Preseason Poll Accuracy

It’s preview magazine season in the college football world, and that means there’s predictions flying around all over the place. does a great job of keeping track of these predictions and seeing how accurate everyone is.

That accuracy is of the predictions of conference standings though, not accuracy of the preseason top 25 polls. I decided to see just how well everyone did in their rankings, since rankings, unlike standings, leave room for debate and subjectivity.

I decided to use the final, post-bowl AP poll as the benchmark to compare against. All polls have inherent flaws, but I think the AP poll is the best of them since every ballot is published for the public to see and the AP is not trying to sell magazines with its poll. Sure the media members have their regional biases, but they all get smoothed out in the end.

With that in mind, I used Stassen’s 2007 preseason consensus page to get some rankings from last season. I only used the magazines and websites I have heard of for my comparison, so not everyone on that page is represented below.

I looked at the average error between each team’s preseason rank on the polls and its position on the final AP poll. I used the “also receiving votes” category to make rankings #26-40, so anyone who ended up not even receiving a vote in the final AP poll got an automatic ranking of #41 when calculating error. I also took into account how many teams on the polls’ preseason top 25 lists ended up unranked in the final AP poll.

For best pick I tried to harmonize the maximizing of the difference between a poll’s rank and the preseason consensus and the minimizing of the difference between the poll’s rank and the final AP rank. That’s why no one who had LSU in the preseason top 3 got that put as the “best pick” since the consensus had LSU #2. For worst pick, I did the same only with the maximizing of the difference between the poll and the final AP rank. That’s why no one has UCLA for the “worst pick” despite nearly everyone having the Bruins in their mid-teens; the consensus had UCLA at #15.

I went ahead and threw my preseason top 25 on there since I wanted to see how I did. It was not that I hoped to brag or anything; after all I had TCU at #22 despite not knowing a single player on the team and Oregon State at #23 despite at the time not knowing who Sammie Stroughter is. It was a poll made of well informed ignorance, built from an SEC bias on the assumption that the ACC and Big 12 would be down.

Now, to the rankings. In the best/worst pick fields, the number in parenthesis is the place in the final AP poll the team ended up with. Also, I eliminated Michigan and Louisville from contention for worst pick as everyone got them very wrong.

2007 Preseason Poll Accuracy vs. Final AP Poll
Rank Poll Avg Err. Unranked Best Pick Worst Pick
1 Year2 9.36 9 BC #15 (10) USF #24 (35)
2 Coaches 11.12 11 Auburn #14 (15) Florida #3 (13)
3 St. & Sm. 11.24 9 Clemson #22 (21) Arkansas #12 (31)
4 AP 11.28 11 Oklahoma #8 (8) TCU #22 (NR)
5 Consensus 11.60 11 Tennessee #12 (12) California #15 (NR)
6 Blue Ribbon 11.72 10 BC #14 (10) Arizona #23 (NR)
T-7 Athlon 12.24 10 Oregon St. #22 (25) G. Tech #12 (NR)
T-7 Lindy’s 12.24 10 Florida #11 (13) Arkansas #14 (31)
T-7 CBS SL 12.24 10 Texas #8 (10) Arkansas #11 (31)
10 Phil Steele 12.44 10 Oregon #18 (23) S. Carolina #13 (NR)
11 TSN 12.92 11 Tennessee #18 (12) Florida St. #11 (NR)
12 14.28 11 Clemson #18 (21) G. Tech #10 (NR)

“St. & Sm.” is Street and Smith’s, CBS SL is CBS, TSN is the Sporting News, and is FoxSports’ and Pete Fiutak’s

Somehow well informed ignorance won out, which makes sense for such a crazy year in college football. The Coaches’ Poll ended up being more accurate than the AP’s own preseason poll, but I don’t know what to make of that.

It was a rough year for the normally accurate Phil Steele, and I have no idea what Pete at CFN was thinking when he made his poll. It wasn’t just GT at #10 that got him; it was also UCLA at #11, Miami (FL) at #12, Oregon State at #14, Oklahoma State at #19, and others.

While having nine preseason top 25 picks unranked at the end of the year was as good as anyone could do in 2007, the 2006 preseason consensus had just eight top 25 teams unranked at the end. Eight isn’t that much better than nine is though, and 2006 was not a chaotic year like 2007 was.

In the end, it just goes to show how inaccurate preseason polls really are. They’re an awful lot of fun to discuss, but no one really knows much of anything. The unpredictability is one of the best things about this sport.

When you see someone claiming to know all or a magazine touting its accuracy, take it with a grain of salt, nay – a salt mine, and gear up for another great season of college football.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: