Home Unfiltered Articles Players Baseball Prospectus
Basketball Prospectus home
Click here to log in Click here for forgotten password Click here to subscribe

Click here for Important Basketball Prospectus Premium Information!

<< Previous Article
Around the Rim (10/25)
Next Article >>
SCHOENE on 2010-11 (10/25)

October 25, 2010
Prospectus Hoops List
NBAPET and Modal Reality

by Bradford Doolittle

Printer-
friendly
Contact
Author

Will the Clippers win the NBA title in the upcoming season? If you were to ask David Lewis, the late American philosopher, he'd say yes. Lewis was a modal realist, which in effect means that somewhere in the infinite realm of time, space and dimension, everything that can be will be. Somewhere under the umbrella of limitless possibility, circumstances will conspire in just such a way to enable the Los Angeles Clippers to be crowned as NBA champions sometime in June, 2011.

How likely is that to happen? Not very, but the possible answer to that question, and others like it, is what we're addressing today. I first became enamored of preseason simulations back when ESPN.com first started running the results of DiamondMind Baseball sims and publishing the results. The great thing about this method of establishing expectation for the upcoming season is that it underscores a couple of fundamental aspects about sports that all thinking fans would do well to remember:

1. There is (almost) always hope for your team.

2. Any given season is merely one outcome amongst an infinite array of possibilities.

The first of these lessons falls under the umbrella of "anything can happen" and is what first drew me to simulations. I am a lifelong fan of the Kansas City Royals. As bad as things have been, the Royals won at least one division title in a DMB sim season each year from 1998 to 2001. Believe it or not, that gave me hope. The Royals went on to lose at least 85 games in each of those seasons, but that's not the point. When those campaigns started, I was armed with the knowledge that if everything (and perhaps more) fell into place, then even the Royals had a chance. Of course, there were subsequent seasons in which the Royals were given zero chance of experiencing October baseball by DMB. That was at least fuel for my inner realist.

I don't have the time, or the computing power, to run endless simulations of the upcoming NBA season. So I ran 10,000, which seems like plenty. Here are the general steps I went through to generate these results.

1. Use NBAPET to generate individual player projections for the upcoming season. Since I'm not using SCHOENE, the official projection system of Pro Basketball Prospectus 2010-11, this gives us an alternate take on the teams as preseason ends. NBAPET calculates projected offensive value in a similar fashion to SCHOENE, though this year I incorporated on-court/off-court data to model the intangible effects of a player's game. Defensive value uses a modified, multi-season dMULT that I've combined with data from Synergy. On-court/off-court data is used to adjust dMULT as well. Geez, this paragraph is a bit heavy on jargon. Sorry about that.

2. Using the same playing time projections as SCHOENE, I calculated a baseline win total for each team. Usage rates were adjusted to fall in line with anticipated rotations and playing time was adjusted to give each team a total of 19,800 minutes. Here's a peek at those baseline projections, which gives us a glimpse at how NBAPET assesses each team's offense and defense:

TEA     W    W% Rnk   ORTG Rnk   DRTG Rnk
atl  39.6  .483  14  109.2  16  109.8  16
bos  53.7  .655   3  110.3  10  105.4   1
cha  36.3  .442  19  106.8  28  108.5   9
chi  38.3  .468  17  108.9  18  109.9  19
cle  33.7  .411  26  109.2  17  112.0  27
dal  49.0  .598   7  111.5   6  108.4   7
den  50.1  .611   4  111.7   5  108.1   5
det  34.6  .421  23  107.9  22  110.4  21
gsw  35.9  .437  20  108.7  19  110.7  23
hou  43.2  .527  10  110.5   9  109.7  14
ind  35.4  .432  21  106.9  26  109.1  11
lac  34.3  .418  25  108.0  21  110.6  22
lal  55.1  .672   2  112.5   4  106.9   2
mem  31.8  .387  27  109.6  14  113.3  30
mia  61.6  .751   1  116.2   1  107.4   4
mil  38.9  .474  16  107.7  23  108.5   8
min  29.9  .364  29  106.9  27  111.2  26
njn  34.9  .425  22  108.7  20  111.1  24
nwo  50.1  .611   5  113.6   3  110.0  20
nyk  40.5  .494  12  109.7  13  109.8  17
okc  39.8  .486  13  109.3  15  109.7  15
orl  49.4  .602   6  111.5   7  108.2   6
phi  34.4  .419  24  107.1  25  109.6  13
phx  37.9  .462  18  110.0  11  111.2  25
por  44.1  .538   9  114.3   2  113.1  29
sac  39.4  .480  15  106.5  29  107.1   3
sas  45.9  .560   8  110.6   8  108.7  10
tor  27.4  .334  30  107.6  24  113.1  28
uta  42.1  .514  11  109.9  12  109.5  12
was  30.0  .366  28  105.7  30  109.9  18

3. Using this baseline of expected wins as inputs and a Monte Carlo simulator I laced together, I ran a simulation of the season by running the probabilities through the actual 2010-11 schedule.

4. The results of the simulated season fed into a playoff simulator, which was based on the log5 engine that I've used to project the playoffs the last few years.

5. Repeat above 9,999 times while tracking the results. Thankfully I wrote a macro to do this for me. The macro was activated by hitting CTRL-W. Rather than doing this over and over and risking rigor mortis in my fingertips, I went high-tech: I stacked some pennies on the 'CTRL' and the 'W' keys on my keyboard, then placed a weight on top of the stacks press the keys and get the macro running. It reached 10,000 in about 30 minutes. Later, Kevin Pelton sent me some iterative code that would automated the process for me. Oh well. As I always say, I'm not a statistician or a programmer, I'm just a writer.

That's it. Nice and easy. Let's run through the simulation results team by team, in order of average wins per sim season, with a word to two on each team. By the way, Clippers fans, there is hope. Perhaps not in this universe, but out there somewhere in infinity.

KEY: (all metrics based on 10,000 season simulations)
BASE_W: baseline wins projection
SIMS_W: average win total
SKED/RNK: schedule strength as measure by ratio of sim wins to baseline wins
>=.500: percentage of sims team finished with at least 41 wins
PLYOFF: percentage of sims team made postseason
FNLS: times reached finals; CHMP: times won championship
HI: most wins in a sim season; LO: fewest wins in a sim season
>=72: sim seasons with 72 or more wins

1. HEAT

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Miami          61.6    62.3  .989/ 16  100.0%  100.0%  6795  5179  75  45   47

NBAPET assigns the Heat an absurdly high Offensive Rating, which is interesting because SCHOENE marked Miami's super team as a defensive juggernaut. NBAPET has the Heat fourth on that end, so it's not exactly in disagreement there. The projected Offensive Rating (116.2) would be among the best post-merger figures, but I'm skeptical became I'm not sure NBAPET is capturing the diminishing returns of combining the mighty trio of LeBron James, Chris Bosh and Dwyane Wade. If the system is right, look out. The sims give the Heat a 52 percent chance of winning the title and a 0.47 percent chance of winning 72 games.

2. LAKERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
L.A. Lakers    55.1    55.5  .993/ 14  100.0%  100.0%  4098  1642  70  40    0

NBAPET generally falls in line with conventional wisdom this year, which I actually don't like. SCHOENE is designed to zero in on breakout players and teams, as well as those primed for collapse. NBAPET is much more conservative, but that's the essence of why I put up these results, because they can be so different from SCHOENE. Ultimately, both systems will have its share of hits and misses. NBAPET likes the Lakers much more than SCHOENE, seeing the defending champs as a the league's second-best defensive team and an almost an even-money bet to get back to the Finals for the fourth straight season. However, the sims clearly cast the Heat as the favorite to take home all the marbles.

3. CELTICS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Boston         53.7    54.6  .983/ 18  100.0%  100.0%  2130  1233  70  40    0

The Celtics' results echo those of the Lakers--Boston is being largely rewarded for its past body of work. NBAPET is simply not built to identify age- and injury-based collapses. Everyone declines on a nice, smooth downslope in the Utopian world of NBAPET. I look at Boston's projection here as what the season will look like if its primary pieces don't all get old at once, which is a part of what plays into SCHOENE's pessimism about the Celtics. NBAPET forecasts the Celtics to be the league's best defensive team. That might not seem like a stretch, but the system doesn't account for the possible void left by former assistant coach Tom Thibodeau. (It does account for a big chunk of missed time for Kendrick Perkins.)

4. NUGGETS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Denver         50.1    50.4  .994/ 13   99.0%   98.8%  1683   474  66  35    0

NBAPET projects Denver as the league's fifth-best defensive team, which I ain't buying. Isn't it a bitch when your own system tells you something you can't believe? If it actually happens, then the Nuggets have about a five percent chance of winning the championship. Of course, the Carmelo Anthony situation would make all of this moot should he be traded. Can't forecast trades--yet.

5. MAGIC

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Orlando        49.4    50.2  .984/ 17   99.0%   99.8%   815   361  64  33    0

My system perpetually underrates the Magic, and I'm not sure why. They are forecast as the seventh-best offense and sixth-best defense, the latter of which seems off, as does the average win total of 50.2. Over 10,000 simulations, team results are invariably pulled towards .500. That's kind of the point--setting in effect an over-under for each team's win total. In this case, I'd take the over.

6. HORNETS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
New Orleans    50.1    50.1 1.000/  7   98.8%   98.3%  1720   517  66  33    0

Easily the most surprising result spit out by NBAPET was the 50.1 average win total the Hornets achieved in the simulations. New Orleans projects as the league's third-best offensive team. Last year, the Hornets were 16th. Sure, Chris Paul is forecast to be healthy, but still. By the way, Jerryd Bayless was still on the Trail Blazers when the sims were run.

7. MAVERICKS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Dallas         49.0    49.0 1.001/  5   97.6%   97.2%  1258   355  66  31    0

There is a lot of buzz over the Mavericks as the season opens, with the prevailing opinion seeming to be that Dallas is the team most likely to overtake the Lakers should the champs stumble. NBAPET is placing the Mavs behind Denver and New Orelans in the West, but there is little to separate the three teams.

8. SPURS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
San Antonio    45.9    45.8 1.002/  3   88.9%   88.4%   457    72  62  30    0

The Spurs are gearing up for perhaps the last run together for Tim Duncan, Tony Parker and Manu Ginobili, but NBAPET sees the Spurs as no better than a second-tier team.

9. TRAIL BLAZERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Portland       44.1    43.9 1.005/  2   77.8%   76.5%   299    50  62  29    0

SCHOENE is far more optimistic than NBAPET about the Blazers. For what it's worth, if there is one team I'd bet would outperform their projection in this piece, it'd be Portland. Actually, it'd be Oklahoma City. But then the Blazers. For sure.

10. ROCKETS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Houston        43.2    43.5  .993/ 15   74.9%   72.9%   226    35  59  26    0

This seems about right for the Rockets.

11. JAZZ

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Utah           42.1    42.1 1.001/  6   64.5%   60.5%   156    27  57  25    0

I think the Jazz will be better than this, as a No. 12 ranking in Offensive Rating is too low for a team with Deron Williams at point guard and Jerry Sloan as the coach. The Jazz did lose a lot of talent, however.

12. KNICKS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
New York       40.5    41.2  .983/ 20   56.5%   82.9%    90    19  56  26    0

NBAPET is seeing the league as more West-leaning and the East as more top-heavy than I think will be the case. We'll see. As it is, the system forecasts the Knicks as the fourth-best team in the East, albeit with a barely break-even mark.

13. HAWKS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Atlanta        39.6    40.4  .980/ 21   48.4%   77.2%    50    12  60  22    0

Like SCHOENE, NBAPET is projecting a big step back for Atlanta. As I wrote in the book, there is something about the Hawks that our metrics just don't seem to pick up on. That doesn't mean we're wrong about a regression.

14. THUNDER

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Okla. City     39.8    40.0  .995/ 12   45.5%   39.0%    39     4  56  22    0

Don't believe this one. It wouldn't shock me if the Thunder fail to win 50 games again. It also wouldn't shock me if they rocketed into the high 50s. This projection is dragged down by the two seasons previous to last year, which is always going to be the case with young, breakout teams.

15. BUCKS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Milwaukee      38.9    39.7  .979/ 22   42.8%   71.5%    43     9  56  25    0

The Bucks have improved so much the last two years that they fall into the same trap as the Thunder in terms of being dragged down by previous poor performance. I do think the system is justified in being leery about the Bucks' offense, which is ranks at No. 23. Despite the additions of Drew Gooden and Corey Maggette, I still think that too much is reliant upon rapid improvement from Brandon Jennings, especially since Andrew Bogut may be diminished in the coming season.

16. BULLS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Chicago        38.3    39.4  .973/ 30   40.6%   68.4%    38     4  57  23    0

It's always interesting to see how a new mix of players comes together, both on the spreadsheet and on the court. In the Bulls' case, the offensive projection is dragged down by the returning core, which was in turn held back by the system of former coach Vinny Del Negro. The defense, however, is held back by the addition of Carlos Boozer. I think Chicago will be better than this, perhaps by a lot. However, this projection does give me pause. But, hey, as I hearken back to my bit about hope, the Bulls did win four of the 10,000 sim championships.

17. KINGS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Sacramento     39.4    39.4  .999/  8   40.2%   32.7%    46     3  53  24    0

Frankly, this one leaves me a bit speechless. I don't mind the Kings' win projection. I actually think the Kings have a real shot at finishing better than .500. I am stunned by NBAPET forecasting Sacramento as the league's third-best defensive team. I know they added Samuel Dalembert (eventually), and the system loves Tyreke Evens, but still.

18. SUNS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Phoenix        37.9    37.6 1.007/  1   24.8%   18.6%    12     0  53  22    0

I don't know whether this projection is high, low or right on the money. I do think it's a very bad sign that the Suns didn't win a single simulation.

19. BOBCATS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Charlotte      36.3    36.9  .983/ 19   20.8%   44.8%    15     0  52  20    0

NBAPET sees the Bobcats in the quagmire of teams trying to nab the last couple playoff spots in the East. That seems about right. However, unless the point guard position is upgraded, I see a potential for collapse here.

20. PACERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Indiana        35.4    36.4  .973/ 29   16.8%   38.8%     4     0  56  20    0

Not much to say about the Pacers, but they do have the league's second-easiest projected schedule, though back-to-backs aren't accounted for in that calculation. That could help Indiana separate from the other teams fighting for seventh and eighth in the East.

21. WARRIORS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Golden State   35.9    35.9  .999/  9   14.9%   10.3%     5     0  53  20    0

NBAPET is not nearly as excited about the Warriors as SCHOENE.

22. NETS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
New Jersey     34.9    35.8  .974/ 28   13.6%   32.7%     5     0  52  19    0

Would any Nets fan be disappointed by this kind of improvement? Shouldn't be.

23. PISTONS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Detroit        34.6    35.4  .976/ 26   11.5%   29.1%     3     0  52  18    0

Right now, the Pistons are the most "meh" team in the league. I'm not even sure what that means, but it's a term that is ubiquitous on Twitter and seems apropos for Detroit.

24. 76ERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Philadelphia   34.4    35.3  .974/ 27   11.4%   27.5%     9     3  52  17    0

The Sixers won three sim seasons, a fact which I can't explain, except to underscore today's theme about everything being possible and, in fact, inevitable. If I'd run enough sims, every team would have a title or two.

25. CAVALIERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Cleveland      33.7    34.5  .978/ 25    8.5%   21.5%     3     0  50  18    0

The Cavaliers have 0.03 percent chance of reaching the Finals, so it's not likely to happen. But wouldn't it be a gas if it did?

26. CLIPPERS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
L.A. Clippers  34.3    34.4  .997/ 11    7.7%    5.1%     1     1  50  20    0

Love it! There is a universe out there somewhere in which the Clippers are going to win the title. This. Season. It's probably the same universe in which Spock is duplicitous and evil and Kirk has a shapely, opportunistic concubine stashed in his quarters. (Trivia: The actress that played that concubine boards her dog as the same place I take mine. Now you know. There's an autographed picture of her on the wall. It's a shot from the episode. She only got one ep but, hey, she's got an action figure.)

27. GRIZZLIES

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Memphis        31.8    31.8  .999/ 10    2.0%    1.4%     0     0  47  16    0

The Grizzlies' have a horrific defensive projection which I don't quite buy, but is also entirely possible. Memphis looked great in the preseason and I think the Grizz will make a playoff run. However, we have to acknowledge that their roster is a potentially combustible mix.

28. WIZARDS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Washington     30.0    30.7  .978/ 24    1.0%    4.8%     0     0  45   8    0

All signs point upwards. Just not this season. Yes, the Wizards won eight games in one of the simulations. That's what we call a fluke.

29. TIMBERWOLVES

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Minnesota      29.9    29.8 1.002/  4    0.6%    0.4%     0     0  46  17    0

No argument here.

30. RAPTORS

TEAM         BASE_W  SIMS_W  SKED/RNK  >=.500  PLYOFF  FNLS  CHMP  HI  LO >=72
Toronto        27.4    28.0  .978/ 23    0.2%    1.0%     0     0  44  18    0

Here either, though I think Toronto's projection as the 28th-ranked defense is actually optimistic.

NOTE: The Hoops List this year will take on a different format. Over the last three years, I've attempted to maintain a weekly rankings piece that would run upwards of 4,000-5,000 words. The size of the task became a grind and, thus, sometimes the list fell through the cracks. This season, I'm shifting to a monthly format, which I'll roll out five teams at a time during the last week of each month, beginning with the end of November. It's not necessarily a better format, but it is more sustainable.

You can follow Bradford on Twitter at twitter.com/@bbdoolittle.

You can order a copy of Pro Basketball Prospectus 2010-11 here.

Bradford Doolittle is an author of Basketball Prospectus. You can contact Bradford by clicking here or click here to see Bradford's other articles.

0 comments have been left for this article.

<< Previous Article
Around the Rim (10/25)
Next Article >>
SCHOENE on 2010-11 (10/25)

RECENTLY AT BASKETBALL PROSPECTUS
State of Basketball Prospectus: A Brief Anno...
Tuesday Truths: March-at-Last Edition
Easy Bubble Solver: The Triumphant Return
Premium Article Bubbles of their Own Making: Villanova, Temp...
Tuesday Truths: Crunch Time Edition

MORE FROM OCTOBER 25, 2010
Trend Watch: Love, the Hawks and Ts
SCHOENE on 2010-11: Final Projections
Around the Rim: Iowa Hits the Gas Pedal
Five Thoughts: Mavericks and Bulls
Five Thoughts: Blazers and Nuggets

MORE BY BRADFORD DOOLITTLE
2010-10-28 - PBP Roundup: Games of Oct. 27, 2010
2010-10-27 - PBP Roundup: Opening Night
2010-10-26 - 2010-11 NBA Preview: Our Picks
2010-10-25 - : NBAPET and Modal Reality
2010-10-25 - Five Thoughts: Mavericks and Bulls
2010-10-25 - Five Thoughts: Blazers and Nuggets
2010-10-14 - Five Thoughts: Raptors-Bulls (preseason)
More...

MORE PROSPECTUS HOOPS LIST
2012-10-30 - : Preseason 2012-13
2010-10-25 - : NBAPET and Modal Reality
2010-03-29 - : Week Ending March 28, 2010
2010-03-25 - : Week Ending March 21, 2010
2010-03-15 - : Week Ending March 14, 2010
More...

Basketball Prospectus Home  |  Terms of Service  |  Privacy Policy  |  Contact Us
Copyright © 1996-2014 Prospectus Entertainment Ventures, LLC.