Posted by Neil Paine on March 8, 2011
The close-game struggles of this year's Miami Heat are nothing if not well-documented. A 5-13 record in games decided by 5 or fewer points has become the team's defining stat, far surpassing LeBron James' gaudy all-around numbers or the scoring brilliance of Dwyane Wade. As far as the mainstream media is concerned, it is now assumed this team will choke until they prove otherwise.
As statheads, we typically detest this sort of cliched, pseudo-psychological nonsense. Part of the sabermetric orthodoxy is to deny the existence of "clutch skills", or at least to minimize them relative to overall factors that impact every minute of the game. But with the Heat so dominant in blowouts and so vulnerable in close games, perhaps there is something to the old sportswriter aphorisms about certain teams being unable to close the deal when the margin gets tight.
As mentioned earlier, Miami is 5-13 (.278) in games decided by 5 points or fewer, while they sport a sterling 38-7 (.844) mark in games decided by 6 or more points. The Heat now have the biggest differential in NBA history between wpct in games decided by 6+ pts and games decided by 5 or fewer:
|Margin>=6 pts||Margin<=5 pts|
|2001||San Antonio Spurs||53||12||0.815||5||12||0.294||0.521|
|2003||New Jersey Nets||45||23||0.662||4||10||0.286||0.376|
|2007||San Antonio Spurs||50||13||0.794||8||11||0.421||0.373|
|2004||San Antonio Spurs||47||13||0.783||10||12||0.455||0.329|
|1992||San Antonio Spurs||40||21||0.656||7||14||0.333||0.322|
|1972||Los Angeles Lakers||59||6||0.908||10||7||0.588||0.319|
|1996||Portland Trail Blazers||33||17||0.660||11||21||0.344||0.316|
|2001||Portland Trail Blazers||43||20||0.683||7||12||0.368||0.314|
Although APBRmetrics teaches us that all teams' records in close games regress toward .500, at a certain point it becomes statistically unlikely that the Heat's "true" probability of winning those types of games is as good as a coin flip.
Using the binomial distribution, the probability of a true .500 team going 5-13 or worse in a given set of 18 games is just 4.8%. In other words, if our initial hypothesis was that Miami is still a true .500 team who simply suffered a spate of colossally bad luck in close games, we now have enough evidence to reject that hypothesis. The alternate hypothesis -- that the Heat are something less than a true .500 team in close games -- seems far more likely.
How much less, though? Well, using the Empirical Rule, there's a 68% chance that Miami's true "close game skill" falls between a 3-15 record and a 7-11 one, and a 95% chance it falls between 1-17 and 9-9.
Of course, you may want to use Bayes' Theorem to inform Miami's expected skill level with additional information (in which case their .844 record in non-close games would come into play). But if you only look at the evidence from close games this season, it is statistically improbable that the Heat's true ability to come out on top is even .500 when the score is tight.