No Gravatar

For the last few years the writers at 3 Shades of Blue have had the privilege to interview a number of interesting people but no one has been as willing to talk to us as a Dave Berri, the noted professor of economics at Southern Utah University and noted author of two books – The Wages of Wins and Stumbling on Wins.

Dave Berri is considered one of the forerunners of statistical observation of NBA games. His insight into the league and who is contributing to team’s success has been controversial at times but generally accepted around the league. As support and interest into statistical analysis has grown (remember that Memphis hired John Hollinger for a major front office position mainly for his statistical analysis of the game) so has Dr. Berri’s stature on the subject. We are honored to be able to discuss with him his thoughts on the Grizzlies this season.

This interview was initiated after the Grizzlies completed their 62nd game of the season. I wanted to point that out because some of the numbers are dated by the time this is published but it is difficult to be timely and in-depth. Berri was very in-depth with his analysis and we do appreciate it.

1)      The Grizzlies have suffered a large number of injuries this season but as of Monday morning we are in the playoffs out West (and would be in 3rd place in the East). How is that possible?

My answer to this question is going to wander a bit.

Let’s start with the simple answer:  I agree that injuries really seem to be the story for the Grizzlies this season.

Now let’s go consider some details.  We start with the following numbers:

After 66 games the Grizzlies are scoring 102.81 points per 100 possessions and surrendering – again, per 100 possessions – 101.44 points.   This means the team’s efficiency differential (or offensive efficiency minus defensive efficiency) is 1.37.  This mark means the Grizzlies should have won about 36 of their first 66 games.

When we look at Wins Produced – which connects a team’s efficiency differential to what each player does on the court– we see that these 33 “wins” can be connected to the play of a number of different players.  Of the 13 players who have played at least 500 minutes for the Grizzlies, eight have posted above average Wins Produced per 48 minute numbers (WP48; average is 0.100).

The Memphis Grizzlies after 66 games in 2013-14

Team is compared to what we should expect, given each player’s performance in 2012-13

Grizzlies Minutes WP48
2012-13
Wins
Produced
2012-13
WP48
2013-14
Wins
Produced
2013-14
Difference
Mike Conley 1903 0.199 7.9 0.133 5.3 -2.6
Mike Miller 1361 0.177 5 0.162 4.6 -0.4
Tony Allen 1007 0.179 3.8 0.216 4.5 0.8
James Johnson 780 -0.054 -0.9 0.229 3.7 4.6
Zach Randolph 2148 0.158 7.1 0.069 3.1 -4
Courtney Lee 1025 0.142 3 0.138 2.9 -0.1
Kosta Koufos 1155 0.184 4.4 0.111 2.7 -1.8
Ed Davis 809 0.178 3 0.143 2.4 -0.6
Nick Calathes* 940 0.113 2.2 0.113 2.2 0
Marc Gasol 1413 0.172 5.1 0.064 1.9 -3.2
Tayshaun Prince 1641 0.075 2.6 0.043 1.5 -1.1
Jon Leuer 596 -0.02 -0.3 0.096 1.2 1.4
Beno Udrih 21 0.08 0 0.286 0.1 0.1
Quincy Pondexter 270 0.129 0.7 0.023 0.1 -0.6
Darius Morris 66 -0.012 0 0.029 0 0.1
Seth Curry* 4 0.014 0 0.014 0 0
Jamaal Franklin* 151 -0.088 -0.3 -0.088 -0.3 0
Jerryd Bayless 650 0.087 1.2 -0.03 -0.4 -1.6
SUMMATION 44.6 35.6 -9

* – rookie did not play in 2012-13.  So 2013-14 WP48 number is the same as 2012-13.

Numbers from 2013-14 taken from boxscoregeeks.com

Numbers from 2012-13 are my own calculation

This is actually somewhat unusual.

For example, consider the Grizzlies last season:

Memphis Grizzlies in 2012-13

Player WP48
2012-13
Wins
Produced
2012-13
Mike Conley 0.199 11.4
Marc Gasol 0.172 10
Zach Randolph 0.158 8.6
Tony Allen 0.179 7.9
Quincy Pondexter 0.129 3.3
Jerryd Bayless 0.087 3.2
Ed Davis 0.207 2.3
Tayshaun Prince 0.084 2.1
Wayne Ellington 0.094 1.3
Rudy Gay 0.03 1
Marreese Speights 0.043 0.5
Austin Daye 0.073 0.5
Jon Leuer 0.213 0.4
Chris Johnson 0.19 0.4
Keyon Dooling 0.046 0.1
Hamed Haddadi -0.013 0
Dexter Pittman -0.46 -0.2
Tony Wroten -0.038 -0.2
Darrell Arthur -0.018 -0.4
Josh Selby -0.297 -0.4
SUMMATION 51.9

Last year the Grizzlies were led by Conley, Gasol, Randolph, and Allen.  These four players produced 73% of the Grizzlies’ Wins Produced in 2012-13.  This is the more typical pattern we see in the NBA, where a few players produce most of the team’s wins.

The Grizzlies this season, though, have not had a major producer of wins.  In fact, Conley – who leads the Grizzlies this year – doesn’t rank in the top 50 in the NBA in Wins Produced.   Normally when the top player on your team isn’t among the best in the league, your season is a disaster.  For example, the top player on the Nets, Sixers, Magic, and Bucks also isn’t ranked in the top 50 in Wins Produced (and those teams are not exactly contending for a title).

Again, the Grizzlies are still in playoff contention because the team has so many above average players (a credit to the people picking these players!).  But the team has still declined since last year.  What explains this decline?

When we compare what the players did last year to what we see this year, we see – in the first table – that Memphis would be much better had certain players not declined.  Specifically, had the players for Memphis maintained what they did last year this team would have accumulated 45 Wins Produced.  Again, the team has not achieved this level so far.  And when we look at the individual players we see that declines in the production of three players – Conley, Randolph, and Gasol –explains the team’s drop-off.

Why have these players declined?  One obvious explanation is the aforementioned injuries.   Last year these players only missed eight games across the entire season.  This year this trio has already missed 35 games.  In addition, Tony Allen has missed 26 games.  Allen ranked 4th on this team in Wins Produced in 2012-13 and 3rd this season.   So again, injuries have been a problem.

2)       20 games remain and the Grizzlies are as healthy as they have been since the first month of the year, do you believe they will make the playoffs and can this team as it is playing now advance out of the first round if they do?

As the old joke goes: Predictions are difficult, especially about the future.

The data in basketball allows us to see how players impacted the team outcomes we have observed.  But the future in sports is difficult to predict.   For the most part –as the two tables illustrate – what you see is what you get.  That is a great feature of basketball.   Players in basketball– relative to what we see in other sports – tend be very consistent.

But what can we say when performance actually changes?  We know that performance is impacted by age. 
And it appears coaching doesn’t matter much  (although there are exceptions!).
But that is not the end of the list.  Clearly injuries also matter.  And that seems to be the story for the Grizzlies.

But how healthy are the Grizzlies?  If the aforementioned players are suddenly completely healthy, then yes, the prospects for the Grizzlies would seem likely to improve.  Of course, even if they make the playoffs, they likely will have to defeat one of the top teams in the West just to get out of the first round.  But a completely healthy Memphis team could pull that off (as I think they did last year).

But are the Grizzlies completely healthy? And how can we tell? Even if a player is on the court, nagging injuries can cause a player to do worse than what we saw in the past.

So to answer your question… since I don’t know exactly how healthy the players are, I am not sure if this team will be a force down the stretch.  I can say that if these players perform as they did in 2012-13 then the Grizzlies have a shot.

3)      Your system estimates the Grizzlies win total should be just short of 36 wins. Is having 39 at this point within the normal range of error or is something else explaining the 10% greater win total than expected?

Efficiency differential and Wins Produced explain about 95% of the variation in wins.  So the average margin of error tends to be between two to three wins across an 82 game season.  So this is normal.  Teams don’t carry over a performance in one game to the next.   But the model – which is using aggregate data – acts as if teams do this.  That is the source of the error.

4) If the majority of teams wins can be explained by the performance of just a few players, does Memphis having a number of higher producing players bode well heading forward or should the lack of a mega-producer spell doom for the team?

It is much easier to win in the NBA if you have at least one player who can produce in double figures.  Not having this player means you have to accumulate a number of above average players. In addition, you have to be quick to get rid of players who produce very low levels of wins. All that makes winning in the NBA difficult.

And if you want to contend for a title, then you have a huge problem.  To win 60 or so games you have to have an average WP48 of 0.150.  If you don’t have many players above that threshold, then getting to 60+ wins is almost impossible.

This means the Grizzlies want to contend for a title, they need someone who can play major minutes and post a WP48 beyond 0.200.  Mike Conley came close to this in 2012-13.  If he cannot do this in the future, though, the Grizzlies will need to add someone to the roster.


Players who produce wins in large quantities include Kevin Durant, LeBron James, Kevin Love (who the Grizzles traded for O.J. Mayo), and Chris Paul.
 But before Memphis fans despair, there are non-scorers who also produce win in large quantities.  That list includes names like DeAndre Jordan, Joakim Noah, Nicolas Batum, and Lance Stephonson.  In sum, the Grizzlies don’t need to win the draft lottery to acquire a major producer of wins.

5) Randolph and Prince have shown declines this season in their play, can the fans in Memphis expect that these veterans will improve their play down the stretch or should they continue to decline?  And Tony Allen is nearly as old as Prince and Randolph but his play improved this season despite injuries and losing his starting position. How do you explain this?

In looking at the past 40 years of NBA player performance, we can see that productivity peaks for an NBA player around their mid-20s.  By the time a player reaches 30, significant declines tend to set in for the average player.  But it is important to remember that this is what we see in general.  Specific players can age slower (or faster).

And again, performance can change for other reasons.  For example, I think attitude – which we can’t really measure – can make a difference.   An example would be what we saw with respect to Lamar Odom in recent years.  In the past, people have also argued Zach Randolph had attitude issues.  Under Hollins, though, he played quite well.  Is it possible that the departure of Hollins specifically impacted Randolph?  Again, we can see in the numbers that his performance has clearly declined.  But it is not clear in the numbers why specifically Randolph is offering less (except that his offensive rebounds are down and his turnovers are a bit up).

So to answer your question… we know how age impacts players in general.  And we know that someday age will result in each of these players being much less helpful to an NBA team.  But how age impacts a specific player isn’t something we can say.  So it is not surprising to see Prince and Randolph decline faster than Allen.

6) One last question. If you were the coach of the Grizzlies, who would you have in the starting lineup?

In recent years there has been some time spent by people on “line-up analysis”.  The problem with this analysis is that for most line-ups the number of minutes played together in a season is very small (less than 100 minutes, or less than about two games).  Furthermore, looking at an entire line-up ignores the basic question we have when we look at data in professional team sports.  We know which team won the game.  What we want to know is how each individual impacted that outcome.  In other words, we want to separate the player from his teammates.  Looking at entire line-ups, though, doesn’t allow us to see the individual.

Proponents of this idea seem to argue that player performance depends on the line-up.  But given the consistency we see in NBA player performance over time, it is hard to believe teammates make that much difference.  For example, Chris Paul has always been a very productive player even though he has played with many different teammates.  And Andrea Bargnani has always been a very unproductive player, even though he has also played with many different teammates.

The consistency of player performance suggests your best bet is to try and play your most productive players (that is the approach I took when ESPN the Magazine asked me this same question for every team before the season started).  For the Grizzlies, that suggests a line-up of

PG: Mike Conley

SG: Tony Allen

SF: Mike Miller or James Johnson

PF: Ed Davis

C: Kosta Koufos

Now if Zach Randolph and Marc Gasol can produce like they did last year, then
they would be better choices at power forward and center.  But if that is
not going to happen, then the Grizzlies are better off playing players who make
a bigger contribution.

Let me close by noting that deciding what line-up to play in the future
requires that we identify the most productive players going forward.  Our
best measure of productivity in the future is what a player has done this
season.  However, if what we are seeing this year is simply a function of
injury, and those injuries have been healed, then you need to dismiss what you
have seen so far in 2013-14 and consider what a player did before he was hurt.

Share →

4 Responses to Chatting with Dave Berri

  1. […] – of Three Shades of Blue – recently sent along some questions about the Memphis Grizzlies.  My answers were posted earlier this week.  The interview touches upon the Wins Produced of the Grizzlies, injuries, forecasting, and the […]

Leave a Reply