No Gravatar

Over the past few years, the guys at 3 Shades of Blue have turned to Dr. David Berri from time to time to get some outside perspective on what we as fans feel about the Memphis Grizzlies. Dave, along with his co-authors Martin Schmidt and Stacey L. Brook, penned the statistical treatise Wages of Wins and Stumbling on Wins to try and objectively look at what the root causes of wins and losses on a basketball team. In doing so they discovered a disconnect between the contributions that help teams win games and the pay the players receive.

With the Grizzlies recent deep run in the playoffs, we turned to Dr. Berri more often to try to predict and explain why this run was happening. His insight was informative and interesting. So much so that we started looking into other works from Dr. Berri. In that investigation we came across a paper Dr. Berri wrote about the impact of coaches on wins in the NBA. As we have come to expect, Dr. Berri’s conclusions are not what one anticipates.

So, with the Grizzlies in the midst of a public relations storm about re-signing Lionel Hollins as head coach (and Dr. Berri on summer break from his job at Southern Utah University), we felt it was appropriate and timely to see what he had to say about what is going down here in Memphis. While discussing Hollins particular situation he also dropped a few cautious words about re-signing Tony Allen as well.

3SOB) In your paper, The Role of Managers in Team Performance, you make the claim that “the role of  the manager in promoting production is a little-understood phenomenon” and “it is difficult to separate manager’s contributions from the abilities of the workers they supervise.”  Are you saying anyone can manage great players to success?
Dave Berri: It is important to understand what we did, and what we didn’t do.

What we did:  It is argued that some coaches are able to elicit better performance from players. To test this hypothesis we examined 30 years of player performance data. We then looked at how coming to a new coach impacted a player’s performance (controlling for past performance, age, performance of teammates, and other factors that would impact player productivity). What we found is that of the 62 coaches investigated, only 14 were found to have a statistically significant and positive impact on player performance. So, most coaches did not have any statistically significant impact.

For those who are interested, here is the pdf to the published paper.

What we didn’t do: People have argued that what we are saying is that teams don’t need a coach. Our research doesn’t address whether or not a coach is necessary. And such a study really isn’t possible. Every NBA team has a coach. So, we can’t see what would happen if a team didn’t have a coach.

All we can say is that it appears most coaches are not able to elicit better productivity from their players. And that suggests most NBA coaches are similar.

3SOB) You base this claim on the work of noted economist Adam Smith. Just what does a 18th century Scot have to say about the Modern NBA team?
DB) We quote Adam Smith in our paper. Smith argued that day-to-day managers of firms are simply “principle clerks”. What this means is that Smith didn’t think that day-to-day managers could really alter the rates of returns for a firm. And this is because – as Smith argued – most managers faced with the same problems would make similar decisions.

We think this applies to coaches in sports. Coaches in sports have similar training and similar levels of information. Consequently, we suspect that most NBA coaches would be able to lead the Heat to one of the best records in the NBA. And most NBA coaches – if given the roster we see in Charlotte – would lose most of their games.

3SOB) Are you saying the General Manager has a larger impact on a team’s success than the Head Coach?
DB) We argue that players primarily determine outcomes in the NBA.  So the person who chooses the players (i.e. the General Manager) is the most important manager on a team.

3SOB) Memphis’ Front Office, which is reportedly adopting many of the ideas of statistical analysis, is debating the team’s coaching situation after their most successful season in history both in number of wins, percentage of wins and playoff advancement. Do you believe they are discussing this very issue of value versus results?
DB) I am not real sure how the Grizzlies are making their coaching decision.

3SOB) Lionel Hollins is credited with creating the culture of “Grit and Grind” that personifies the Grizzlies but he recently said he doesn’t even know what “Grit and Grind” means. Can any coach, given the correct type of players and a basic knowledge required, produce similar results? Does Lionel Hollins have a positive impact on the production of his players?
DB) Hollins was not part of our study (we looked at data from 1977-78 to 2007-08). One suspects – given our results – that most coaches would get similar results from the same set of players. More on this issue below…

3SOB) Hollins was an outspoken critic of the Rudy Gay trade and yet was the beneficiary of the team’s superior performance after the trade. Do you believe he did a great coaching job under difficult circumstances or did he just benefit from a more productive team despite his coaching?
DB) One wonders if Hollins is still opposed to the Gay trade. It was clear that the Grizzlies didn’t miss Gay’s production. His ability to launch shots at the basket is easy to replace in the NBA.  So it is not surprising that losing Gay didn’t hurt. But people tend to fixate on scoring totals. Consequently it is hard for people to see that Gay isn’t as valuable as his scoring totals would suggest.

As for Hollins coaching… it will take some time to update our study.  For now, let’s just do a simple exercise.

Imagine if we knew before the season started exactly who the Grizzlies would play this year and how many minutes each player would receive.  Given what the players did last year, we could then see how many wins the Grizzlies could have expected to receive from each player.  When we compare this to what we actually saw in 2012-13, we can see how player performance changed from 2011-12 to 2012-13.

The following table illustrates this exercise (this numbers are my own calculation, so they are slightly different from



Wins Produced




Wins Produced





Mike Conley






Marc Gasol






Zach Randolph






Tony Allen






Quincy Pondexter












Ed Davis






Tayshaun Prince






Wayne Ellington






Rudy Gay






Marreese Speights






Austin Daye






Jon Leuer






Chris Johnson






Keyon Dooling






Hamed Haddadi






Dexter Pittman






Tony Wroten*






Darrell Arthur**






Josh Selby










* – rookie, so numbers are the same for each season.

** — did not play in 2011-12, so numbers were taken from 2010-11

*** – team won 56 games. The team’s offensive and defensive efficiency in the regular season was more consistent with a team that would win 52 games.

On the left we see each player’s Wins Produced and Wins Produced per 48 minutes (WP48) for the 2012-13 season. An average player will post a WP48 of 0.100, so Memphis employed seven players who were above average this past season (and 13 players who were below average in 2012-13). Of these, five players – Mike Conley, Marc Gasol, Zach Randolph, Tony Allen, and Quincy Pondexter – produced 41.2 wins (or about 80% of the team’s wins).

When we look at these players in 2011-12 we see that this quintet – given last year’s performance – could have been expected to produce about 35.1 wins (one should note… this is a very naive projection since we would consider more than just last year’s performance if we wished to forecast this year). So these five players – more specifically, Conley and Gasol (and by himself, Gasol accounts for about half this difference) – account for much of the difference we see between our 2011-12 expectation and what we actually saw in 2012-13.

Is this difference coaching? This exercise doesn’t quite get at that, since we are not controlling for the other factors that cause player performance to change.  But it does show that for most players what we saw in the past is a good indicator of what we will see in the future (across all players we see about a 0.80 correlation between box score statistics from year-to-year in the NBA).  We could argue that Hollins “caused” Conley and Gasol to play better, but then we would have to ask…

  • … did Hollins also cause Gay to decline somewhat?
  • … Gasol posted a 0.185 WP48 in 2009-10, and then marks around 0.120 the next two season.  Is Hollins responsible for the declines the past two seasons in Gasol’s performance?  In other words, if Hollins got Gasol to produce more in 2012-13, why didn’t he stop the declines we see in 2010-11 and 2011-12?

Again, it might be the case that once we update the coaching model from 2007-08, we will see that having Hollins as a coach is associated with better player performance.  But I suspect – given what we see for most coaches – that the changes we see in performance can be either a) explained by the other factors in our model or b) simply random fluctuations (after all, players are not robots so we should expect some changes in production over time just because are examining human beings).

3SOB) We at 3 Shades of Blue have correlated what happened in Oakland in the MoneyBall days to what is happening in Memphis today and drew the correlation that Art Howe, despite a successful record as the manager of the A’s, was not resigned after a 103 win season and the A’s never suffered from his leaving. Can Grizzlies fans expect similar results if the team keeps their players in place yet replace their Coach?
DB) Yes. Again, I suspect that these players would perform in similar fashion for any coach. One issue you would have to consider, though, is that a key performer – Tony Allen – is getting old.  The decline we saw between 2011-12 and 2012-13 might continue. Age is the ultimate determinant of performance.  All players eventually decline with age (it is inevitable, although the speed of the process can vary from individual to individual).

3SOB) Do you believe it’s a mistake for the Grizzlies to not renew Hollin’s contract as a PR move? Do you believe it’s a mistake for the Grizzlies to not renew Hollin’s contract as a basketball move?
DB) Ultimately what matters for demand is whether a team wins or not. Since I do not think there is much evidence that Hollins actually changes player performance, I am not sure keeping him for PR reasons makes sense. And I think that answers the second question.

Again, it is possible that Hollins is one of those coaches who systematically alters player performance. But that seems unlikely.  Therefore, one would suspect another NBA coach – given this same roster – will lead the Grizzlies to similar results.

Consequently, the Grizzlies need to consider the costs of keeping Hollins to the cost of a different coach.  If the different coach is much cheaper, then the Grizzlies might be better off making a change.  If not, keep Hollins. Either way, I think what we see is that it is the players that will ultimately determine the future fate of the Grizzlies. Counting a coach to substantially change outcomes is probably not realistic.

Thanks again to Dr. Berri for his thoughtful and interesting thoughts about coaching in general and the Grizzlies coaching in particular.


Share →

5 Responses to What’s a Coach Worth?

  1. StevedNo Gravatar says:

    He forgot to mention a coaches greatest role, choosing who gets to play. In that, Hollins has done a reasonably good job overall giving the lion’s share of the minutes to the top performers (excepting his love affair with Gay). The FO did Hollin’s a great favour by removing his ability to give Gay so many minutes.

    • Red ColemanNo Gravatar says:

      I think that might be the biggest complaint amongst Hollins’ detractors, too — that the front office has to trade guys (or possibly let them walk in free agency?) so that he cannot play them anymore. Won’t play Tony Allen? Ok, say bye-bye to Xavier Henry and O.J. Mayo. Not playing Marc Gasol enough? So long, Marreese Speights and Hamed Haddadi. I’m not saying they’re right in that assessment — but it is worth exploring.

      • GrizzFan2005No Gravatar says:

        None of those guys were traded because Hollins was playing someone too much or not enough. They were traded because the front office thought we had other needs that needed to be addressed and someone else could do it better, or the money the player wanted was in line with what they felt his worth was.

  2. […] Click through to the article to see Dave’s simple exercise (or “very naive projection”, as he calls it). […]

  3. GrizzFan2005No Gravatar says:

    You can’t really judge a coach by this metric, because it takes the individuals out of the context of the whole. Someone may have what is perceived as decreased production, but it is allowing someone else to have increased production which is leading to better overall performance of the team. This is a team game, and it’s about how well the parts work together, that’s what coaching is about, the ability to making all these separate piece fit together, and move toward a common goal. How do you evaluate that other than looking at how the team is performing as a whole? I would say this is not one of the most talented teams in the nba but the team is greater than the sum of it’s parts. that doesn’t happen by accident.

    I use the analogy of building a house, you can have the best materials money can buy, but if you don’t use them correctly than the house is going to not serve it’s intended purpose very well, it may not even stay standing. Sometimes using less precious materials the proper way leads to a better structure. Would pure gold being a good material to make a support beam, definitely not. But the less valuable steel does the job. But generally the best house involves using the best materials, in the best way….That is coaching.

Leave a Reply