The Atlanta Hawks in 2005-06 scored 103.5 points and surrendered 108.5 points per possession. So their efficiency differential – offensive efficiency minus defensive efficiency (103.46-108.52) – was -5.1.
In 2006-07 this team scored, again per 100 possessions, 99.8 points while surrendering 104.9. This gives the team an efficiency differential of -5.1. In sum, although the team won four more games in 2006-07, the Hawks were essentially the same team each season.
Understanding the Past
When we look at Wins Produced we see the same story (this is of course no surprise, since Wins Produced simply relates what we know about efficiency differential and wins and applies this to the evaluation of players). As Table One indicates, the summation of Wins Produced for the Hawks in 2005-06 was 28.2. In 2006-07 it was 28.3.
Table One: The Atlanta Hawks in 2005-06 and 2006-07
In each season the Hawks were led in Wins Produced by Josh Childress. And in each season Childress was the only player to receive substantial minutes and post a WP48 [Wins Produced per 48 minutes] above 0.200. Again, average is 0.100. A team of average players will win 41 games. A team of players at the 0.200 mark would expect to double the win total of an average team, or win 82 games. In sum, Childress is really good. Surprisingly, though, Childress doesn’t start for this team.
After Childress this team has Josh Smith (a great small forward and an average power forward), Joe Johnson (slightly above average shooting guard), and Shelden Williams (a slightly above average power forward). And that was the extent of above average talent on this team in 2006-07. Given this collection, it’s not surprising that the Hawks have struggled to win 30 games each of the past two seasons.
Fans of this team know that the problems extend back quite a bit farther than just two seasons. Since 1999-00 the Hawks have averaged only 27.3 wins per season, with the team’s best campaign being a 35 win campaign in 2002-03. Prior to the last eight seasons, though, the Hawks were generally an above average team. From 1977-78 to 1998-99, Atlanta won 55% of its regular season games and only finished below 0.500 four times. So winning has a tradition in Atlanta.
And this leads us to ask, is it possible for Atlanta to return to its “glory” years?
Guessing About the Future
The only moves the Hawks have made this past summer is the drafting of Al Horford and Acie Law in the 2007 lottery. That’s really it. Atlanta, coming off eight consecutive losing seasons, is bringing back twelve players from a team that won 30 games.
This was essentially the same strategy the team followed last summer. The Hawks in 2006 added S. Williams, Speedy Claxton, and Lorenzen Wright to a team that won 26 games. And the results were predictable. With virtually the same roster this team produced virtually the same results.
Is there any reason to believe this strategy will produce different results in 2007-08? Surprisingly (at least, it was a surprise to me) there’s a small glimmer of hope. Horford actually posted slightly better numbers than Greg Oden in college last year (0.420 Win Score per-minute for Horford vs. 0.406 for Oden) so it’s possible that Horford can step in and help right away. Horford, though, does more than just offer the promise of future production. He plays in the frontcourt. And this means the Hawks are going to have to take minutes from someone else to get Horford on the floor.
How Horford gets his minutes might indeed be key for this team. If the addition of Horford pushes J. Smith and Marvin Williams (a below average player at every position and a disaster at power forward) out of the power forward position, then it’s possible the Hawks can improve a bit (let’s say 10 wins). If the Hawks continue to use J. Smith and M. Williams at power forward, though, any improvement will be somewhat less than 10 wins.
Of course, if J. Smith and M. Williams go to small forward, what happens to Childress? Ideally the Hawks will just sit M. Williams. Yes, M. Williams was the second player taken in the 2005 draft. And yes, he was taken ahead of both Chris Paul and Deron Williams (and no, Acie Law doesn’t project to be as good as either Paul or D. Williams – and yes, there are too many people named Williams in this post). But after two below average campaigns, it might be time to move on. The Hawks should devote the bulk of the shooting guard and small forward minutes to Childress, J. Smith, and Johnson. The minutes at power forward and center should go to Zaza Pachulia (a slightly below average center), S. Williams, and Horford.
These six players range from about average to quite good, and give the Hawks some hope going forward.
You will note, of course, that I have not mentioned the point guard spot. At that position the team has Speedy Claxton, Tyronn Lue, Anthony Johnson, and Law. Lue has never been above average, so he’s not likely to help much. And again, Law doesn’t project to be an above average player. Claxton apparently is healthy again, and he has been above average in his career. So it’s possible that Claxton could be productive.
So let’s put it all together. Can this team post a winning record? It’s possible, but it requires the Hawks be healthy, that they put their most productive players on the court, and put these players in the right position. So far all that hasn’t happened very consistently. Still, if it did happen, the same old Hawks might actually win a few more games.
- 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
19 responses so far ↓
Paulo // October 5, 2007 at 2:09 am
I don’t remember as clearly, but I read somewhere that Childress used to play some PG in Stanford. Joe Johnson also used to play some PG in PHX (and was originally signed by ATL with the intention of making him a PG). So with that, a lineup of Childress and JJ at G, Smith, Horford, and someone to play PF/C (Shelden Williams?) with Horford, I think would be a decent start, considering that they have a lot of interchangeable parts.
As for Marvin Williams, I do agree it’s time to pawn him off for a veteran. But I wouldn’t be shocked if he stays, and I’m hoping for an improvement. It was really a big disaster for them not getting CP3 (who at the time was the best PG prospect) or Deron Williams.
mrparker // October 5, 2007 at 7:54 am
I think you will be pleasantly surprised by what Law does. He played 4 years in the big 12 so his production should show up by January.
Chirstopher // October 5, 2007 at 10:31 am
David, I’d love to see you guess more about the future. How about guessing WL records for next year using your model? Hollinger has done this (as usual for this time of year) but I’d love to see how yours differ and use that as a discussion springboard. Hollinger’s stuff is here:
http://sports.espn.go.com/nba/news/story?page=HollingerForecastIndex0708
He has the Rockets with the league’s best record 61-21. I find that very hard to swallow.
Christopher
Chip // October 5, 2007 at 10:48 am
Was that a typo about adding Lorenzo Williams? I believe the Hawks added Lorenzen Wright to last year’s team after the C/F washed out in Memphis who was in desperate need of a center.
Not much of an addition in my opinion.
Jeff // October 5, 2007 at 11:12 am
David, thanks for the Hawks review! I think, as you stated, that using the right players at the right positions is the key for the Hawks. Trading Marvin Williams for a good point guard (Calderon?) either now or by the trade deadline would make this team a lock for the playoffs. Thanks again.
dberri // October 5, 2007 at 11:26 am
Chirstopher,
I thought of doing this, but picking an exact number for a team involves a fair amount of guessing. Just looking at the Hawks we see that if the team puts people in the right spot (and for some, that is on the bench) they could push 40 wins. But if they don’t, its back to around 30 wins. How are you supposed to know before the season starts how the coach is going to allocate minutes? Having read some of Hollinger’s forecast I would note (in addition to the obvious problem with PER) that he is often engaging in pure speculation. Part of his improvement for the Hawks depended on the idea that a team drafting in the lottery year after year has to get better eventually. Certainly that is true if you are picking good players. But if not, just being in the lottery year after year is not going to help.
Chip,
Thanks for catching the typo. I fixed it. The Hawks probably would have been better off with Lorenzo Williams. These typos are going to happen more often since I started posting these late at night rather than waiting until the morning.
Jeff,
Glad you liked the review. Is there any talk of trading Marvin Williams in Atlanta? For this to happen, Atlanta has to believe M. Williams isn’t that good. Not sure they believe that, but I am not paying that close attention to what is being said in Atlanta.
John // October 5, 2007 at 11:52 am
OK, I gotta say I think you are way off the mark on a couple of your comments. Joe Johnson is a “slightly above average” shooting guard? But he was selected for Team USA. And Marvin is a much better player than you give him credit for. He was hurt last year in training camp, and he was much more productive the last month or two of the season. I believe the Hawks are coming together and will be a much better team than you are predicting. I respect your opinion, just don’t agree with it.
Chirstopher // October 5, 2007 at 12:08 pm
Oh, I understand that it’s an exercise fraught with peril. But I think you could easily do it. Just take last year’s WP metrics for everyone on the roster and sum. Use carrier averages if injury makes last year’s numbers “bad” and use either average of last draft’s rookie class for this year’s or 80% of college stats (scaled to 82 games) for Durant and Co. I know it’s rough but I think seeing a table of such a WL prediction would be interesting and I think you could generate fairly quickly given the database you have. Who knows, maybe you’ll have Houston winning 61 games too
Jeff // October 5, 2007 at 12:37 pm
David,
I think Atlanta is still very high on Marvin Williams. They’re hoping he turns the corner and realizes his ‘potential’. If I can continue to request reviews; Christopher’s suggestion of Houston would be a good choice for your next review. Thanks.
John // October 5, 2007 at 1:31 pm
Atlanta is very high on Marvin. Why invest all of these resources into him, just to dump him off on some other team so he can have his break-through year (remember Boris Diaw)? They see this guy and, despite his average numbers the past two years, still honestly believe that he will become a star. Joe Johnson has carried this team and his numbers are outstanding, and he is an unselfish player. Not sure what more that anyone could ask of him. The reports from the Hawks training camp is that Acie will be hard to keep out of the lineup and that Horford is, to quote a couple of the Hawks players, a “beast.” Shelden is hurt, so it’s hard to say what his impact will be, and Josh Childress and Speedy have been playing very well in training camp by all accounts. In short, I think the Hawks are maturing, and by mid-season could be an above average team, maybe even good. That being said, I certainly agree that staying healthy will be key for the Hawks. Time will tell how this all plays out.
dberri // October 5, 2007 at 6:14 pm
Chirstopher,
I know it sounds easy, but actually what you suggest seems like quite a bit of effort. At least, as the person who would make the effort, this sounds like some work.
For each team I would have to assemble the players who are on the roster on that team this year, make projections for the rookies, and figure out minutes. And I think this process overstates wins because you are ignoring the end of the bench players who play a few minutes here and there. Plus there is no accounting for injury. One way to overcome this is to do the forecast for every team and then adjust the forecast downwards so that the sum of all forecasted wins equals 1230. Or some such approach.
All in all, sounds like quite a bit of work. We do have to remember that doing this stuff is Hollinger’s job. For me, it is just something fun to do to waste my time at night.
The Franchise // October 5, 2007 at 8:42 pm
The question I have is this:
What about Marvin Williams convinces Atlanta Hawks fans to be optimistic about him?
And John, don’t be too offended by Dave’s conclusions; they’re not his opinions, they’re the conclusions based on a model that has proven to be extremely accurate. As mentioned in a recent post, the model isn’t perfect at predicting the future–it’s usually more accurate than an “expert opinion”, but it’s not perfect, so a fan can still hold out hope that theirs will be a team that has unpredicted success.
mrparker // October 6, 2007 at 6:41 am
dberri,
When I do this I always set max minutes to 2400 for every starter. Then I add wins for “credible” bench players. My system is a little different from yours in that I don’t evaluate stats as representing wins. I just have them as representing good or bad. This would give a pretty safe assesment of most teams but leaves wins on the table. (i.e, when you add up every teams wins there are less than 1230 wins).
While this doesn’t give definite win/loss predictions it does result in having a clearer picture of which teams are better than others and by how much.
I am finished with the Eastern Conference predictions but have had some difficulties with the Western conference so I’m not predicting anything there yet.
Christopher // October 6, 2007 at 9:24 pm
Oh, it’s possible that I underestimate the effort but I think there are some shortcuts. I’m assuming you know who is on what team, i.e., offseason moves are incorporated in your database. Then, literally, just add up everyone’s wins from last year by team. You would also need to add up minutes and scale total wins to 82*5*40. That’s it, all players get the same treatment. Even use last year’s metrics for Yao and McGrady. In effect assume that they will be similarly injured this year. Use average rookie wins produced from last year too, one # for lottery picks vs another # for non-lottery. This is easier than my initial suggestion. Note: players that have left the L (retired or injured like Elton Brand and Sean May) do not contribute wins produced. You catch this roster void by the scaling trick. And I’d only worry about players where it’s common knowledge that they will miss the whole season, so there will be a little bit of manual flagging here. The hardest part is mapping players to teams but if you have last season in some database… Again, I’m not the one writing the script but I know the summation/scaling bit would be easy. It’s just a big pivot table. In any event, I do not want to beat a dead horse but predciting the future is, imho, the next logical step for your modeling framework. If you make it explicit that you are predicting this year’s team performance based soley on last season’s individual wins produced what is there to lose? To be honest, I’m surprised to have not seen a table along these lines in your book or here. You could batch process all years for which you have data in this manner and I’m sure there is an interesting story there.
Owen // October 6, 2007 at 11:03 pm
“I do not want to beat a dead horse but predicting the future is, imho, the next logical step for your modeling framework.”
Christopher – Horse dead, beaten, flogged, and reduced to glue. I don’t know if you were around when DB predicted the Sixers final record, or when that turned out to be correct, but I can assure you that doing what you suggest, beyond being a lot of work, is also a textbook no-win proposition. Instead of receiving encomiums for his predictive acume, DB ended up getting criticized roundly in many circles for achieving exactly what you suggest is “the next logical step,” (which i have to say has strangely sinister or Spockian overtones.) That was because his prediction was partly based on Chris Webber being there. How could he possibly know he would be cut?
Given the vagaries of fortune and the role chance plays in every NBA season, there really is no benefit is there to trying to pick one number. And it’s a major pr problem People hearing a specific number prediction will inevitably impute a level of accuracy and predictive prowess that DB would never claim. It’s dead certain to be misinterpreted as his thinking he is smarter than everyone else, that he thinks his system perfect, that it’s the final word in basketball analysis, that you can use it to beat vegas, etc etc…
Also, imho, this post is extremely clear. He doesn’t think they will be worse than the last two years. He thinks it very unlikely they will be above 500%. So that’s a range of 28 to 40. A lot depends on the play of Horford, the minutes given to Childress and Williams, the play of the guards, the durability and play of Smith, etc. Maybe 35 wins if things go reasonably well, 42 if they go perfectly, and 28 if things don’t work out well again, 36.5 – How does that sound?
Also the point of the blog is not the numbers, but the stories the numbers tell us. Isn’t it weird that Atlanta’s best player is on the bench? How good will Horford be? Does playing on Team USA mean you are actually an elite player? How do you have lottery picks for many years and not improve? etc….
Alright, enough…
dberri // October 6, 2007 at 11:55 pm
Owen,
Well said.
Nevertheless, I am going through the numbers and seeing what I can do. For the reasons Owen stated, I am reluctant to give a number for each team. But I think we can say something about relative quality.
For example: Celtics, Bulls, Pistons, or Cavs. Who is better in this group? I think there are differences but the forecasts I have seen seem to treat all four the same. Not sure these teams are all the same.
By the way, I missed all the criticism of the 76er forecast. Owen is right, you can never make some people happy.
Chirstopher // October 8, 2007 at 9:56 am
Well, you always stick out your neck when you make a prediction. In any event, I did what I suggested for the Hawks and Celtics. For the Hawks the exercise was underwhelming. ATL did not have an eventful offseason so predicting this year based on last year is fairly trivial. ATL will be 27-55 this year assuming the roster is trimmed down to 15 by cutting the non-rookie players who played the least last year and by assuming that both rookies have a WP48 of 0.07 and play 20 mpg.
For Boston, things were different. Boston should go 74-8 next year. That is what the simple scaling exercise yields. Here, the two recent undrafted players were cut, I assumed 10mpg and WP48 = 0.035 for the remaining draft picks, and pegged Tony Allen at WP48 = 0.100 (no data on Chandler’s site). What’s interesting is that total minutes played for all players was 58 minutes less than total possible minutes. The line-up then (player and mpg):
Batista, Esteban 0.99
Pollard, Scott 1.35
Powe, Leon 8.82
Davis, Glen 10.00
Pruitt, Gabe 10.00
Allen, Tony 11.07
House, Eddie 11.80
Scalabrine, Brian 12.49
Perkins, Kendrick 19.21
Jones, Dahantay 20.32
Pierce, Paul 21.23
Rondo, Rajon 22.30
Posey, James 25.51
Allen, Ray 27.10
Garnett, Kevin 37.10
Obviously not spot on but I don’t think it’s far off. Dahantay Jones will not get 20+ mpg, some of those minutes will go to Pierce and Ray Allen (which will increase WP overall). OTOH, I would not be surprised to see the top 5 in mpg start with Garnett as center… Also, Pollard is much better than his last year’s metric indicate. Ideally some way to adjust for smaller rotations in an unsupervised manner would be useful but I’m starting to drink Danny Ainge’s Kool-Aid…
Owen // October 8, 2007 at 12:28 pm
Christopher – See “the Pareto Principle and the Boston Celtics” and specifically:
But our focus is the Celtics. Right now we are not entirely sure who the Celtics are going to play along side Pierce, Allen, and Garnett. If the Celtics surround this trio with a collection that is equal in productivity to what the average NBA team received from “The Rest” (i.e. a WP48 of 0.042), the Celtics can expect to win between 50 and 53 games in 2007-08. If the Celtics are able to find complementary players as productive as what we see in Houston, the wins for this team jumps into the 60 plus range. Of course, if “The Rest” for the Celtics is in the New Jersey Nets range, then we are looking at a team that is going to win less than 45 games.
In sum, this team does look like a playoff team regardless of who else it employs. Still, Bob Ryan is correct to note that the complementary players do matter. Although the Celtics only won 24 games last season, a 45 win season in 2007-08 will be considered a disappointment. For this team to truly contend for a title, “The Rest” is going to have to do something in Boston.
Christopher // October 8, 2007 at 9:58 pm
Yeah, different assumptions = different outcomes. *shrug* I’m getting interested in mining the data to determine which line up is most likely. One of the issues in the best 5 above is that there really is no front court in the traditional sense. Maybe after rosters are trimmed to 15 I’ll look at some set of rules to find a rotation. If scoring is overvalued then perhaps coaches fall for this too?