Monday, July 10, 2017

Polling Differences: Addendum

Per request, this is an addendum to the most recent post regarding Power Five teams and the difference between their finishes in the postseason and preseason AP Poll since 2005. Previously, I used net difference between their postseason and preseason cumulative finishes, but a reader requested I use percentages instead. I have done just that. Obviously, larger numbers equal more improvement. Numbers less than one indicate a team has under-performed relative to their preseason positioning in the AP Poll and vice-versa.

In the previous posts, I began with the top and bottom ten and then went through each conference individually. I am going to reverse that formula here and begin with each individual conference and then finish with a caveated top ten. We’ll go alphabetically as always and begin with the ACC.
You can probably tell why I started with individual conferences. Thanks to sample size issues, the numbers can be a bit wonky. Wake Forest has exactly one ranked finish in the past twelve seasons, but thanks to continuously being ignored by the national media (and rightfully so), ranks second in the ACC in percentage increase. However, in my opinion, Clemson has the more impressive accomplishment. The Tigers have actually over-performed relative to their preseason prospects despite being burdened with pretty high expectations. And of course, if you paid attention in high school algebra, you know the other problem with using percentage increase instead of net – dividing by zero. Technically Duke and NC State have exceeded their preseason expectations by infinity (at least if you try to take it to the limit one more time), but I digress. Let’s just move along to the Big 12.
I don’t have a problem with Kansas State appearing at the top of this list as they always seem to defy preseason expectations. However, wonky issues with sample size also put their Sunflower State brethren near the top as well. Moving to the Big 10.
Here we have more wonkiness at the top with Northwestern grabbing the pole position. Penn State, Michigan State, and Wisconsin have been impressive in consistently out-performing their preseason expectations despite also consistently having some preseason expectations. Ohio State has basically broken even despite massive preseason expectations. And finally, Purdue, along with Virginia, is the only Power 5 team to not finish a season ranked despite at least one ranked start in the preseason poll. Here is the Pac-12.
Based on how many postseason points they have accumulated under Kyle Whittingham, I would rank Utah as the most overachieving Power 5 program. They are technically behind Boston College, but the Utes have nearly 40% more postseason points and the Eagles have not finished ranked since 2007. And finally, here is the SEC.
As I mentioned last week, Alabama has not only been super-elite since 2005, they have also exceeded their preseason expectations. That is very tough to do. The other over-achievers in the SEC are the perennially underrated teams in Starkville and Columbia respectively.

Since using percentages can make the data a bit wonky, I am going to list the top and bottom teams with a few caveats thrown in so that teams are compared to similar teams. It makes no sense to compare Alabama, a perennial top-ten team with Vanderbilt, a team that has not been in a single preseason poll in the time period examined here. So with that out of the way, here is the first grouping. The following table sorts elite teams by the percentage difference in their standing in the postseason and preseason AP Poll. Elite is defined here as having at least 100 preseason AP Poll Points. Those parameters yield fourteen teams and while some may not really be ‘elite’, attaining triple digits in preseason points seemed like a good arbitrary cutoff.
Not only is Alabama great, but for teams that enter each season with lofty expectations, they do the best job of exceeding them. Oregon and Clemson are the only other two elite teams to have exceeded expectations in this time span and Ohio State has basically broken even. Every team below Ohio State with the exception of Oklahoma has fired/forced to retire/exiled to the Phantom Zone at least one coach in the past twelve seasons. It seems with great power comes great responsibility.

This next grouping is what I would consider ‘good’ teams. These teams have accumulated at least 60, but less than 100 preseason points since 2005. This arbitrary cutoff yielded a nice round number of teams (ten).
For teams just outside the elite, TCU and Stanford have been the gold standard. The Horned Frogs and Cardinal have combined for four Power Five conference titles since 2012. The other two ‘good’ teams to exceed their preseason projections are Michigan State and Wisconsin. The Spartans and Badgers have combined to win or share six Big 10 titles since 2010 and at least one of them has appeared in every iteration of the Big 10 Championship Game. Look away Nebraska and Tennessee fans. Pretend it is still the 90’s and things are great.

Before we get to the top over and under-achievers regardless of preseason points, I wanted to list the five teams that accrued postseason points despite never appearing in the preseason poll during this time period. These teams have out-performed their expectations by infinity.
This is mostly a function of one good or decent year with no national expectations since 2005.

Now here are the top ten teams by percentage difference between their postseason and preseason point totals regardless of the total number of preseason points they received.
The top eight teams on this list are pretty similar. No team is a consistent national power, but they have each had their share of time in the spotlight over the past twelve seasons. However, the last two teams on this list seem to be different. Both Penn State and Washington can claim national relevancy (and titles) since the mid-80’s. Both teams finished in the top-ten last season and appear to be trending in the right direction under quality head coaches. If we revisit these numbers in five years or so, I think Penn State and Washington may find their way into the elite sphere.

And finally, here are the teams with the worst percentage difference between their postseason and preseason point totals.
Virginia and Purdue have not finished ranked despite receiving a little bit of love in the preseason polls, but for my money, Cal has been the biggest underachiever among Power 5 programs the past twelve years. The hippies from Berkeley have held on to just a quarter of their preseason poll points and have just two winning seasons since 2010.

When all is said and done, I think the addendum ended up being longer than the original post. Oh well. I hope you enjoyed it. In the next post, we’ll look at the most overrated teams (postseason poll) of the past twelve seasons.

Wednesday, July 05, 2017

Summer of Polls Part III: Polling Differences

In the first two posts on the AP Poll, I looked at the preseason poll and the end of season poll and calculated which teams accrued the most ‘points’ in each over the past twelve seasons. I awarded 25 points for teams ranked first, 24 for teams ranked second, and so on with the 25th ranked team receiving a solitary point. Now comes the next logical step: determining which teams have exceeded expectations the most and which teams have failed to live up to their lofty preseason projections. The methodology is quite simple. Just take the number of Postseason AP Poll Points (POAPPP) each current Power 5 team has accumulated and subtract their Preseason AP Poll Points (PAPPP) from that number. Positive numbers indicate teams that have over-performed and negative numbers indicate teams that have under-performed. So with the methodology out of the way, here are the top ten teams that have over-performed relative to their preseason ranking since 2005.
Congratulations Kyle Whittingham. Despite receiving almost no preseason love during his tenure in Salt Lake City (ranked 19th in the 2009 preseason poll), the Utes have racked up 49 points in the postseason AP poll over five ranked finishes. Missouri, with four conference title game appearances since 2005, is a close second and former mid-major TCU ranks third. It is interesting to note that the top three teams in this metric all upgraded to better conferences during this time period. You could have probably guessed most of the remaining teams on this list. For the most part, they are solid programs that are underrated nationally (Boston College, Kansas State, Wisconsin, and Mississippi State) or a non-traditional power that has exploded onto the national scene thanks to some great hires (Stanford). However, it is a bit surprising for Penn State and Alabama to appear here. The Nittany Lions have finished ranked in the top-ten four times in the past twelve seasons (2005, 2008, 2009, and 2016), but in two of those seasons (2005 and 2016) they were not ranked in the preseason poll at all and in another (2008), they were only ranked 22nd in the initial poll. For Alabama, this only serves to further Nick Saban’s legacy. As you may remember from a previous post, Alabama ranks fourth since 2005 in preseason AP Poll points. Yet, they have exceeded even that lofty standard with their final rankings. As we will see in a moment, most other preseason darlings have not been so fortunate. Speaking of, here are the top ten teams in failing to meet their preseason poll expectations since 2005.
This list features six teams (Oklahoma, Georgia, Florida State, Texas, LSU, and Southern Cal) that were in the top ten of total preseason AP Poll Points, with the recently retired Bob Stoops grabbing the pole position. That makes perfect sense. Teams that receive a large number of preseason points obviously have the most to lose. The teams that should really be shamed by this metric are Cal, Nebraska, and Tennessee. Those three all accrued 69 preseason points or less and still managed to make this list. For comparison’s sake, while Oklahoma lost 65 points from their preseason total, they still managed to hold on to more than 70% of their total points. Meanwhile, Nebraska and Tennessee only held on to about a third of their preseason points and Cal kept only a quarter of theirs. As with the past two posts, we will now examine the difference in POAPPP and PAPPP by conference. We’ll start with the ACC.
Outside of Boston College, no ACC team drastically exceeded their preseason expectations. And for Boston College, this is a function of three consecutive ranked finishes a decade ago. The Eagles have not finished in the final poll since 2007, but they have only appeared in a single preseason poll (2005). Clemson has done a good job of living up to their relatively lofty preseason expectations. Among teams with at least 100 preseason points, only the Tigers, along with Oregon and Alabama, have a positive differential. The rest of the conference is just not that interesting from an analytical perspective.

Now here are the Big 10 rankings.
Joining Penn State and Wisconsin as the only Big 10 team with a double-digit positive differential is Michigan State. If I had done this look back prior to their forgettable 2016 campaign, the Spartans would have ranked even higher. Michigan State began last season ranked twelfth and thus lost out on 14 points when they finished 3-9. While Ohio State does have a slight negative differential, they have pretty much finished in line with expectations on average, holding on to more than 98% of their preseason points.

Here are the Big 12 rankings.
I feel like West Virginia has a reputation for not meeting lofty expectations, but the numbers do not appear to support this claim. This undeserved reputation is most likely due to the 2008 and 2012 seasons when West Virginia began the year ranked eighth and eleventh respectively. The Mountaineers finished near the bottom of the poll in 2008 (23rd) and outside the poll in 2012, but in the other ten years, the West Virginia outpaced their initial standing in the poll.

Here are the Pac-12 rankings.
The Pac-12 owns the distinction of being the only conference to have five teams post a double-digit positive differential. Washington State is one of only six teams (Indiana, Iowa State, Kentucky, Minnesota, and Syracuse) to not appear in either the preseason or postseason AP Poll since 2005.

And finally, here are the SEC rankings.
Are SEC teams overrated by the preseason poll? Nine conference teams produced negative differentials between their preseason and postseason rankings. Traditional powers like Florida, Georgia, and LSU produced negative differentials thanks to being consistently ranked in the preseason poll. However, mediocre teams like Arkansas, Ole Miss, and South Carolina also produced negative differentials.

This concludes our look at differences between the preseason and postseason polls. In the next post, I’ll run a regression analysis between team record (Power 5 only) and final poll ranking. I’ll use the resulting formula to rank the most overrated teams of the past twelve seasons and determine which teams probably should have finished ranked. Until next time, thanks for reading.