This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.
«
»

Team Positional Production Allocation… or Something

Posted by Neil Paine on August 7, 2009

A quick post/data dump for Friday... When I was putting together some of the Win Shares-based lists we've been posting recently, I wondered how heavily each team relied on their frontcourt and backcourt, and whether or not that correlated to success. So, first I added up WS by a team's guards and their non-guards, making each into a percentage. Here are the most guard-heavy teams in terms of production this decade:

Rank Year Team Pct_G Star Guards TeamWS
1 2007 SAC 69.4% Kevin Martin Ron Artest Mike Bibby 37.9
2 2006 MIL 65.5% Michael Redd Bobby Simmons Charlie Bell 39.0
3 2008 MIA 63.7% Dwyane Wade Chris Quinn Dorell Wright 18.7
4 2009 NJN 63.5% Vince Carter Devin Harris Keyon Dooling 35.7
5 2005 ATL 62.8% Josh Childress Tyronn Lue Tony Delk 17.3
6 2003 MIL 62.3% Sam Cassell Michael Redd Ray Allen 42.9
7 2005 GSW 62.3% Jason Richardson Mike Dunleavy Jr. Derek Fisher 37.0
8 2002 CHI 61.9% Fred Hoiberg Jalen Rose Trenton Hassell 19.2
9 2002 CLE 61.8% Andre Miller Wesley Person Ricky Davis 34.5
10 2006 GSW 59.1% Jason Richardson Derek Fisher BaRon Davis 39.4
11 2001 IND 58.8% Reggie Miller Jalen Rose Travis Best 42.0
12 2006 PHO 58.6% Steve Nash Boris Diaw Raja Bell 59.6
13 2005 NJN 58.4% Vince Carter Jason Kidd Travis Best 37.1
14 2000 TOR 58.2% Vince Carter Doug Christie Muggsy Bogues 42.2
15 2000 NJN 57.7% Stephon Marbury Kerry Kittles Kendall Gill 40.2
16 2007 NJN 57.6% Vince Carter Jason Kidd Eddie House 41.2
17 2009 PHI 57.2% Andre Iguodala Andre Miller Louis Williams 42.1
18 2008 PHI 57.2% Andre Iguodala Andre Miller Louis Williams 42.3
19 2008 MIL 57.1% Michael Redd Mo Williams Charlie Bell 23.8
20 2004 SEA 56.6% Ray Allen Antonio Daniels Brent Barry 39.9

Overall, the correlation between Pct_G and total team WS was 0.091, which is not very big -- but at least it's positive.

Now, the most frontcourt-dependent teams:

Rank Year Team Pct_FC Frontcourt Stars TeamWS
1 2009 WAS 98.4% Antawn Jamison Caron Butler Darius Songaila 20.9
2 2003 DEN 90.7% Juwan Howard Nene Hilario Ryan Bowen 19.5
3 2002 MEM 89.2% Pau Gasol Shane Battier Stromile Swift 21.3
4 2006 UTA 86.0% Mehmet Okur Andrei Kirilenko Matt Harpring 34.6
5 2003 MIA 85.5% Brian Grant Eddie Jones Caron Butler 27.5
6 2004 MEM 85.0% James Posey Pau Gasol Shane Battier 48.4
7 2006 MEM 84.0% Pau Gasol Shane Battier Mike Miller 50.0
8 2009 OKC 83.5% Kevin Durant Nick Collison Jeff Green 26.8
9 2005 CLE 83.1% LeBron James Drew Gooden Zydrunas Ilgauskas 45.0
10 2009 MIN 82.6% Kevin Love Al Jefferson Mike Miller 27.6
11 2004 CLE 82.4% Carlos Boozer Zydrunas Ilgauskas LeBron James 35.7
12 2006 CLE 81.8% LeBron James Zydrunas Ilgauskas Drew Gooden 47.9
13 2005 UTA 80.9% Mehmet Okur Andrei Kirilenko Matt Harpring 28.4
14 2000 SAC 80.7% Chris Webber Vlade Divac Scot Pollard 50.0
15 2001 ORL 80.4% Tracy McGrady Bo Outlaw Mike Miller 45.4
16 2003 CLE 79.1% Carlos Boozer Zydrunas Ilgauskas Jumaine Jones 18.1
17 2003 UTA 79.0% Karl Malone Andrei Kirilenko Matt Harpring 48.2
18 2002 ORL 78.6% Tracy McGrady Horace Grant Pat Garrity 47.0
19 2003 ORL 78.2% Tracy McGrady Pat Garrity Mike Miller 43.4
20 2004 LAC 77.8% Elton Brand Corey Maggette Chris Wilcox 29.3

The correlation between Pct_FC and team success is actually -0.091, meaning the more reliant you are on your frontcourt for production, the more you actually tend to lose games. However, these correlations are truly microscopic in both cases, which means there is no real relationship between a team's allocation and its success. (We'll see later that balance is by far the more important determining factor in team wins.)

Anyway, I wanted to look at this one other way, since our positional designations are pretty generic and aren't always the most accurate in the world. The 40th percentile height of every player from 99-2000 to 2008-09 was 78 inches, so I classified every player 6'6" or shorter as "smalls" and 6'7" or taller as "bigs", just to add another angle that our positions were perhaps not picking up. Here were the teams most centered around their "smalls":

Rank Year Team Pct_S Star "Smalls" TeamWS
1 2003 NYK 72.8% Allan Houston Clarence Weatherspoon Howard Eisley 36.2
2 2007 MIL 72.4% Michael Redd Ruben Patterson Charlie Bell 29.9
3 2000 SEA 67.3% Gary Payton Brent Barry Ruben Patterson 43.8
4 2001 SEA 64.3% Brent Barry Ruben Patterson Shammond Williams 42.0
5 2001 CLE 64.1% Clarence Weatherspoon Andre Miller Trajan Langdon 30.2
6 2002 NYK 62.6% Mark Jackson Allan Houston Clarence Weatherspoon 30.1
7 2003 MIL 62.3% Sam Cassell Michael Redd Ray Allen 42.9
8 2002 CLE 61.8% Andre Miller Wesley Person Ricky Davis 34.5
9 2009 BOS 61.8% Ray Allen Paul Pierce Rajon Rondo 61.3
10 2008 LAC 61.1% Corey Maggette Brevin Knight Cuttino Mobley 23.2
11 2000 NYK 61.0% Allan Houston Latrell Sprewell Charlie Ward 43.4
12 2005 BOS 60.1% Paul Pierce Gary Payton Ricky Davis 45.0
13 2001 HOU 59.4% Steve Francis Cuttino Mobley Moochie Norris 47.7
14 2002 WAS 58.0% Chris Whitney Richard Hamilton Michael Jordan 38.0
15 2006 BOS 57.7% Paul Pierce Delonte West Ricky Davis 38.9
16 2009 PHI 57.2% Andre Iguodala Andre Miller Louis Williams 42.1
17 2008 PHI 57.2% Andre Iguodala Andre Miller Louis Williams 42.3
18 2004 SEA 56.6% Ray Allen Antonio Daniels Brent Barry 39.9
19 2009 MIA 55.7% Dwyane Wade Mario Chalmers Daequan Cook 43.0
20 2002 SEA 54.8% Gary Payton Brent Barry Earl Watson 49.5

Some crossover there with our guards table, except where C-Spoon seems to be concerned. The correlation between smalls' production and Win Shares was 0.094, essentially the same as we found for the guards. And that means it's essentially the same for the "bigs":

Rank Year Team Pct_B Star "Bigs" TeamWS
1 2009 WAS 98.4% Antawn Jamison Caron Butler Darius Songaila 20.9
2 2003 DEN 95.7% Juwan Howard Nene Hilario Ryan Bowen 19.5
3 2004 TOR 94.1% Donyell Marshall Chris Bosh Vince Carter 32.9
4 2002 MEM 90.9% Pau Gasol Shane Battier Stromile Swift 21.3
5 2007 ATL 90.0% Josh Childress Joe Johnson Zaza Pachulia 29.3
6 2009 SAC 86.9% Kevin Martin John Salmons Jason Thompson 20.7
7 2008 ATL 86.9% Josh Childress Joe Johnson Josh Smith 37.7
8 2006 UTA 86.0% Mehmet Okur Andrei Kirilenko Matt Harpring 34.6
9 2004 CLE 85.8% Carlos Boozer Zydrunas Ilgauskas LeBron James 35.7
10 2004 MEM 85.0% James Posey Pau Gasol Shane Battier 48.4
11 2009 OKC 84.8% Kevin Durant Nick Collison Jeff Green 26.8
12 2006 ATL 84.5% Joe Johnson Josh Childress Zaza Pachulia 31.1
13 2002 IND 84.5% Reggie Miller Jermaine O'Neal Jeff Foster 42.6
14 2005 CLE 84.4% LeBron James Drew Gooden Zydrunas Ilgauskas 45.0
15 2008 NJN 84.0% Vince Carter Richard Jefferson Josh Boone 29.4
16 2008 IND 83.6% Mike Dunleavy Jr. Danny Granger Jeff Foster 40.4
17 2006 CLE 82.9% LeBron James Zydrunas Ilgauskas Drew Gooden 47.9
18 2009 MIN 82.6% Kevin Love Al Jefferson Mike Miller 27.6
19 2001 IND 82.2% Reggie Miller Jalen Rose Jermaine O'Neal 42.0
20 2003 CHI 81.8% Donyell Marshall Tyson Chandler Jalen Rose 28.1

Obviously, the correlation between "big" production and total team WS is -0.094, in keeping with our earlier results with the positional designations.

Finally, I calculated a metric called "balance", which is simply the difference between a team's allocation and the "ideal" allocation of 2/5 of the team's WS going to guards/smalls and 3/5 going to bigs/frontcourt players. The correlation between "balance" and WS is essentially 0.3 for both the positional study and the height-based one, confirming the conventional wisdom that a team with balanced production between its backcourt and frontcourt will tend to do the best. This is still not a major correlation by statistical standards, mind you, but it is much higher than the correlation we were seeing for the teams biased toward either bigs or smalls.

What does any of this mean? Well, again, it's best to balance your production between your frontcourt and your backcourt, your bigger players and your smaller ones. But, it appears that there's a very slight trend over the last decade that says teams who rely on their guards and smaller players tend to win a few more games over the course of a season. This makes sense, given that the league spent most of the Oughts trying to tip the advantage in favor of perimeter scorers with modifications to the rules on hand-checking and more liberal foul calls on drives in general.

Related posts:

    Win Shares and Aging
    Youth Movements
    Biggest Differences Between a Player’s Best & Second-Best Seasons
    More Turkey Day Specials
    The Bird-Dirk Advanced Stat-off

This entry was posted on Friday, August 7th, 2009 at 5:17 am and is filed under Analysis, History, Win Shares. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

9 Responses to “Team Positional Production Allocation… or Something”

  1. KneeJerkNBA Says:
    August 7th, 2009 at 3:08 pm

    Love the site. But did Artest play any guard for Sacto? Every box I've checked has him at SF. And the only time I can remember him getting extensive time at guard was his rookie year for the Bulls.

  2. Ben Smith Says:
    August 7th, 2009 at 3:41 pm

    Wouldn't the fact that the two most dominant teams of the decade do not appear on any list seem to negate this premise? San Antonio and LA have won 8 of the last 11 titles.

  3. jd15 Says:
    August 7th, 2009 at 3:42 pm

    @KneeJerkNBA
    Yeah but SF is almost (basically) a guard. Depending on the offense it definitely can be. I think they were just taking a little liberty.

  4. Neil Paine Says:
    August 7th, 2009 at 3:47 pm

    @Ben Smith
    The fact that they don't appear just confirms that balanced teams have the most success, which we kind of already proved with the very low correlations between wins and a heavy allocation in either direction.

  5. Neil Paine Says:
    August 7th, 2009 at 3:53 pm

    As far as positions go, yeah, they're not always accurate to each specific season. When a player is considered one position, we don't really change it across their career even if they played a different position in some seasons. Artest has played both G and F, but he's considered a G here for his entire career. That's why I threw in the height thing too, just to catch guys that might be incorrectly assigned a position. Positions are so fluid in basketball that any designation is going to be arbitrary anyway.

  6. Jason J Says:
    August 7th, 2009 at 5:03 pm

    I guess it's not surprising that being overbalanced w/ bigs or smalls is not a good formula for winning.

    It's a corruption of the idea behind the post, but I would be very interested to see what WS has to say about the best Trios in NBA history. I'd love to see how Magic, Worthy, Kareem stacks up against Timmy, Manu, Parker or Jordan, Pippen, Grant and KG, Allen, Pierce. Or maybe less successful trios who have zero help would dominate the list like Run TMC and the current Wiz (but I doubt it).

  7. Kevin Says:
    August 7th, 2009 at 5:40 pm

    However, these correlations are truly microscopic in both cases, which confirms that a team with balanced production between its backcourt and frontcourt will tend to do the best.

    Am I missing something? I don't think that's confirmation at all. I accept that (a) there is essentially no correlation, and that (b) more balanced teams are probably the best (based on other evidence), but don't see at all how (a) leads directly to (b). If there were truly no correlation, then it wouldn't matter how frontcourt/backcourt-oriented you were, right? (the main problem here being the use of a linear model, which logically does not reflect reality)

  8. Neil Paine Says:
    August 7th, 2009 at 6:45 pm

    Thanks for the catch, I knew I forgot something in this post -- I also wanted to report what the correlation between "balance" (the difference between the ideal distribution, 2/5 guards and 3/5 FC, and the team's actual allocation) and wins was: 0.31, which while still small by statistical standards, is much higher than the 0.09 and -0.09 we reported for the guard-heavy and big-heavy teams. That's the missing piece; you're right that a correlation of 0 simply means no relationship in either direction (you could have a lot of wins with a guard-heavy team or few wins, but there's no relationship between wins and allocation), but when compared with the other correlation for "balance", we see confirmation that balance is far more correlated with success than heavy overbalance in either direction. I've changed the post to reflect this.

  9. Matt Says:
    August 9th, 2009 at 12:40 pm

    Yeah...just saw that it was odd that Artest was counted as a guard for Sac in '07 while Butler was counted as a big for Was this past season - though both played similar roles on their respective teams. I know this isn't an exact science, but I'd be interested in seeing what the numbers would be if Artest was counted as a frontcourt and Butler as a backcourt player.

«
»