Free Report: 10 Powerful Technical Chart Formations

How accurate is the NCAA preseason poll?

The dog days of summer are in the rear view mirror. The sun is meeting the horizon more swiftly with each passing day. And the kids are heading back to school. Of course, that includes the big kids too. I’m talking about the really big kids, like Michigan State’s starting offensive line, for example.

Yes, college football is right around the corner. The season kicks off on the Labor Day long weekend. There is a certain buzz heading into this year thanks to last season’s inaugural playoff which included an unexpected championship from The Ohio State University Buckeyes. And the momentum picked up some steam this week as the Associated Press (AP) released its highly anticipated and often debated preseason poll. For the first time, a program has been unanimously presented with the top spot. Not only must the Buckeyes defend their title, but they will be out to justify this historical preseason perch.

But how accurate is the AP preseason poll? I know the probability that the season-ending poll will match the preseason poll is inconceivable. However, rankings must wind up being a fairly good proxy for things to come. On average, can you not expect as many teams to exceed expectations as there are schools who fall short? Consistency is a vessel for greatness. Surely, certain coaches and/or programs will not deviate far from expectations. The following analysis takes a look at these considerations and evaluates how reliable the preseason projections are.


Before analyzing the accuracy of the AP preseason poll, I figured it was best to understand the current landscape. The following is a summary of the 2015 AP preseason poll released on August 23rd

2015 NCAA Football Associated Press Preseason Poll






Ohio State






Arizona State




Georgia Tech






Michigan State




















Florida State


Boise State


Notre Dame












For the 1989 college football season, the AP moved from the traditional Top 20 that our fathers grew up with and expanded to a Top 25 poll. Dating back to the first preseason Top 25 poll, I determined how many preseason ranked teams remained on the poll when the bowl games were all said and done. For example, since 1989, 19% of the teams ranked #25 in the preseason poll remained ranked when the season ending poll was released.


The eye gravitates to the top two rankings. 96% of teams ranked #1 or #2 in the preseason poll have still hung on to a poll position come January. Over the past 26 college football seasons, only the 2012 preseason #1 USC Trojans and the 1994 preseason #2 Notre Dame Fighting Irish failed to hold on.

The graph depicts potential for teams being over and underrated. Teams ranked #3-4, #9 or #18 show potential for being overrated – not a great sign for Alabama or Arkansas this season. Likewise, teams positioned #5-7, #14 or #17 have shown greater likelihood of staying ranked all season long. Could Auburn and Ole Miss be those teams this season?

The next graph is similar, but more focused on the accuracy of the preseason poll. Now I look at whether, at season’s end, the school remained within five positions (up or down) in comparison to the preseason poll. To illustrate, 62% of schools with a #5 preseason rank stayed in the Top 10 (i.e., either +/- five spots) in the season’s final AP poll.


Again, those top two rankings prove to be reliable in comparison to most other poll positions. This is especially noteworthy considering a #1 ranked team, for example, can only go down, thereby limiting its likelihood of remaining within the plus/minus five-spot cushion. Equally odd from a sheer statistical perspective, #13 ranked teams – the very center of the poll – have had such meagre consistency. Over the past ten seasons, only the 2013 Oklahoma State Cowboys remained within the margin after dropping to #17 following a loss to Missouri in the Cotton Bowl.


The charts above only depict the end result. But a story’s climax can be taken out of context without understanding the journey from beginning to end. For the chart below, I placed the preseason ranked programs into five groups: those ranked #1-5, #6-10, #11-15, etc. I followed each school’s progress throughout the season and charted the group’s average Top 25 ranking for each week.

There was one obstacle with this approach: How to include teams in the average if they become unranked. Obviously, adding them as a zero in the average wouldn’t work since it would lower the average when the goal would be the opposite effect. Instead, I assumed that each unranked college would be the equivalent of a program ranked #30.

I also needed to account for the fact that seasons have had varying durations over the past 26 years. Consequently, I lumped any results from the 14th week up to bowl games into a single category (see 14-16 below), thus avoiding any issues with small sample sizes.


The results were fairly consistent across the board. The average ranking within each group steadily decreases over the first half of the season and settles about 6-8 spots lower than the preseason poll. For example, schools ranked in the Top 5 in the AP preseason poll will finish the season around #10, on average. The quandary I faced with unranked teams starts to affect the results for the #21-25 group by flattening the line, but the analysis stays true with each group below this level.

Next, I looked closer at the group ranked #1-5 by splitting out the results for each of the top five preseason poll positions in the graph below. I had already established that preseason #3 and #4 ranked teams are more often overrated than a #5 school, but the chart below shows how they arrive at that conclusion.


#5 ranked teams tend to hit a sharp decline early into the season, but then settle around #10, on average. In contrast, #3 and #4 ranked teams will stumble further into the season eventually averaging approximately ten positions lower than expected. The progression of the two preseason favorites is far more gradual with both teams staying within the Top 5 for much of the season.


My focus continues to narrow as I examine the most recent history of the preseason #1. Can the progression of these teams give any insights as to how Ohio State will fare this season? The chart below looks back at the previous five preseason top picks.

2014 – Florida State
2013 – Alabama
2012 – USC
2011 – Oklahoma
2010 – Alabama


The five teams shown in the chart above can be categorized into three groups. The previous two preseason favorites – the 2014 Seminoles and the 2013 Crimson Tide – remained highly ranked and unbeaten until suffering a loss in either a bowl game or their conference championship. The 2011 Sooners and 2010 Crimson Tide squads are also similar in that both teams failed to live up to expectations but had respectable three-loss seasons otherwise. The 2012 Trojans are the exception to the rule. Lane Kiffin and USC limped to a 7-6 mark, becoming the first preseason #1 to finish the season unranked in 48 years.


For my final chart, I have analyzed the accuracy of preseason rankings for each college. To keep the results relevant, I only considered data from the past ten football seasons. The chart below assesses the average swing in AP rankings from preseason to season-end over the past ten years. I have charted the five teams (and ties) that averaged the most favorable swing in results and the five teams whose ranking shifted the furthest in the other direction. Note that only schools appearing in at least five of the past ten preseason polls were included in the results.

Teams with an average descent down the polls are shown with a lighter shade. Those who climbed up further, on average, than preseason expectations are darker. As an example, Nebraska has seen an average 8-spot drop from their preseason ranking since 2005. Of all the college football programs, only Oregon and Alabama saw an average uptick from their preseason rankings. Of course, Bama’s three national championships helped their results, but so does a swing in 2008 from a preseason #24 all the way to #6. Over the same period, the Ducks reached the Top 5 on four occasions and the Top 10 six times.


On the other hand, consistently failing to meet expectations could be a sign that a program is on the decline. Three of the schools on the right side (Nebraska, Texas A&M and Michigan) failed to make the 2015 AP preseason poll and another, Oklahoma, has its poorest showing (#19) in 15 seasons.

But what about the top two ranked squads this preseason? How will Ohio State and TCU hold up? The aggregate figures shown earlier were favorable. And, according to the chart above, each program should continue to track closely to expectations based on recent historical results. Although, can the historical data be trusted when we are dealing with the unprecedented. After all, the first kick-off hasn’t even taken place. The AP has never unanimously ranked a school #1 during the preseason. And they have never ranked the Horned Frogs this high.

The experts anticipate that this will be a college football season for the ages. I can’t wait to see how accurate they are.

Bob Sullivan writes periodically for and can be followed on Twitter at @mrbobsullivan.

Latest Articles