU got super lucky.

I tend to lean towards option one. Perhaps that is overly optimistic on my part, but most of the games in which OU won by seven or fewer points really should have come down more definitively on the Sooners’ side of the scoreboard, according to Connelly’s stats. OU had postgame win expectancies of 84% or better against Tulane, Nebraska, Kansas State and Iowa State. Intuitively, that matched the vibe of the ‘21 season. It often felt as though bad decisions and a lack of discipline – stockpiling penalties, clumsy special teams, settling for field goals, et cetera – kept OU’s overmatched opponents within striking distance.

Even if that is the correct interpretation, it certainly doesn’t guarantee a better record for OU in ‘22. But I’d argue that how OU wins this fall might be the most important measuring stick for Venables in his first season. OU got away with lackluster performances in ‘21, but history tells us that teams with high aspirations can’t live that way. If Venables can get the team to play sharper football on a weekly basis this season, that likely bodes well for the health of the program going forward.