Across the past few years the Nuggets have had three problems building a championship contender. First, the Nuggets have put their money into unproductive scorers (i.e. Carmelo Anthony and Allen Iverson). Second, the players who do produce wins for this team have a history of injuries. Finally, and perhaps the strangest trend, is that the Nuggets have an odd habit of giving very good players to the 76ers. Consequently, this team has not quite performed as well as Denver fans may have liked.
Today I want to review the recent history of the Nuggets, a review that will highlight these tendencies. Then I will discuss where this team might be heading.
Reviewing 2006-07
After 23 games in 2006-07 the Nuggets record was 14-9. Such a pace would result in 49 victories over an 82 game season.
And then the Nuggets sent Andre Miller and Joe Smith (and draft picks) to the 76ers for Allen Iverson. Carmelo Anthony was suspended for 15 contests the game before this trade occurred. Consequently Iverson’s first 14 games in Denver were played a bit shorthanded and the Nuggets only went 6-8.
When Anthony returned to the line-up it was expected that Denver would take-off. After all, Denver now had two of the top scorers in the league on the roster. But this isn’t what happened. Over the next 34 games the Nuggets were only able to post a 15-19 record. Such a pace would only result in 35 wins over an 82 game season. Given that the 76ers clearly improved without Iverson, it was beginning to look like Denver made a mistake acquiring The Answer.
But over the last eleven games of the season Denver awakened. The team only lost one game, and defeated such giants as San Antonio and Dallas. Iverson fans suddenly felt vindicated. If the Nuggets could win 91% of its contests it would win 75 games in an 82 game season. Clearly having the two best scorers on one team meant you had the best team ever.
On further review the Nuggets success at the end of the season was a bit of a mirage. The Spurs did not play Tim Duncan, and when Duncan did arrive for the playoffs the Spurs dispatched Denver in five games. Looking at the Dallas game we see that Josh Howard – a player who finished second in Wins Produced on the Mavericks — only played seven minutes for Dallas. Additionally, the Nuggets wins over Minnesota and New Orleans were eased by Kevin Garnett not playing for the Timberwolves and Tyson Chandler not taking the court for the Hornets. Finally, Denver’s one loss was to the Grizzlies, the worst team in the NBA last season.
In sum, the eleven games at the end of the season probably didn’t paint a very realistic picture of this team. Perhaps a better picture can be painted if we look at the productivity of the players Denver employed last season.
Odd Trends in Denver
The productivity of the Nuggets across each of the last two seasons is noted in Table One.
Table One: The Denver Nuggets in 2006-07 and 2005-06
When we look at these past two seasons we see that the star is not on offense, but defense. The 2006-07 defensive player of the year – Marcus Camby – led Denver in Wins Produced each of the past two seasons. Actually, he has led the Nuggets in Wins Produced each of the past four seasons. In all, Camby has produced 58.2 wins – or 33% of the team’s Wins Produced – since 2003-04. One should note that prior to 2003-04, injuries had held Camby back. In seven seasons he had never played more than 63 games in a single season and averaged only 50 per year. But the last four seasons Camby has been fairly healthy. Yes, he has missed at least ten games each year. He has averaged, though, 66 games per year and consequently his Wins Production has been outstanding.
In 2005-06 Camby was joined at the top of Denver’s Wins Produced rankings by Andre Miller. Miller produced 33.1 wins for the Nuggets in his three full seasons in Denver. After he departed in the Iverson trade, second place in Wins Produced on this team was taken by Reggie Evans. And like Miller, Evans was shipped to Philadelphia for a player of lesser value (Stephen Hunter). Yes, apparently Billy King (general manager of the 76ers) has something on the brain trust in Denver.
The loss of Miller and Evans doesn’t leave the cupboard bear in Denver. Anthony and Iverson, although not as good as widely believed, are capable of being at least average players (average Wins Produced per 48 minutes is 0.100). And average, as is often noted here, is “not bad.”
Beyond the scorers, the team also has Nene Hilario and Kenyon Martin. Heading into last season, Hilario had only produced 11.1 wins in his career and posted a below average WP48 of 0.088. Like Camby, injuries appeared to be part of the story with Nene. Last year, though, Hilario took a leap forward and posted a WP48 of 0.158. He only averaged 26 minutes per game in 64 games, so his Wins Produced was just 5.6. Still, if healthy, Hilario looks like someone who can help.
Joining Hilario and Camby in the frontcourt is former number one pick overall, Kenyon Martin. In Martin’s last year in New Jersey he produced 8.7 wins and posted a 0.184 WP48. But this player has yet to show up in Denver. As a Nugget Martin has posted a 0.081 WP48 and only produced 6.6 wins across three seasons. And last year he missed all of 62 minutes due to injury. Still, if healthy, it’s possible that Martin can be quite good.
Reading along you might have noticed the themes mentioned at the onset. The Nuggets have a collection of good players. Often, though, these good players are either sent to Philadelphia or lost to injury. If Denver could stop sending its better player to the Sixers, and it could somehow keep these better players healthy and productive, the Nuggets might be a team that could surpass 50 wins in 2007-08. Injuries in the frontcourt, though, and this team -despite Melo and The Answer – might begin to disappoint their fans.
Changing Performance
Before closing, let’s ask this question – who is the smartest guy in the NBA? This is a hard question to answer. It might be Wayne Winston, creator of adjusted plus-minus and a consultant to the Mark Cuban and the Dallas Mavericks. But another candidate is Dean Oliver. Dean – author of Basketball on Paper — currently works with the Denver Nuggets.
Much of the work on NBA statistics tends to focus on how productive a player is (or has been). Wins Produced is a statistic that answers the “how” question. Although Oliver has chimed in on this issue, his focus tends to be on “why is the player productive?” And his work on this question threatens to wreck our ability to forecast in the NBA.
In looking at a team for 2007-08 I assume that what we saw in the past is what we will see in the future. For most players this is a pretty good assumption. Player performance in the NBA – unlike what we see in baseball and football – tends to be pretty stable. Dean, though, is trying to figure out how to change performance. And if successful his work might simultaneously help Denver win more games and ruin our ability to see the future in the NBA. Dean’s work is something to keep in mind as we watch the Nuggets this next season.
By the way, you will note I limited my candidates for “smartest guy” to two stats people. Although this obviously reflects a bias of mine, it also probably reflects the list of people I know who work in the NBA.
Okay, I know more than two people in the NBA. Really. I probably know about five people (it might even be six).
– DJ
For a discussion of other teams see NBA Team Reviews: 2006-07
Our research on the NBA was summarized HERE.
The equation connecting wins to offensive/defensive efficiency is given HERE
Wins Produced and Win Score are discussed in the following posts
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Wages of Wins Journal 
Rashad
October 25, 2007
I’m skeptical of Oliver’s ability to make players more productive. While scoring is clearly overvalued, coaches do know the importance of the other main inputs to win score. You hear coaches all the time talking about reducing turnovers and increasing rebounding. After games a coach will say, “well, when you get outrebounded by 10 you’re probably not going to win the game” or “well, when you have 8 more turnovers than the other team you probably won’t win.”
While the player incentives may favor scorers, coaches clearly have an interest in their team winning, and while they often overvalue scorers, they surely know that someone creating 3+ turnovers per game is not ideal.
Kevin Broom
October 25, 2007
Rashad: I did a simple study of what factors matter most in determining individual playing time last season. The data shows that the most important factor for 20 of the 30 teams was per minute scoring.
For 6 of the 10 teams where scoring was not #1, assists showed up as most important.
No team had rebounding show up as having the strongest correlation with playing time, but 6 teams had a stronger correlation between turnovers and playing time than they did between scoring and playing time.
Random notes on this: Charlotte’s strongest playing time to stat correlation was with steals; Indiana’s was blocked shots.
Only 4 teams had 2 stat categories that showed up as more important than scoring.
This is a measure of how coaches decide playing time. They may talk about lots of other stuff, but their tendency is still to give minutes to scorers.
Evan
October 25, 2007
Kevin —
That’s pretty fascinating for Charlotte and Indiana. Are there any possible explanations for this? Were the correlations lower than for the teams that had strongest correlations with scoring?
It seems your data could also give us a model for how an average bench player could shoot more to maximize his playing time, and thus maximize his pay.
Kevin Broom
October 26, 2007
Evan — I think that it’s possible the correlations for Charlotte and Indiana may be related to what those coaches valued. Rick Carlisle puts a big emphasis on defense. However, while this was the top correlation for Indiana, it was still quite weak, and it was really basically in a tie with scoring (.35 for blocks; .34 for scoring).
Charlotte’s steals correlation with playing time, I think reflects their defensive system. They placed a lot of emphasis on pressure defense and stealing the ball. Which doesn’t make much sense given the relative unimportance of steals to having a good defense.
Charlotte’s correlation between steals and playing time was .52, which would rank 24th if it was in the scoring column. It was the 3rd highest correlation in the league for steals behind Washington and Cleveland.
One thing I was curious to examine was the relationship between shooting from the floor and playing time. Curious because shooting from the floor is the most important factor — the team that shoots best from the floor wins almost 80% of the time in the NBA.
eFG showed up 4th in correlation with playing time, behind per minute scoring, per minute assists, and per minute turnovers. Several teams had negative correlations between eFG and minutes — in other words, guys who shot better played less.
For only 4 teams does eFG show up as more important than scoring — Phoenix (by a WIDE margin); Milwaukee, Chicago and Minnesota (just barely).
NOTE: I did not include eFG when I made my previous response to this article.
Overall, per minute scoring had a correlation with playing time that was 2.9 times higher than the correlation between eFG and playing time.
dberri
October 26, 2007
Kevin,
Could you send me a copy of your study? This sounds very interesting.
Kevin Broom
October 26, 2007
I emailed you the links to relevant blog entries. It’s not a formal study that might be submitted to an academic journal. I was just trying to see if the numbers would tell me anything about what earned playing for the Wizards. I later looked at the entire league.
dberri
October 26, 2007
Kevin,
Thanks, I got it. I will try and look this over soon. I am interested in doing a study like this in the near future and your stuff should help.
Brett
October 26, 2007
Kevin, could you either put a link here in the comments or email it to me as well.
Kevin Broom
October 29, 2007
This is basically what I sent to Dr. Berri:
What I did was not a formal study in the sense of publishing an academic paper. My guess is that Berri & company will find better ways of looking at this. Basically, the whole thing grew out of a desire to answer the question: What earns playing time with the Wizards? I ended up doing 3 blog entries on the subject —
http://kevinbroom.wordpress.com/2007/05/01/what-earns-playing-time-with-the-wizards/
http://kevinbroom.wordpress.com/2007/05/02/more-on-production-to-playing-time-correlations/
http://kevinbroom.wordpress.com/2007/05/04/even-more-on-the-relationships-between-per-minute-stats-and-playing-time/
As I point out in the final link, I wasn’t looking for (or expecting to see) a statistically significant relationship between a particular stat category and minutes. I was looking for tendencies or preferences.
In each stat category, I’m running the correlation between each player’s per minute statistics and his minutes per game. I used only players with at least 500 total minutes. The league averages I cited previously came from simply averaging the team-level correlations. I get slightly different numbers when using the actual player population, but the differences are very small and do not change any of the conclusions or points I’ve previously made.