What if you identify the disease, but there’s no cure?

… I wanted to see if I could find any connection between a team playing disproportionately well or poorly in the fourth quarter (as compared to the rest of the game) and their performance the next season. What I assumed I would find is a connection in which teams’ performances in the first three quarters carried over. That’s evidently not the case.

I grouped teams together based on how different their Q4 S&P+ was from their overall S&P+. The categories you see are as follows:

-20 or worse: Q4 S&P+ ranking is at least 21 spots worse than their overall S&P+. Obviously in Georgia’s case, they qualify both on defense (23 spots worse) and offense (69).
0 to -20: Q4 S&P+ Q4 ranking is 0 to 20 spots worse.
1 to 20: Q4 S&P+ Q4 ranking is 1-20 spots better.
21 or better: Q4 ranking is at least 21 spots better.

What I expected to find was that the “-21 or worse” crowd improved the next year. I found the opposite.  [Emphasis added.]

Damn you, Bill Connelly.


Filed under Georgia Football, Stats Geek!

14 responses to “What if you identify the disease, but there’s no cure?

  1. Hogbody Spradlin

    Connelly’s numbers don’t factor in starter pulling, which was meaningful for Georgia. Georgia had 3 or 4 games in which the game was over by end of 3rd qtr. Richt pulled starters en masse. That’s a significant percentage.

    Looking at the issue another way: in what games that were actually at issue, besides Auburn, did this really become a major issue. I know we lost games in the 4th, but we got in our share of licks.


    • Doesn’t every program that gets a big lead in a game pull its starters early? I would think that sort of washes out.


      • Hogbody Spradlin

        Conceded to a degree.I thought while writing my comment that some coaches are quicker to pull starters than others.


      • This would be my primary response, yeah. Honestly, if the reason behind this odd finding has to do with what it reveals about depth, then it’s probably saying that the teams that pull starters and fall apart might have reason for concern. I’m still a bit at a loss and trying to figure out what this means … it really wasn’t what I expected to see.


        • Mayor of Dawgtown

          It was all about conditioning last season. The starters were out of shape, particularly the O-line and D-line. They didn’t have gas in the tank for the 4th quarter. A bad off-season conditioning program was the main culprit but doing away with two-a-days didn’t help either.


    • Hogbody Spradlin

      To amplify I thought that last year, the ULaLa game and the three in October where we scored 40, were clearly over by late third or early fourth quarter, so the starter pulling was pretty distinct, and the offensive production clearly dropped. That many games, with benign reasons for the dropoff, might affect the conclusion from the stats. I have absolutely no explanation for defensive dropoff.


  2. The defensive numbers make sense. But, on offense, it seemed to me that they more often started out really slow and had to score in a flurry at the end to make it a game. Arkansas, comes to mind. MSU was the same way. South Carolina never really got going. I never felt that the offense fell apart late in games, save for the Colorado game. In the Auburn game, the scoring stopped (virtually) after the first quarter.


    • W Cobb Dawg

      Absolutely. In a season where we scored over 30 points in multiple games, much of what I recall were the struggles to score when it came down to a W or an L.


  3. Dog in Fla

    “What if you identify the disease, but there’s no cure?”

    Not sure but in other cases of terminal pathology, some have been known to just click their heels three times, say their work here is done and leave to spend more time with their family


  4. The real question is what percentage of teams that promote Joe T in the offseason rebound. We better how it’s 100 percent.


  5. Cojones

    You always have to do the math. Interesting result, but can’t find fault with the premise.

    I’m sure that the variables can never be sequestered in a game involving teenagers and hormones. Good example can be found in late win games with Martinez and Halanger- trained Dawgs. Ramble back a few years and place your statistics to account for year to year. You may get something out of it that’s worth a formula to apply to more recent games.

    This isn’t rocket science, but who knows where it could lead? Can’t hurt and will prevent too high of an expectation that would be unfair to our Dawgs and fans. That way you get one helluva lot more from an unexpected win.

    Anyway, it’s imaginative and not dusty old crap. I’m glad you did it. In science when you don’t get the answer you expected, the good news often is that the model was set up objectively.

    GO DAWGS!!