About two weeks ago we volunteered to take questions from the readers at Freakonomics.com. The request for questions generated 60 responses. Our responses – to ten of these questions – have now been posted. So please check these out and let us know what you think (and we want to thank Freakonomics for the opportunity to interact with their audience).
Again, there were 60 responses and we obviously did not respond to most of these. This is primarily because Stephen Dubner told us to focus on just a few of these questions (in fact he seemed somewhat worried that we might try and answer all of these). Our answers run for about 3,000 words, so even though we only focused on a few, we wrote quite a bit.
Despite the volume of our writing, though, we thought there were still a few more questions we could answer. So below are answers to five more questions from the Freakonomics readers (and some readers from here as well).
My question has to do with the BCS and college football: Is it in the best interest of colleges to stick with the BCS system financially? How much would they lose from switching to a playoff system? I don’t know if you specialize in professional sports, but I’m just curious. — Erika
Can you estimate the financial impact of college football adopting a 32 team play-off? In college basketball the universities love the tournament so much they’re expanding it to 96 teams yet the same college presidents are scared of the financial impact of a football tournament – seems like the NCAA is speaking from both sides of their mouth. — Steve White
These two questions are related. The NCAA and its member institutions generate most of its sports revenue from men’s basketball and football. The outcome of each season, though, is different. In basketball, the champion is determined by a tournament. This tournament generates a tremendous amount of revenue, funds that are distributed to each of the members in the NCAA. In football, the last game of the season – for the good and average teams – is a bowl games. Proceeds from these games are distributed to the participants and to the respective conferences.
From our understanding, if the NCAA adopted a playoff system for football – similar to what we see in basketball – revenues to the NCAA would increase. But the revenues going to the top teams and the top conferences might actually decrease. This is because we think revenues from a playoff system would have to go to all member schools. Consequently, the top colleges are opposed to a playoff system.
One should add that the coaches are probably not that enthusiastic about a playoff system either. Right now there are more than 30 head coaches who get to finish their season with a win in a bowl game. If you go to a playoff system, only one head coach gets to be a bowl game champion. And the other coaches get to answer questions about why their team isn’t as successful as their fan base desires.
Given these pressures, we suspect a playoff system for the top division of college football is not going to happen.
As for expanding the NCAA men’s basketball tournament… that seems more likely (although only a minor expansion is occurring this year). Such a move would increase revenues for the NCAA and increase job security. So on both fronts, the idea is a winner.
By the way, if you are looking for a good book on the economics of college sports, we would recommend “The Economics of Intercollegiate Sports” by Randy Grant, John Leadley, and Zenon Zygmont.
If there were one player action that you would want to be tracked and added to the NBA box score, what would it be? — brgulker
Your work currently does not include charges taken by defensive players. Hoopdata.com now tracks charges taken and Tom Haberstroh has argued (rather convincingly) that taking a charge should be considered the same as stealing the ball because it forces a change of possession. Are you planning on re-running your regressions to find the value of box score stats with the addition of charges drawn? It seems like it would further increase the accuracy of the WP48 measure and better account for some aspects of individual defense, which the current system doesn’t totally account for (because of the limitations of the box score). Thanks! — Kevin
These two questions go together. The answer for brgulker is charges. Tom Haberstroh published a great article at ESPN on the importance of the charge in the NBA (insider access required). This article clearly explains why we should track how often a player takes a charge. And as Kevin notes, this data is available at Hoopdata.com (but it is not yet part of the standard box score). To incorporate the data, though, one doesn’t need to run a new regression.
As Stumbling on Wins notes, the basic regression employed involves regressing wins upon offensive and defensive efficiency. From this regression we can see the value – in terms of wins — of points, field goal attempts, free three attempts, offensive rebounds, defensive rebounds, steals, turnovers, and personal fouls. This model also provides values of the opponent’s turnovers that are not steals. Part the opponent’s turnovers would include taking a charge, so this impact is already measured. All one would need to do is download the data.
As for the other elements of defense, we also know – from the aforementioned regression – the value of an opponent’s scoring as well. All we need is some method to assign this impact to individual players. So far, though, the measures provided are not very stable over time. This suggests that defense in the NBA is primarily a team activity. Consequently, defense is best evaluated at the team level. One should note that most of the box score statistics are quite stable over time. This suggests it is a good idea to credit these statistics to individuals.
Why haven’t the Cubs won a World Series in over 100 years? Do they keep making the same mistakes every year? — Doug
Josh Peter – of Yahoo! Sports – wrote an article about the sale of the Chicago Cubs last fall. In this article Peter addressed the Cubs reputation as “lovable losers.” As the following quote emphasizes, this is really brilliant marketing.
“It is pure genius from a marketing perspective, that we support a team that is bad,” said David Berri, a professor at Southern Utah University who specializes in sports economics. “It’s the only entertainment where, ‘We have failed, but we would still like you to pay us.’
“Imagine if you had a restaurant like that. We serve crappy food! That’s what’s so great about us!”
As noted last fall, this is part of the problem in Chicago. Stefan Szymanski – author of Soccernomics (and a number of other books on sports and economics) – once estimated the link between wins and attendance for each team in Major League Baseball (in a paper presented at the Western Economic Association). His results indicated that the fans of the Cubs are one of the least responsive to an additional win.
So we would argue that one problem with the Cubs are the fans of the team. The team knows the fans will show up whether the team wins or loses. Since losing requires less money and effort, the team tends to be less successful.
One should add, a similar problem exists for the Detroit Lions. The support for this team only started to falter when the team lost every game in one season. Such loyalty is considered admirable by some. But really, it just gives a team less incentive to try.
Your interesting inter-generational comparisons of NBA players reach back a little over three decades, a period for which you are able to use a full complement of variables to develop your models of player performance. Do you have any plans to develop models using fewer variables that would allow comparisons of players going back five decades or more? While the explanatory power of these more parsimonious models would almost certainly be lower, they it would allow for fascinating inter-generational comparisons — and would allow you to develop complete career wins estimates for players like Abdul-Jabbar, Julius Erving, and the like. — Brian Taylor
This is an interesting idea. The data presented in the standard box score goes back to 1977-78. We can get everything except turnovers for the individual player (turnovers were tracked for the team) back to 1973-74. So a few more years could be easily added, and those years would almost complete the careers of Abdul-Jabbar and Dr. J.
Before 1973-74, though, we don’t have steals, turnovers, or blocked shots. But one could probably construct a measure from what we do have and tell some stories. One would have to assume that the value of the statistics tracked is the same as what we observed from 1973-74 onwards.
As noted, we do have measures back to 1977-78. It might be a good idea to write another book detailing the stories this data tells. Perhaps something like “An Alternative History of the NBA.”
Love the blog Dave. Management make moves all the time with the implicit belief that they will be putting more fans in their seats and/or selling more tickets. These moves include player changes, coaching changes, management changes, adjusting game times, involving cheerleaders/dancers, promotions, and half-time entertainment. Other than winning more games or (for teams that sell out) building more seats, is there any other act that has been shown to sustainably bring more fans into the building? — TBall
Glad you like the blog. As for your question… the factors that we find determine gate revenue include team performance (wins and wins last season), star power on a team (measured with all-star votes), arena capacity, age of stadium, being an expansion team, having won a championship (or championships) in the past, and employing a player taken with one of the first two choices in the NBA draft (in the past year or two). Of these, wins, stadium capacity, and having a new stadium seem really important.
What is needed is a study of how all the other things a team does impact attendance. That would probably make for an interesting marketing study. From talking to people in sports, it is clear that they understand that winning is very important to their fan base. But perhaps the half-time shows generate some fan interest. It would be interesting to find out how much.
Final Thoughts
Even with these additional questions, we still have not responded to the majority of comments. Hopefully we can get to more of these soon.
Let me close by noting that none of these questions at Freakonomics apparently came from the APBRmetrics community. Members of that community, though, did comment on the Q&A. I will leave it to members of this community to think of an appropriate response.
The WoW Journal Comments Policy
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
Wages of Wins Journal 
brgulker
April 30, 2010
Thanks for the in-depth responses here and in the linked article. Makes for some great reading.
And thanks for addressing my question directly!!
Anon
May 1, 2010
As someone who doesn’t spend a great deal of time in online basketball communities, I don’t know why the APBRmetrics people dislike D. Berri. Can someone explain it to me?
Josh Smith
May 1, 2010
It’s largely because he refuses to engage with them on the apbrmetrics forum where most people with high level interest in this topic congregate, including the bulk of the people who work for teams. I understand Dr. Berri’s refusal to do so to some extent, the internet is usually not conducive to well reasoned discourse on complex topics, but generally the tone over there is fairly professional and the participants tend to be knowledgable and well educated. Dr. Berri would likely garner less criticism if he was less dismissive of anything not published in a peer reviewed economics journal. That said, I want to reiterate that I understand his issues with the general level of discourse on the internet, and his preferences for a peer refereed environment.
dberri
May 1, 2010
You are correct that I do not go to the Apbrmetrics forum to discuss topics. I am already here on the Internet. It seems silly to say that someone should dismiss an argument because it wasn’t left in the proper forum.
The Apbrmetrics people are free to leave comments here. In the past they have and we discovered their arguments were not quite as sound as the Apbrmetrics community apparently thinks.
I would add…. I have also already addressed much of what the Apbrmetrics community has said. And this was done in a peer reviewed journal and in Stumbling on Wins. My sense is that most members of this community have read none of this. I also suspect, few ever read The Wages of Wins.
SA
May 1, 2010
Great question about going farther back in NBA history. I was always wondering how Wilt Chamberlain would compare to other players, including modern players, in WP48.
At Basketball-Reference it says Chamberlain AVERAGED 50 points a game and 25.7 rebounds per game in the 1961-62 season. To average something that has probably never been achieved in a single game by probably 98 percent of NBA players is pretty amazing.
Tindall
May 1, 2010
SA – it is amazing, but it is less so when you factor in minutes played and pace. For instance, Chamberlain led the league with 18.6 rebounds in his final season. However, his total rebounding percentage, the percentage of available rebounds a player grabs while on the floor, was “only” 20%. To put that in perspective, Dwight Howard’s TRB% was 22% this season. While this slightly de-mystifies Chamberlain, it’s still clear that he was a dominant rebounder, and it’s also clear that Dwight Howard is the dominant rebounder and defender of his generation (and perhaps under-appreciated).
Although steals, blocks, and turnovers are necessary to calculate wp48, I would bet you a steak dinner that he would have a Wp48 in the .400 range several times during his career. His willingness to pass, his limited personal fouls, and his defense would propel him into that range. He had several seasons as a phenomenally efficient offensive player, albeit – bad word incoming – in a more low usage role.
As for an estimate of blocked shots from that era – http://sonicscentral.com/apbrmetrics/viewtopic.php?p=17149&sid=4108aad9ff888e20522dae8d38e33bb4
Fun to speculate.
SA
May 1, 2010
Tindall — All really good points. Related to that, it is also amazing that for a guy his size that Chamberlain played almost the whole game consistently during his career and seemed to avoid injuries for the most part. In his last season, at age 36, he averaged over 43 minutes a game. His scoring was down but his shooting percentage was .727. He still averaged 18 rebounds a game. I understand his career ended after a bit of a fiasco where he jumped to the ABA and the Lakers blocked him from playing due to his contract.
marparker
May 1, 2010
Prof,
Finally picked up the signed copy of stumbling on wins from my mail room. Thanks
reservoirgod
May 1, 2010
The APBRmetrics forum was so hostile to Dr. Berri that Justin Kubatko (from basketball-reference.com) called the forum members behavior “shameful.”
simon
May 1, 2010
Yeah the APBR thing is too bad as I used to read a lot of Kevin Pelton stuff back in his Hoopsworld days. Like many others I discovered sports stats through ESPN’s Neyer which led me to Bill James and later Moneyball, moving to basketball with Dean Oliver and then Kevin Pelton.
For some reason, I always chuckle whenever I see APBR people call dberri “arrogant.” In any case To see the APBR’s attitude on dberri, this is a comment from one of the members a few years back:
“He’s putting himself into a shell, trying to show he’s totally self convinced he’s right, rather than recognize he forgot something in the building process of the metric.
…”
And a current thread:
http://sonicscentral.com/apbrmetrics/viewtopic.php?t=2564
So “arrogance,” “Rodman,” and predicting lineup performance(urg, thanks Dr. Rosenbaum) have always been themes for years and will continue to be for years to come. Book it.
Alvy
May 2, 2010
I order Stumbling on Wins directly to my local Barnes n Noble.
Anyway, I don’t know the history between the APBR community and D. Berri’s work, but I can see how Wages of Win would really hurt all those +/- NBA-employed people.
shawn
May 2, 2010
I assume that by a “NCAA basketbal style tournament” you mean expanding the brackets to include teams that are clearly not among the best. In college football there is a playoff now – it is just that it involves only two teams.
While this discussion centers on the economics of expanding the college football bracket. It might also be ice to examine the effect on “fairness” as relates to the ability of a single elimination tournament to actually crown the “bes” team champion.
Josh Smith
May 2, 2010
Simon,
So a player performance metric SHOULDN’T be able to predict lineup performance? I’m not sure why that shouldn’t be a major issue.
simon
May 2, 2010
No, I just don’t think WP is a proper tool for that purpose. It’s a simple model with a decent year-to-year predictable power for individual players’ performance that can be summed back to the team efficiency in a given season, and that’s it’s purpose stated by its creators. I know Dr. Rosenbaum and APBR people have asserted that using the WP for predicting adjusted +/- and line up performances, and such is the “out of sample” test for it, but to me that seems testing a tool outside its intended purpose and complaining it’s a poor tool. If you can find a model that does a better job of predicting year-to-year performance of an individual player and link it clearly back to the team wins, that I’m all ears, but evaluating the WP model by using it to judge line up performance and other samples from specific game situations just do not sound right to me since they weren’t what the models were meant to solve.
marparker
May 2, 2010
Sour Grapes
Daniel Leroux
May 2, 2010
Hey Dave,
Do you have a way of counting how many players have scored over 50 in one game and over 25 boards in a game (either the same one or different games) in their career?
My guess is that it’s less than 20.
dberri
May 2, 2010
Daniel,
Couldn’t you find this out at Basketball-Reference? If they don’t know, I am not sure who would.
Westy
May 4, 2010
Unfortunately, from this bystander’s perspective, at this point the dialogue between APBR and Mr. Berri has regressed to the point of being unhelpful.
I agree with Justin Kubatko that their tone has sometimes been shameful, but I might also agree with some folks on their board and reviewers of the books in seeing how Dave could be seen as arrogant. I have yet to read Stumbling on Wins (waiting for library copy to become available), but certainly would suspect that members of APBR would not agree with the above statement that Dave has, “…already addressed much of what the Apbrmetrics community has said.”
My hope is that going forward the differences can continue to be researched and some progress be made. And I hope more APBR folks would engage here respectfully to clarify for Dave precisely the issues they continue to see.
dberri
May 4, 2010
Westy,
Since I addressed this in a published article (which I doubt you read) and in Stumbling on Wins (which you have not read), how can you disagree with what I am saying?
The APBR community has serious problems. The members simply do not understand basic regression analysi. Models are presented where they do not control for anything. They use residuals to evaluate models. They regress their models on other models (as if that would tell us anything). They fail to acknowledge the serious problems with APM and PER. I have seen models where standard errors are not presented or not interpreted correctly. All of these problems I have noted. My sense is that like you, none of what I have written has ever been read.
Essentially the APBR methodology is as follows: If you are in the group, no matter what you do, it is fine. If you are not in the group, no matter what you do, it must be wrong. The group consists of the “true” experts. The fact that these “experts” make mistakes my students can easily avoid is irrelevant.