Every player in the NBA is not an equal producer of wins. Last summer I noted – consistent with the work of 19th century economist Vilfredo Pareto (which did not mention basketball) – that about 80% of wins in the NBA are produced by 20% of the players. This means that the top three players on a 15 man roster tend to be the driving force behind a team’s success.
Last summer I posted a list of the top trios – defined as the three leaders on each team in Wins Produced – for each team in 2006-07. The list – re-posted below – revealed that the top trios were in Phoenix, San Antonio, Chicago, Dallas, and New Jersey.
Table One-A: The Top Trios in 2006-07
Table One-B: The Top Trios in 2006-07 (in alphabetical order)
The Top Trios in 2007-08: Wins Produced Edition
For 2007-08, Phoenix still finished in the top five. But the top spot was held by the New Orleans Hornets. Chris Paul, Tyson Chandler, and David West (mostly Paul and Chandler) produced 48 wins for the Hornets last year. Unfortunately the rest of the roster only produced seven victories.
Table Two: The Top Trios in 2007-08
Joining New Orleans and Phoenix in the top five were Orlando, Boston, and San Antonio. Back in November I argued that the Boston trio of Kevin Garnett, Paul Pierce, and Ray Allen would lead the league. Not only did this not happen, but Ray Allen lost his spot in Boston’s Top Trio to Rajon Rondo. So that’s one prediction that didn’t quite work out (of course I was still closer to the mark on the Celtics than Bob Ryan of the Boston Globe).
The Top Trios in 2007-08: WP48 Edition
As noted, Top Trios has been defined in terms of Wins Produced. What if we focused instead on WP48 [Wins Produced per 48 minutes] of the top trio [calculated by adding the Wins Produced of the top three together, dividing by minutes played, and multiplying by 48]?
Focusing on WP48 reveals that Phoenix – with Steve Nash, Amare Stoudemire, and Shawn Marion – led the way last season (as it did in 2006-07). Unfortunately the trade for Shaq, who posted a 0.227 WP48 with the Suns last year (Marion’s mark was 0.306), probably means Phoenix will not top the trio rankings next season (the new Suns trio, though, might still rank in the top five).
If Phoenix fades from the top spot, though, who is likely to take their place? To find candidates to lead the WP48 Trio Rankings for 2008-09 we first should look at the other top five threesomes this past season. And that list includes the trios in San Antonio, the LA Lakers, Boston, and New Orleans. If I had to guess, I would suspect the Lakers Trio of Kobe Bryant, Pau Gasol (or Lamar Odom), and Andrew Bynum will lead the way in 2008-09. Of course we should remember that my last attempt to predict this contest failed.
Three More Top Trios?
It was argued by Chris Broussard at ESPN that the addition of Ron Artest gives the Rockets a top trio. The data, though, suggests otherwise. In 2006-07 Ron Artest, Tracy McGrady, and Yao Ming posted WP48 marks of 0.114, 0.168, and 0.225 respectively. This works out to an average of 0.169, or a mark that would rank 19th in the NBA this past season. If we look at the 2007-08 WP48 marks (0.081, 0.107, and 0.232), we see an average of 0.140. This would rank 29th this past season. In sum, the acquisition of Artest does not give the Rockets a top trio.
Fortunately for Houston (and the team’s fans), the Rockets had the best supporting cast last season. So even though this team didn’t have a top trio last year – and probably won’t have one this next year – the Rockets will still be a “good” team. Probably not “Laker or Celtic good”. But Houston should win more than 50 games in 2008-09.
If we are looking for a new top trio, we need to get our minds off of Houston and start thinking about Miami. If the Heat keep Shawn Marion, Miami will have a trio consisting of a healthy Dwyane Wade, Marion, and rookie Michael Beasley. If Wade returns to what we saw in 2006-07 (and Beasley delivers on the promise of his college numbers) Miami’s new trio could be a top ten combo. And this should transform the Heat – who were led by Wade, Jason Williams, and Chris Quinn this past season — into an Eastern Conference playoff team.
After the Heat, another top threesome has also been assembled by the Clippers. As I noted a few weeks ago, the combo of Marcus Camby, Chris Kaman, and Baron Davis does put another top trio in LA. Unfortunately, the supporting cast in Clipperland is going to make it difficult for the Clips to join the Heat in the playoffs.
A Pareto Prediction
Let me close by looking at the link between the productivity of a team’s top trio and a team’s final record. To address this issue, I regressed a team’s actual wins last season on the Win Produced of each team’s Top Trio. The results indicated that the Wins Produced of the top three players explains 66% of a team’s actual wins. In other words, as the Pareto Principle suggests, the top players on each team tell us quite a bit about a team’s ultimate success.
Of course, if all you know is the performance of the top three on a team, you don’t know everything about that team’s ultimate level of success. In other words, 66% is something, but not everything. A strong supporting cast – like we see in Houston, Boston, and Detroit – can overcome a relatively weak Top Three (or in Boston’s case, give the Celtics the top team in the league). That being said, the Heat – if The Matrix stays in Miami – might be back in the playoffs in 2009. And that return is primarily due to the strength of this team’s Top Trio.
- DJ
The WoW Journal Comments Policy
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.
There are a handful of guys that might begin to appear better by your measure if you included offensive fouls drawn. It is effectively the same as a steal, which has a major impact on wins produced and win score. Please consider it. Anderson Varejao, Shane Battier, and I believe just a few other guys would see their numbers change considerably. I just thought of it looking over the ‘box scores don’t demonstrate how effective Shane Battier is’ arguments. I disagree, but I think fouls drawn is probably the most important thing missing, at this point, for more conservative defenders who are still effective.
Does this only take into account wins produced by a player for that partiuclar team? I see Jason Kidd is in the top 3 for both NJ and Dallas – does that mean that Jason Kidd produced more wins for Dallas in his 1/3 of the season he played there than did (e.g.) Josh Howard in his entire season? Or are you counting all of Jason Kidd’s wins produced for the entire season for both NJ and Dallas?
In any event, interesting that, for the second year running, “the rest” in NJ produced negative wins. No wonder Rod Thorn blew up the entire team.
Chris Quinn?
you can’t be serious….
so in conclusion…your analysis is flawed
What’s that! A below average Chris Quinn was the third most productive player on the worst team in the league! Why that’s…that’s…um…entirely plausible. Thanks for the enlightenment, Jim.
Kevin, I ran Shane Battier’s numbers and I get an unadjusted WP48 of 0.102 for 07-08. When I credit him with drawing 50 offensive fouls during the 08 season and credit them as steals, his WP48 goes up to 0.112.
(The 50 offensive fouls is the same number that he drew during the 06-07 season according to 82games. He drew 48 fouls during the previous season.)
The 06-07 offensive foul drawing leader Anderson Varejao has an unadjusted WP48 of 0.150 prior to accounting for offensive fouls drawn and 0.180 after (for the 06-07 season). He drew 99 offensive fouls.
Kevin, WOW already accounts for all forced not stealed TOs (including off. fouls drawn) in a team defense adjust, but the distribution is not made by individual play by play, but by players’s minutes.
If the relationship between Battier’s forced turnovers and team’s is not the same as the minutes relationship, then he could be losing some credits.
Although all FTO and steals have the same possession value, the marginal FG% scoring value produced by steals is a bit more.
Pingback: Top Posts « WordPress.com
But of course Kevin, the subjective marginal value of the foul is not added to the forced turnover-charge. That is, when sending some important players to the bench, out of game, reducing their hustle, or putting the opp. team on FT penalty, etc.
Flowers: Thanks.
Harold: Picked up to book and read of English until comprehending can I of you the better sum.
Flowers: So Battier goes from average to ever so slightly better (1 steal to 1.59 forced turnovers). Since you’re obviously no novice, tell me, is pace controlled for or mostly unimportant?
Flowers: Battier’s simple win score (with steals converted into FTO (steals + charges taken) goes from .174 to .189, yes? How do you calculate Wins produced? I just took WP48 at your word and did this: (Battier’s minutes/48) * WP48 = 6.175 WP w/o FTO, and 6.78 with it, just as a rough test. It’s a small, but perhaps meaningful difference (assuming I didn’t fup by doing it that way), when sorting out a tight pack of middling or similar players.
Oh, and while the difference is small when adding forced turnovers, it’s actually a bigger difference than Battier makes with his shot blocking in the simple win score measure.
Uh, scratch mister happy face, that’s supposed to say Battier’s minutes divided by forty-eight.
I’m pretty sure that the WS per minute is 0.174. In trying to derive (fudge) Battier’s WP48 (0.096) from the WS48, I divided by 82 (I should have used 87 for Battier and 74 for Varejao). If the relationship between WS48 and WP48 is linear or nearly so then the numbers with drawn offensive fouls should be good approximations of WP48. Using 87 as the divisor for Battier gives 0.096/0.105 for 07-08. Using 74 for Varejao gives 0.166/0.200 for 06-07.
Oh Jim. You are a fool
The logic to bump based on Offensive Fouls Drawn is flawed.
Right off the bat, I can logically think, you have to compare their fouls drawn to the NBA average at their position, otherwise, someone who only draws 1 foul per season, would still get an increase in WP48, which logically makes no sense.
So you’d have to calculate the average fouls drawn per minute, and create a calculation based on the difference between the said player, and that NBA average.
Nick, check out 82games.com. You’re right about needing the averages. They report it over at 82games, and you can see for yourself Battier and Varejao draw an above average number of charges, and many other guys draw a very low number of charges. So we can learn a little bit more about many players on both ends of the spectrum, so long as we can get the averages. I don’t quite get what needing averages has to do with logic, but ok, we need averages.
Nick, the original hypothetical was that drawn offensive fouls should count roughly the same as steals. Seeing as we can plug them into the WinScore formula and look at overall productivity (using a hack) why do we need to know anything about the averages? Also, what exactly is the problem with a player’s rating increasing when they do something additional that helps their team win? Player X + 1 additional steal per year is indeed more productive than that same player without that additional steal, whether you’re dealing with total production or average production.
This is an incredibly good framework for statistically valuing NBA players, http://hardwoodparoxysm.blogspot.com/2008/08/arbitrarian-individual-contributions-to.html Absolutely incredible.
ilikeflowers the key is that everyone on the court cannot have an increase in winscore…
because at the end of each game, only 1 team wins. So how can you increase the number of wins in the league?
the guys who draw more than the average amount of fouls should increase in wins produced, and the guys who draw an under average number should decline.
Otherwise, a guy who only draws 1 offensive foul per season, would have his wins produced go up, when the NBA average player maybe drawing like 10, meaning that the player is below average.
Kevin, perhaps I could have worded it better, but when i read the other comment, I just “knew” something was up, so I thought about it, and the best way for me to word that feeling was “logic”.
Nick, none of that is going to meaningfully affect the change in Battier’s productivity. The original point was trying to assess how much Battier was being undervalued by not crediting him with offensive fouls drawn, and the approach used was logical. The resulting answer for Battier was that he’s slightly undervalued but still right around average.
If we are interested in the entire league then treating the charges drawn as steals would logically lead to charging the players committing those fouls with additional TO’s (and subtracting those TO’s from the team statistics since – according to the book – turnovers that are not caused by steals are assigned to the team). Regardless, as long as Battier (or anyone else) is not committing a large number of drawn offensive fouls himself (relative to what he draws) then the positive change in his productivity is not affected significantly, although it may be lessened. To my knowledge Battier doesn’t commit a large number of offensive fouls, but even if he did that would only serve to reduce the positive impact of his offensive fouls drawn and reinforce the fact that he is essentially what the unmodified model already says that he is, namely an average player (which is not a bad thing).
Sometimes math is fantastic and proves a point. Sometimes it doesn’t. In this case, I am utterly unconvinced that McGrady, Yao and Artest isn’t a top 5 trio based on them being healthy
The New Orleans trio was better than the rest, but their bench was one of the weaker ones. Boston, Detroit and the Lakers had stronger benches than the rest of the league. It makes me wonder what the best approach for a GM would be; building around a few “stars”, focusing on depth (i.e., six to eight 0.100+ players) or somewhere in between? What teams have managed to stay competitive through the years, avoiding the spikes like we’ve seen from the Heat? What have their rosters been like? Just food for thought…