BBR Mailbag: The Most Offensively Balanced (and Imbalanced) NBA Finalists
Posted by Neil Paine on June 3, 2010
Today I have a mailbag question from our friend David Biderman at The Wall Street Journal:
I had an NBA playoffs question. I took a quick (really quick) glance at the Celtics, and the first thing that jumps out is how balanced their starting five was during the regular season. Is there any way to quantify how balanced a team is? If so, could we do this for NBA finals teams?
Absolutely; in fact, it's a concept I looked into a bit for these posts:
Spreading It Around
Championship Usage Patterns and “The Secret”
Championship Usage Patterns II
Championship Usage Patterns III: Regular-Season Teams Built For the Playoffs
However, I hadn't looked at NBA Finalists in particular yet, and I also would like to take this opportunity and try out a new metric to measure how balanced teams' starting lineups are.
The new metric (well, new to me at least) is called the Herfindahl-Hirschman Index (HHI), which measures how concentrated a market is among its firms. A monopolistic market will have a high HHI score, indicating that the majority of the market power rests with only a select few firms; conversely, a lower HHI score indicates a competitive market in which all firms have relatively equal shares. If you think of a basketball team as a market, you can apply this logic to a 5-man unit -- every player is working together to create points, but they're also "competing" against each other for touches and shots. A "monopolistic" lineup would be one where only one or two players take the majority of the possessions (think the 2000-02 Lakers with Shaq & Kobe), while a "competitive" lineup would be one where the offensive chances are distributed relatively evenly among the 5 players on the court.
So here's how I applied the HHI to Finals starting 5s... First, for every player in playoff history, I calculated Modified Shot Attempts (MSA):
MSA = 1.00 * (FG) * (1 - ((tmAST) / (tmFG)))+ 0.50 * (FG) * ((tmAST) / (tmFG))+ 1.00 * ((FGA) - (FG))+ 0.44 * (FTA)+ 0.50 * (AST)
This is essentially a measure of possessions used, except without turnovers because -- regrettably -- we don't have individual turnover stats for years before 1978. I then calculated the percentage of the team's MSA the player used when on the court (%MSA). For every Finals team ever, I isolated just their top 5 players in minutes -- not always necessarily their 5 starters, but usually the two groups are basically the same -- and scaled their %MSA to add up to 100% for that 5-man unit. I called the scaled %MSA term "share" -- as in market share, if the 5 man unit is a market and each player is a firm competing to use possessions. That way we can calculate the HHI, which is equal to the sum of the squared shares for a team. Here are the most imbalanced NBA Finals teams ever:
Wow, lots of MJ there. And at the other end of the spectrum, here are the most balanced Finalists ever:
It's truly remarkable how balanced the 1973 Knicks were -- consider that a "perfect" HHI for a basketball team would be 20% (5*0.2^2), which was almost exactly the Knicks' score that year (technically it was 20.0328%, but that's close enough). Their top player by "share" was Walt Frazier with 20.8%; their starter with the lowest share was Bill Bradley with 18.7%. That's amazing, because it means on any given Knick possession, any of their starters were basically just as likely to put pressure on the defense. With easily the league's best estimated Offensive Rating in the playoffs (102.1, 3 pts/100 poss. better than the #2 Celtics), it's easy to see why those Knicks were a nightmare to defend that postseason.
But in total, how does team balance affect the outcome of a Finals series? Here is every Finals matchup since 1952, and how their HHI scores compared:
In the 58 NBA Finals from 1952-2009, the more balanced team won just 43.1% of the time -- and that's bad news for Boston, whose 21.2% HHI is lower than L.A.'s 21.9% mark. In general, this backs up what I found here, which was that teams with a more imbalanced attack, concentrating most of their possessions among only 2 players, were more successful that those who evenly spread their possessions among more of their starters. But before you write off the Celtics, remember that this group bucked that trend once before: in the 2008 Finals, L.A. was a team constructed much more like a typical NBA champ than Boston, but the Celtics prevailed anyway. Over the next few weeks, we'll find out if they can beat the odds again, or if the Lakers' traditional championship construction wins out this time around.
June 3rd, 2010 at 11:10 am
Rafer Alston was the top usage man on last year's Orlando squad. Might be one of the reasons that they lost.
June 3rd, 2010 at 1:57 pm
I think the key for that Knicks team was that everybody could really shoot, and nobody cared who got the ball. Pretty funny that the coach who won 10 titles with uber-usage kings Jordan and Kobe learned at the feet of Red Holzman who coached the champion with the most evenly distributed usage in league history.
June 3rd, 2010 at 2:37 pm
I'm surprised the 04 Pistons weren't one of the more balanced teams; I guess we can blame Ben Wallace...
All this proves is that there's more than one way to skin a cat; dominant players win championships, but plenty of balanced teams have won it all too. I would compare the current Celts to the 82 Lakers; they have a balanced attack, but they are also a star-studded team of HOFers
June 4th, 2010 at 5:38 am
shoot,this was supposed to be a sweep.OK hope 4-1 Celts..
June 4th, 2010 at 4:44 pm
It seems to me that there are a couple of interesting parts involved here. My theories are based on my current understanding of the salary cap (not sure about how that has worked historically)
1) Every generation, there are anywhere from one to five truly dominant, high usage players, whose win total far surpass their peers. Lebron and Chris Paul (Durant this year as well) offer so many wins above replacement that they set they their teams far above the pack. These teams SHOULD naturally win more titles with their cap related competitive imbalance (caused by individual player max contracts).
2) Relying on fewer players leads to greater team variance. Injuries and foul trouble (in the playoffs) are obvious issues here. But I would also ask if team construction as a whole leads to greater season long variability for many of the 2 superstar teams. I suggest that because, due to cap restrictions, these teams often have to depend on low usage, relatively cheap role players. They simply can't afford to spread the wealth as much due to their high investment in their star players. These role players can either be rookies or undervalued veterans, about whom there is less league wide information (especially given their future usage context). In any given year, these relative unknowns are probably more likely to under/over perform their projections.
If their are four legitimate two superstar teams out their any given year and these teams are, on expectation, equal to their less variable egalitarian peers, it would makes sense that one of the high variance teams would hit it big on the variance express and win the finals. That probably depends on sample size (not a great stat guy) but I doubt 53 finals is enough to smooth it out.
My guess is that some combination of theories 1 and 2 explain why 2 superstar teams tend to win more championships--at least in the current salary cap context.
June 5th, 2010 at 4:47 am
I'm not sure if anybody has noticed the following.
But for most Neil's position or player related lists lately, the weight of Michael Jordan is massive. He is misconstruing almost all lists. It's only fitting that the Alpha Dog takes his rightful throne in this category.
June 5th, 2010 at 9:24 am
Bob (and Neil alike), I suggest you watch this piece of commentary with Bill Walton:
It's a great piece. He covers more ground on this subject, in laymen terms, than you would expect.