Analysis – The Extra Point

SEC Football Fan Frenzy

December 17, 2018 No Comments

From the roar of the Crimson Tide to Vanderbilt’s sparse stadium… how can the fan experience change so greatly within the same conference? Is there a leading factor that contributes to these differences? In order to answer these questions, I turned to a platform that holds invaluable feedback and statistics on SEC fans…Twitter.

Twitter has become a place where raging fans can broadcast their opinion to an audience that may or may not be listening. There’s no barrier to creating an account and very few opinions that can go unsaid. When a quarterback fumbles the third down play, where can raging fans go to unleash their criticism? Twitter. Where can a lifelong fan go to express their support? Twitter. Through the highs and lows of SEC Football, Twitter is there to document those feelings. For that reason, this is the platform that I chose to analyze and compare how the fan experience changes among 10 SEC schools.

MY METHOD

14 teams make up the SEC. I decided to focus on 10 of those, removing the four schools with the lowest enrollment size. The figure below shows the enrollment size of the remaining 10.

I studied 50 of the most recent tweets mentioning the team’s name. For example, for the Tennessee Volunteers, I used the keywords “Tennessee Football” as my search query. For certain teams, these keywords brought up tweets that were not relevant to my analysis. I excluded those tweets that did not pertain to the team. For example, during the Tennessee Football search, I surpassed anything to do with the NFL Tennessee Titans.

I wanted to analyze tweets from the middle of the week and not on a game day. I chose Wednesday, November 28th because I did not want the feed to be full of statistics around one single game. Then, I organized the 50 most recent tweets into the following categories: fan photograph, criticism, support, news source and player/official accounts. An example tweet of each category is below.

Fan Photograph:

Criticism:

Support:

News Source:

Player/Official Account:

**********

After categorizing the tweets, I compared my findings to a study by Michael Lewis on college football fan base strength. In his study, he viewed college teams as a brand and ranked their strength in the college football market. His first metric was based off of revenue premium. He controlled for quality differences like team performance, alumni size and stadium capacity and argued that this metric showed the intensity of the fandom. Second, was the team’s return on investment. The stronger the team’s brand, the more benefits they would yield in the market, like recruits. The ROI measures the notion of brand efficiency. The last metric was football revenues or market share for each school. Lewis argued that this metric told us the most about the scale of each team. Lewis’ results of the 10 best football program brands is shown below.

I wanted to see if the tweets of the 10 SEC schools I analyzed followed the pattern and rankings of Lewis’ brand strength findings. Would Tennessee, #2 on Lewis’ list show the most support on Twitter? To analyze the correlation between my findings and Lewis’, I will break it down into categories. For the purpose of my survey, an unsatisfactory team will be those with a worse record than the prior year or one with a losing record. A satisfactory team will be one that has improved or one with a winning record. The categories are 1) Unsatisfactory teams with a top 10 college brand ranking 2) Satisfactory teams with a top 10 brand ranking 3) Satisfactory teams with a sub-10 brand ranking and 4) Largest program.

1) Unsatisfactory Teams with a Strong Brand

Lewis ranked Tennessee 2^nd and Auburn 9^th on the best collegiate football brands list. Despite being in the top 10 for this metric, Tennessee’s record is 5-7 and Auburn’s 7-5. The tweets about these teams made it clear that these were some of the most passionate fans in the league. In this case, passion did not always correlate to support. The criticism category exploded for both programs.

At first, it may seem contradictory that the best brands in college football got the most critique from fans. However, when a team’s brand is producing so much money and the team is still underperforming, of course these fans will be the most vocal. They support their team better than most other programs, so they become the most disappointed after a long season.

Tennessee Auburn

2) Satisfactory Teams with a Strong Brand

Now, let’s discuss the teams that were in the top 10 brands and also had a successful record. LSU (9-3), Georgia (11-2), and Florida (9-3) were among this group. Unsurprisingly, this group saw the most support and news coverage. All three schools ranked in the top 10 for revenue premium, ROI and market share. When they produce the results that the fans want, or at least have a decent record, the fans are there on Twitter showing their support. With each win, fans and news stations were there to tweet about it and spread the message.

LSU:

Georgia:

Florida:

3) Satisfactory Team with a Sub-10 brand

There’s one team that often stands out from the crowd; Alabama. In my findings, this trend was no different. Despite Alabama having a 13-0 record, they fell short of making Lewis’ best college football brands. Alabama’s tweets were particularly interesting. Of course, support was still relatively high. These fans are known to be vocal. However, the criticism category held different context. None of the critiques were about the actual performance of Alabama. They were more about the conference in general and how the system is flawed. For example, the tweet below with the hashtag “#CFBisrigged.”

This differed greatly from Tennessee and Auburn whose main critiques came from fans who were not pleased with their performance. For Alabama’s team, there was much more content on the specific players and coaches, almost like they were celebrities. This makes sense when the team is the nation’s best program in arguably the sport’s most competitive era.

Alabama:

4) Largest Program

And finally, Texas A&M. The team that has nearly 18,000 more students than the 2^nd largest SEC team, Florida. I wanted to find out if pure size of enrollment and alumni affect the fans Twitter spread? The categories for this team were much more even. Perhaps this is because the team is doing about as good as they have for the past few years. Nothing too much worse or better. There was a pretty even split between support and criticism. One trend I did notice was the difference in the accounts tweeting about Texas A&M. A majority of the tweets came from students and alumni themselves. At most of the other schools I saw a lot tweets from news sources and outside fans. Perhaps because Texas A&M has so many students, they have a larger presence on the platform.

MISSING VARIABLES

It’s important to identify the variables that were not accounted for in this study. Perhaps the reason the Tennessee Volunteers have such a strong brand is because they are the only major program in the state whereas Alabama has another major team, Auburn, less than 3 hours away. Or the history of a team and if this season is a disappointment. If a team has notoriously been successful and then has an underperforming season, their fans may be more vocal with their critiques on Twitter.

WRAP-UP

The biggest shock of Lewis’ study was perhaps the fact that the Alabama Crimson Tide did not make the top 10 college football brands list. In Lewis’ attempt to explain the omission of Alabama on this list, he poses the GREAT question of, “what would happen if Tennessee had a run like Alabama’s? Would the Volunteer fan base be as intense as the Crimson Tide?”

It is hard to answer or even muse questions like this. Mostly because the day that Tennessee has a program and record like Alabama is unforeseen. But, my results combined with Lewis’ analysis show that if Tennessee ever did have a run like Alabama, the magnitude of college football fandom would change forever.

Analysis

Drafting a “Winner”: Does the NBA overvalue NCAA champions in the draft?

December 15, 2018 No Comments

In the draft, every NBA front office is searching for that edge. While there are relatively tangible attributes that all teams look at to judge a player, such as shooting ability or athletic prowess, talent evaluators often try to look beyond the numbers for an indication that a particular player is destined for greatness. This usually comes in the form of “intangibles,” a player’s overall mindset. One rationale that has been pervasive throughout the NBA community is that X player is going to be great because “he’s just a winner.”

The intangible value of having “a winner’s mindset” is something often discussed when evaluating college prospects. The popular theory is that players who have won the NCAA title have proven through their experience that they can lead a team to success. Heading into the 2015 NBA draft, one of Jahlil Okafor’s perceived strengths was that “he’s proved to be a winner at every level,” which included winning the NCAA championship in his freshman year at Duke.

He and fellow champion Justise Winslow were both taken high in the draft, 3 and 10 respectively, but so far neither has quite panned out in the league. NCAA champion Jalen Brunson was another player lauded for his intangibles heading into the 2018 NBA draft. He was selected to the Mavericks 33rd overall, and in coach Rick Carlisle’s initial press conference on the selection, he made a point that “[Brunson’s] won the national championship two out of the last three years, he’s obviously a winner…the character is obviously there with what he’s been able to do winning two in the last three years.”

Jalen had many of the traits that would dissuade teams from drafting a player. He’s undersized, un-athletic, and old for a prospect, having spent three years at Villanova. Even still, according to Jalen’s former coach Pat Ambrose, “you can’t measure heart and that brain Jalen has. Wherever he has gone, he has been a winner.”

This all sounds nice, and it’s a good talking point to give to fans on draft night, but do NBA teams actually buy into this theory? More specifically, are prospects that have won an NCAA title viewed more favorably by NBA teams, and should they be?

To answer these questions, I used a dataset of all players drafted to the NBA between 1989 (when the draft went to two rounds) and 2007 (so that players who have only played a few years in the league are not included). I then created a dataset of all drafted players during this period who won an NCAA championship. To determine whether or not NBA teams are drafting players who won a championship in college too high, I had to compare the output of each championship player to the expected value of the draft position where they were selected. My calculations were made in career win shares¹ to best reflect the total added value of the player over the course of their career. I averaged all the differences between actual win shares of the championship player and expected win shares for the draft position where they were selected. The results were that, on average, championship players underperform their draft position by approximately two win shares.²

Because the career production of any given player in a draft class varies from year to year, we cannot know the true value of each draft position. Since drafted players who won an NCAA title is a subset of all drafted players, we have to account for variability in both samples.³ Taking this into account, we are 77% confident that college basketball players who won the NCAA championship actually perform worse than the average player at that draft position, with only a 23% chance that this difference is due to sampling variability.⁴ Therefore, it is more likely than not that NBA teams attribute too much value to the “winner’s mindset” when evaluating prospects.

To quantify roughly how much additional value is given to prospects who won the NCAA title, it’s helpful to determine how much an NBA team is overvaluing these players in terms of draft position. Since the calculations or player performance are done in career win shares, we have to estimate the average expected decline in career win shares between draft picks in order to translate the average difference in production from win shares to draft picks.

In a plot of average win shares per draft position, results show that there is a .73 decline in career win shares between each pick as you go further back in the draft.⁵ Since the results of our testing indicated that NCAA champions underperform their draft stock by two win shares, we can conclude that players who won a college championship are drafted somewhere between two and three picks too high.⁶

(-2.03 difference in win shares) / (-.73 decline in win shares per draft position) = 2.73 or ≈ 2-3 difference in expected draft position

So, why are NBA teams making the mistake of drafting championship winners too high? Why are NBA general managers incorrect to add additional value for act of winning an NCAA championship?

One player alone cannot win a championship. The ability of the rest of the team and the unpredictability of the NCAA tournament have an impact over a team’s chances of success. Still, this is not enough to explain why NBA teams are overvaluing NCAA champions. These guys are usually elite college players contributing heavily to their team. Thus, their performance would likely have a large impact over the outcome of the season. Therefore, if their impact helps the team win the NCAA title, their abilities should be looked upon positively.

This is a logical line of thought, however, when a player wins the title, their contributions are actually over-magnified because of double-counting. NBA general managers are counting abilities plus winning when evaluating these prospects. But to count winning separately is to double count that player’s value, because the amount that a player adds to winning is a result of their abilities on the court. These abilities are already being accounted for in the evaluation of the player, so winning should not receive any additional weight.

Basically, players who win NCAA championships are typically good players, but their talent should be evaluated based on the merits of their game, not the fact that they won the title. Championship players have been drafted 21st overall on average, which is well above the expected average draft position for all picks between 1989 and 2007 of 29.⁷ In reality, they should be getting drafted closer to 23 or 24 overall, which is still considerably above average. The abilities these players have, abilities that contributed to winning the championship, do make for a valuable prospect. However, the fact remains that these players are being overdrafted relative to their actual skill level, which suggests that NBA front offices are giving unwarranted value to the act of winning alone.

The value of winning continues to be a talking point in the basketball community. As a result, NBA teams often fail to evaluate these prospects correctly, which can end up in them selecting inferior talent. All front offices want that edge. Unfortunately, there is no secret formula for evaluating prospects. If there was, it most likely would have been found by now. Front offices need to stick to evaluating players based on the merit of their abilities. For those GM’s who look for added value in a “winning mindset,” they have a fundamentally flawed conception of talent evaluation.

Footnotes

¹ Win Shares is a stat that calculates a player’s individual contribution to total team wins. Full description in link

² Data taken from basketball-reference.com. Dataset and summary of statistics in link

³ Two sample t-test was used for statistical testing. Complete test in link

⁴ 23% is the probability of Type 1 error. Type 1 error occurs when a premise is incorrectly accepted. This is also known as a “false positive”

⁵ This is not a perfect estimation because the true decline in value of a draft pick appears to be nonlinear. However, a linear estimation of decline is still useful with the goal of roughly estimating a single value to quantify average decline per draft position.

⁶ Average of the all the differences between actual win shares of championship players and the expected win shares for their draft position divided by difference in win shares per draft position equals the difference in expected draft position and actual draft position. -2.03/-.73=2.73 or ≈2-3

⁷ The average draft position across all drafts from 1989-2007 is 29.07. This is because as expansion teams have entered the league the draft has expanded from 54 to 60 picks. The average draft position for championship players of 20.62 is still well below this average. Complete data in link

Analysis

Blind Sided: The Playbook for a Legendary Sports Movie

December 13, 2018 No Comments

When looking at the top ten grossing sports movies in recent history, you might ask exactly what propelled these feature films to the top of the box office. Was it because the true stories were just the inspiring tale that the world needed? Did the love stories capture the hearts of women while the sports element kept the men rapt? Or was it simply the presence of Adam Sandler and his bottomless production budget? From personal experience, I would say that the inspiring aspect of the true stories like Seabiscuit and Unbroken frequently moved me to tears, and is admittedly the main reason that Remember the Titans is at the top of my “Movies to Watch When Sad” queue. Others, like Rocky IV and Rocky III, I had never encountered before conducting this data set to analyze, and still find little to no connection with their plot besides the inevitable defeat of whoever Rocky decides to fight in that particular sequel. Slapstick comedies such as Adam Sandler’s Waterboy and The Longest Yard, or Will Ferrell’s famous Blades of Glory and Talladega Nights seem to rank in the top ten for other factors, quality of humor aside. So, what is it that connects these hall of famers?

While rewatching all of these movies with a critical eye, I focused on the narrative elements that incited a positive reaction, whether that be empathy or pure laughter. In this manner, I wanted to get to the heart of what landed these movies in the hearts of so many people around the world (see Appendix A). The greatest split within the ten movies comes with whether or not they are based on true stories. Elements such as the use of profanity, the classification of drama or comedy, and the time period the movie is set in had little to do with the popularity of the movie and everything to do with the fictionality of the story itself. Whereas the non-fiction drama films (Unbroken, The Blind Side, Seabiscuit, Remember the Titans, Rocky III and Rocky IV) used little profanity and were set in their appropriate time periods, the comedies (Waterboy, The Longest Yard, Blades of Glory, and Talladega Nights) were much looser with their f*&% bombs and were always set in a modern day America. By analyzing the ten movies as two separate groups, fictional comedy and non-fictional drama, it was clear that each had distinct elements that were necessary to drawing in the audience. As a whole, all ten movies had an element of love, an adult lead, and were rated PG-13 or lower. Each of these consistent elements appeals to a broad audience, which may have helped bring in maximum revenue. The element of love or romance can appeal to anyone, but none of these movies are strictly romance films, which simultaneously does not exclude anyone with particular apathy to such films. Having an adult lead does not exclude the interest of the younger demographic, but may make the prospect of viewing the film much more enticing to adults themselves. Lastly, the PG-13 or below rating does not limit the majority of movie-goers from seeing the move in the first place.

Stepping out of the audience and into the directors chair, I then analyzed the elements of the movies that had nothing to do with the plot (see Appendix B). Initially, I investigated individual production budget and studio, noting that the range went from Rocky III at $17 million to The Longest Yard at $82 million. Studios that produced these films include big names such as Warner Brothers, Paramount, Universal, Columbia, Buena Vista, and United Artists. Due to the wide range of money spent by a wide range of bigtime studios, I concluded that neither budget nor studio had a grand effect on the success of these box office hits. The conflict resolution is an element that is consistent throughout all ten movies and provided a sense of ending and a sense of satisfaction that came with a “happy” ending. Even if this conclusion did not result in the team or individual “winning”, it saw through a larger resolution such as reconciliation with family or loved ones, the overcoming of racial disparity, or a broader internal conflict. While I recognize that any feature film needs a notable studio to produce it with a sufficient budget, the variance of these factors signaled to me that there was not one consistent production team or amount of money that led to these movies’ place in sports film history.

On average, wide-released movies play in around 2,000 theaters. Among these ten sports films, each were playing in an average of 2,937 theaters world wide, almost 1,000 more than average. The least exposed movie was Rocky III, playing in only 1,317 theaters, while Will Ferrell in Talladega Nights raced across 3,807 theaters in 2006. Overall larger exposure plays a role in accessibility, a large factor in ticket sales. Average run-time for these movies was between 1 hour, 30 minutes and 2 hours, 17 minutes. This run-time gave each film a chance to develop a conflict and resolve it, generally within the last 10-20 minutes.

In addition to the direct external elements, I analyzed the overall economic status of America at the time of release of each movie. Using the NASDAQ composite as a gauge for economic growth, I plotted one year of historical data from 11 months prior to 1 month after release, as most movies stayed in theaters for an average of four weeks. By creating line graphs of the data, I noted that 9 of 10 graphs showed a positive sloping trend line, indicating a period of economic growth on the stock exchange. During these time periods, personal income rises and families have more money to spend on movie tickets which are, in economics, defined as normal goods: goods for which demand rises when income rises.

In regards to internal elements of sports movies, I concluded that the only real internal distinction lies between comedy and drama, and within each of those there is a clear script to follow in order to bust the guts or tug the heartstrings of viewers, respectively. Contrarily, the external elements seem to have little correlation to success, as most data points are erratic and not directly related to the eventual box office revenue. However, the external influence of economic health provides the most consistent trend in correlating the success of one of these predictable, albeit inspiring, flicks. Or, maybe it is just Adam Sandler.

Appendix A:

Appendix B:

1 2 3 … 5 Next »