checkerboard03
Well-Known Member
- Joined
- Dec 28, 2008
- Messages
- 206
- Likes
- 0
Tl;dr version
1) TN will probably lose to Auburn, but beat Vandy and UK and be bowl eligible. Yeah, lots of people could tell you that, but...
2) Computer rankings will tell you UT is 40th or so. I'd argue that TN is actually closer to 25 or 30, but for one reason or another simply can't compete with the very best teams, and that makes us look worse that we really are.
Long-ish post, read as much or as little as you find interesting
So... how does TN beat SC and get killed by Missouri?
Could be that TN quit at Mizzou, could be home field, could be the QB.
Or maybe it's none of that. There's a pretty simple method that explains which games TN has won and lost, and by how much. We look at the final margin of TN's games, and the Massey computer ranking of their opponent. At first this does a bad job or predicting results. But if you ln-transform (more in a second) opponent's Massey rating the relationship is very strong. There's an attached graph showing this.
Basically, the fact that ln-transforming opponent rating makes the fit much stronger tells us that TN can't hang with the very best teams, the teams rated in the top 10 or 15 by the computers, but we should expect to be able to compete with anyone ranked around 25 or lower.
This is not good news for next week's game vs. Auburn, but does mean that we've got a really good shot of beating Vandy and UK and getting to a bowl game.
If you look at the graph, the exceptions make sense. In a few games we did worse that we should have (points above the diagonal line) USA? We really half-a$$ed that game and would do better if we played them again. UF? We'd do better if we didn't start Peterman. In a few games we did better (points below diagonal). WKU? Loads of turnovers. USC? Best game of the year for TN.
About computer rankings - lots of us, including me, look at these and call out things that look wrong. But honestly, after 9 games, these guys have a TON of data to work with and are generally really good at their jobs. Of course it's not 100% and you still have to play the games, but they are pretty good.
Here's hoping for at least 6-6 and a bowl game!
1) TN will probably lose to Auburn, but beat Vandy and UK and be bowl eligible. Yeah, lots of people could tell you that, but...
2) Computer rankings will tell you UT is 40th or so. I'd argue that TN is actually closer to 25 or 30, but for one reason or another simply can't compete with the very best teams, and that makes us look worse that we really are.
Long-ish post, read as much or as little as you find interesting
So... how does TN beat SC and get killed by Missouri?Could be that TN quit at Mizzou, could be home field, could be the QB.
Or maybe it's none of that. There's a pretty simple method that explains which games TN has won and lost, and by how much. We look at the final margin of TN's games, and the Massey computer ranking of their opponent. At first this does a bad job or predicting results. But if you ln-transform (more in a second) opponent's Massey rating the relationship is very strong. There's an attached graph showing this.
Basically, the fact that ln-transforming opponent rating makes the fit much stronger tells us that TN can't hang with the very best teams, the teams rated in the top 10 or 15 by the computers, but we should expect to be able to compete with anyone ranked around 25 or lower.
This is not good news for next week's game vs. Auburn, but does mean that we've got a really good shot of beating Vandy and UK and getting to a bowl game.
If you look at the graph, the exceptions make sense. In a few games we did worse that we should have (points above the diagonal line) USA? We really half-a$$ed that game and would do better if we played them again. UF? We'd do better if we didn't start Peterman. In a few games we did better (points below diagonal). WKU? Loads of turnovers. USC? Best game of the year for TN.
About computer rankings - lots of us, including me, look at these and call out things that look wrong. But honestly, after 9 games, these guys have a TON of data to work with and are generally really good at their jobs. Of course it's not 100% and you still have to play the games, but they are pretty good.
Here's hoping for at least 6-6 and a bowl game!
