Most Frequently Overrated/Underrated ACC Teams
This week's Bourbon Shots included a link to an article from the GT blog called Margin of Hype vs. Margin of Excellence (From The Rumble Seat) that examined which teams tend to get overrated and underrated by the media following ACC Media Days. A good idea, but in the spirit of ACC academic rivalry I thought I'd post my own analysis, which unlike the GT blog is based in "theory" and "mathematical correctness" and other useful things that come along with a degree from the finest "Tech" in the ACC!
Let's start by examining shortfalls in the existing analysis by looking at Miami, which From the Rumble Seat claims is the second most ovverrated team in the conference. Presented as evidence is a graph showing that the range of preseason predicted finishes goes from 2nd to 5th, with the average actual finish being 3.4. Average is irrelevant here - if in four seasons they were picked preseason to be 2nd, 3rd, 4th, and fifth and those predictions were correct every time, then displaying an average finish of 3.5 is meaningless and does nothing to say whether or not they were underrated or overrated.
So what we are really most interested in is whether or not the difference between the predicted and actual finish each season tends to be around 0 (meaning there is no evidence of bias seen in the preseason picks for that team), positive (meaning there is evidence that the team tends to be underrated), or negative (meaning there is evidence the team tends to be overrated).
I calculated this difference for every team for the 2005-2009 seasons. Before jumping into who finished where, it's important to first do a statistical test to determine whether or not the differences between teams are real, or just randomness fooling us. It turns out there is ample evidence that one or more teams in our conference are rated differently than the others relative to their ultimate performance (for the geeks out there I ran a One-Way ANOVA in Minitab 16 with a resulting p-value of 0.032).
So who turned out to be perennially overrated and who turned out to be underrated?
At the "most overrated" end of the spectrum, we have Florida State in a runaway. They are ranked in the preseason, on average, about one and a half spots ahead of where the ultimately end up. At the "underrated" end? Boston College, who on average gets ranked about two spots below where the end up in their divisional standings. Although there appear to be differences in the others, keep in mind that the significant overlap in the bars on this picture between any two other teams means that we cannot statistically tell them apart (with a few more seasons we might be able to if a difference exists). To be fair to Georgia Tech, they have been underrated usually.
As for VT, we're sitting in a nice spot - not typically over- or underrated. Of course, when you're picked #1 most seasons it becomes difficult to be underrated...
FanPosts are user-generated content not necessarily endorsed by Gobbler Country.
11 comments
|
0 recs |
Do you like this story?
Comments
Please explain...
…how I was wrong? First, you proved that FSU and Miami were the most overrated and BC and Wake were the most underrated. Secondly, I always base my overrated/underrated statements on the difference between preseason predictions and postseason finishes. How else would you determine an underrated/overrated squad? I think you may have just misread the post and you didn’t read the links referred to in the post. I never referred to the preseason predictions as the sole indicator of whether or not a team was overrated.
I write stuff From the Rumble Seat.
Use statistical tests, not averages
Statistically you cannot differentiate between any two teams other than to say FSU is ovverrated compared to BC (or conversely, BC is more underrated than FSU). We don’t have enough data, or there is not enough of an effect, to say that Miami is the second most overrated team or to pass judgment on any teams other than FSU and BC.
Stating any other teams are over- or underrated based on a simple average is like flipping two quarters five times each and saying one is more likely to come up heads because three flips were heads on that one and only two on the other. Just because the average of a small sample came up that way doesn’t mean the effect is true. So I did not “prove” what you say I did.
Maybe after five more seasons we will have enough data to see that Miami is in fact overrated compared to most teams or that Wake is underrated, but right now you can’t say based on the data with alot of confidence that Miami is any more overrated than Wake.
Statistical certainty is uncertain
We are talking about semantics now.
FSU was picked to finish 1st, 1st, 1st, 3rd, and 1st from 2005-2009. They finished 1st, 5th, 4th, 2nd, and 3rd. I don’t need a statistical test to know that FSU is consistently overrated. Over-applying objectivity to the most subjective, unscientific football stat (a preseason poll run by a bunch of clowns in North Carolina) is dubious.
You also don’t want to use a statistical test over such a small sample size. The inaccuracy of the test would outweigh the benefits of using said test. And unfortunately this is generally the case in a lot of football stats.
I write stuff From the Rumble Seat.
Don't use data
First off, I have no idea what the phrase “Statistical certainty is uncertain” is intended to mean. I, or any other statistician, would never use the term “statistical certainty”. I think that’s a phrase people throw out to try to sound like they understand statistics.
There is nothing at all invalid about applying statistical principles to subjective information such as rankings, as long as it is done properly. In this case, the whole point of the analysis is to demonstrate whether or not there is bias in those subjective rankings and to put objective numbers behind that bias – I believe that was the point of your article. I would love to hear in what way that is dubious.
If statistical analysis cannot differentiate items because of a small sample size, then it is very misleading to then use that same data to back up a subjective opinion. Either you have statistical evidence of something and should present that evidence, or you do not and should then simply stick to presenting opinions and not pretend that data proves anything. I agree that many football stats lack sufficient data to be useful, especially when comparing two players or teams in a brief period of time – that’s what keeps opinion articles on the sport fun!
Georgia tech is the most over-rated b/c they are really bad in bowl games.
i agree wake forest is the most under-rated. They are the second smallest fbs university too. Smallest is rice universirty. I like WF’s coach too.
I'm all about covering the spread and moneylines. Glory favors the bold. Chance favors the prepared mind. Luck, well i have that too. University of Utah goes to the Pac-12 conference in 2011. I expect them to compete immediately for the conference CG. I still will always follow the Mountain West Conference. Brock Lesnar will defeat Cain Velasquez and face the winner of Junior Dos Santos vs Roy Nelson where he will defeat JDS and stake his claim as pound for pound champion. Womens MMA, the next big thing in sports. 1 month till the first game of college football. UTAH vs Pitt. September 2nd 2010.
by wolfmanshowlforever on Aug 5, 2010 4:47 AM EDT up reply actions
Over/Under rated
You can not determine whether a team is under/overrated using math because the claim that a team is either under/overrated is subjective in nature anyway. It is based on collective opinions from sports journalists, pollsters, and even tv coverage.
by Leonard Thompson on Aug 5, 2010 9:21 AM EDT reply actions
Ordinal Data
Since your dependent variables are ordinal and not nominal, I think, technically, you cannot perform ANOVA.
by Chazz Micheal Michealzz on Aug 5, 2010 10:33 PM EDT reply actions 1 recs
BURN!
(I have no idea what any of what he just said means.)
A bullhorn, a bottle of whiskey and a dream. GobblerCountry.com
by furrer4heisman on Aug 6, 2010 2:24 AM EDT up reply actions
Ouch!
It’s considered acceptable as long as specific intervals on the differences are not reported. Just a “yes this one is lower than that one”.
Is football starting yet?

by 













