The Dallas Mavericks won 67 games last year. Generally when teams win this many games the franchise gets an NBA title. The Mavericks, though, were bounced in the first round by the Golden State Warriors.
This year the Mavericks started out 9-2, which would put them on pace to win 67 games again. And then the Mavericks played the Pacers, Bucks, and Wizards – three teams that are not going to win an NBA title this year — and proceeded to lose three straight. Does such a streak suggest a problem in Dallas? Has the window closed on this team? Should Mark Cuban focus all his energy on buying the Cubs and winning a World Series?
Projecting Dallas
Before we get too excited about three games, let’s remember the wisdom of Dean Oliver. Dean makes the following observation in Basketball on Paper: There is actually a 90% chance that a 0.300 team will win three in a row sometime in a season.
See, even bad teams will win three in a row. And conversely, good teams can lose three in a row.
This is a point that needs to be emphasized. In sports we tend to focus on what just happened. But evaluating a player or a team requires that we take a look at the big picture. We don’t focus on one play, one stat, one game, or one week. No, to get a sense of how “good” or “bad” a player or team is we need to consider all that has happened over a period of time.
So let’s take a step back and look at what the Mavericks have done this season (yes, I know, this is still too small of a sample). As I have done each time I looked at a team this season, I am going to offer two projections. The first assumes the players on the team will perform as they did last year. The second will look at what will happen if the players on the team keep playing as they have this season. Each perspective is offered in Table One.
Table One: Projecting the Dallas Mavericks
If each player on the Mavericks maintained his per-minute performance from last year, this team would be on pace to win 55 games. Given what these players have done this year, this team is on pace to win 53 games. In sum, this team hasn’t really changed.
Luck and Dampier
Wait, that doesn’t make any sense. This team won 67 games last year. Clearly something changed.
To understand what’s different, we need to take another look at those 67 victories. When we look at efficiency differential (offensive efficiency – defensive efficiency), we see that the Mavericks had a mark of 7.8 last year. This translates into 61 wins, not 67 victories. In other words, the Mavericks 67 wins were a bit of a mirage last season. Yes, they were good, but not that good.
Of course, 61 still trumps 53 or 55. Where did the other wins go?
The key is the injury to Erick Dampier. In 2006-07 the Mavericks allocated 79% of its minutes at power forward and center to Dirk Nowitzki, DeSagana Diop, and Erick Dampier. This year, because Dampier has been hurt, the team has only given 72% of the minutes at the four and five spots to these three players. If these players each played as well as they did last year, and we simply return to the allocation of minutes we saw in 2006-07, the Mavericks would be projected to win 58 games this season.
In sum, there have basically been two changes in Dallas. Dampier got hurt and the team is now getting the number of wins from its efficiency differential that you would expect. So if you are a Mavericks fan, don’t panic. Your team is still good.
Can Dallas Finally Win a Title?
Is this team good enough, though, to win a title? That’s not quite as clear. When the season started I said the following teams would make the playoffs in the West: San Antonio, Phoenix, Dallas, New Orleans, Utah, Houston, Denver, and the LA Lakers. After a few weeks, here are the top eight teams in the West: San Antonio, Phoenix, Utah, Dallas, New Orleans, Denver, LA Lakers, and Houston. In essence, the Western Conference is falling right in line with what you would project before the season started (of course, don’t ask me about the East).
For one of these teams to wins a title it’s going to have to defeat three teams out West, and then defeat the Eastern champ (who will probably be Boston). For any team, that’s going to be difficult. We do know that one of the eight teams I mentioned will come out of the West. Dallas has a better chance than most, but probably not better odds than San Antonio or Phoenix. In sum, this could still be the Mavericks year. But just like last year, life would be so much easier if the Mavericks could move the city of Dallas to Rhode Island.
- DJ
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.
12 responses so far ↓
Rashad // November 28, 2007 at 9:28 am
So here’s a complete non sequitur that I’ve been wondering about for a while. Clearly talent evaluation in the NBA is flawed. But what if one of the reasons is that bad players can sometimes be REALLY productive, and those good games bias the observers. Take Iverson for example. The last 5-6 games he’s been atrocious, but before that he had this gem:
http://scores.espn.go.com/nba/boxscore?gameId=271112007
FG 14-20, 3PM 2-4, FTM 7-8, REB 3, AST 8, STL 2, TO 4
That’s a win score of 37+3+2+4-20-4-4=18 with an astounding .500 WS per minute across 36 minutes.
I bet if you did an analysis of individual games, overrated gunners like Durant and Iverson will have a smattering of amazing games when their shots are falling. And when they do, they look amazing. GMs think, “We want someone that can do that on our team!” The problem is that can does not mean will.
I think this is the crux of the talent vs. production problem. I mean, isn’t it tempting to think that with better coaching and improved shot selection Iverson, even at 32 could be an extremely productive player?
dberri // November 28, 2007 at 9:34 am
Rashad,
I just finished looking at Durant’s game log (probably as you wrote your comment). He has had four above average games this year. The list of four “good” games includes his last two. Clearly he has turned the corner!!! Okay, maybe not (then again, maybe he has).
I think you are correct. I think visual observation dominates talent observation in the NBA. If you believe a player is “good”, and you see him having a “good” game, your belief is confirmed. And you probably do think as you suggest. “If only he played like this every night, we would have the greatest player ever on our team.”
Mike H // November 28, 2007 at 10:45 am
Along those lines, it would be interesting to look at w48 game variance and range over a season and see how this correlates to player perception.
Owen // November 28, 2007 at 10:55 am
That’s an interesting idea Mike. What would you use as a measure of player perception?
People seem to make a distinction between how productive a player actually is, and how much “talent” he has. I struggle to understand what people mean when they use the word, but it seems to be something about how high a player’s ceiling is, how good he can be when playing at his very best. The fact that a player can put together a few scintillating games per year seems to count very heavily in their favor in the court of public opinion.
For instance, the fact that Jamal Crawford has had a few 50 point games is enough to convince a lot of people that he has a lot more “talent,” than the many guards who are more productive than him but will never have that kind of monster game.
Lior // November 28, 2007 at 12:23 pm
I’ve always thought that standard deviation is an important statistic for evaluating players. A player who plays well every game should have a different value than a player who plays an amazing game once a week, even if their average production per minute is the same. The difference depends on the needs of the team — if you only have one good player in a position, you need him to produce every night. If you have three, you’d rather have three high-variability players and give minutes to whoever is doing well that day.
As to the quantitative side, let’s take the boscore WS/min as a measure of how good the preformed that night. Since the season WS/min is the average of these number, weighted by the number of minutes played in each game, we should calculate the season standard deviation with the same weighting (adjusting for the fact that we are taking the sample std.dev. not a distribution std.dev.).
However, this standard deviation measures the wrong thing (persumably the player will get more minutes on nights he’s playing better). This leads me to propose using the uniformly weighted std.dev. over the 82-game season instead (assign WS/min zero for games not played). That value might tell how how “dependable” a player is.
A separate project (perhaps I should start taking up undergrads to do these things?) is the following: What is the correlation between WS/min and minutes played? [that is, are better players getting more playing time]
Lior // November 28, 2007 at 12:31 pm
Follow up on the correlation question: clearly you need to use the difference (WS/min – expected_WS/min) with the expected value adjusted for position.
MarkT // November 28, 2007 at 2:45 pm
Is anybody keeping a site that reports winscore for each game this season? The one that did that last season doesn’t seem to be repeating it this year.
dberri // November 28, 2007 at 3:07 pm
Mark T,
Hopefully something will be done soon. And no, I can’t define what soon means. People are thinking about this, though.
JChan // November 28, 2007 at 3:35 pm
Plans are in the works….from a few of us. For right now, you can usually see the new win score stats each day at
http://www.winsproduced.com/schedule.php
It’s hastily thrown together, but you can usually find what you are looking for. We’ve got some big ideas though, and hopefully some of that should be coming in the next few months.
Top Posts « WordPress.com // November 28, 2007 at 5:01 pm
[...] Are the Dallas Mavericks Doomed? The Dallas Mavericks won 67 games last year. Generally when teams win this many games the franchise gets an NBA […] [...]
Mike H // November 28, 2007 at 5:36 pm
Owen,
I wasn’t thinking of perception as anything concrete but all-star/this-or-that-award-of-the
-year consideration might be useful. Salary (after the rookie contract) could be used as well although it won’t be as consistently useful from year to year.
The Franchise // November 28, 2007 at 7:08 pm
Mike H. has a good idea–comparing the WP of the first half of a season with All-Star ballot returns would give a pretty good approximation of the relationship between public perception/popularity and actual production. (Of course, this method only works for players listed on the All-Star ballots.)