ESPN has the most broken NBA analytics model of all-time

It seems unlikely ESPN’s NBA analytics formula has ever watched basketball before.

While it is certainly surprising that the Heat are beating the Celtics so summarily that Magic Johnson is flat-out saying Boston has “quit,” the idea that they could win the series was not nearly as unlikely as ESPN’s analytics forecasted.

Before the series, ESPN said the Heat had just a three percent chance of winning the series; gambling odds had the Heat at +400 and the Celtics at -550, which would mean the market saw Miami as having about a 19 percent chance to pull off the upset.

Even after Miami jumped out to a 2-0 lead, with both wins coming in Boston, ESPN’s analytics claimed the Celtics had a 65 percent chance to win the series, a staggering number that defied logic.

This was severely at odds with the gambling odds, where the Heat were -180 and Celtics +150, meaning the market had the Celtics with about a 39 percent chance to come back.

Now that the Heat are up 3-0, ESPN is still giving the Celtics a comically large chance to pull off the unprecedented feat of overcoming this deficit in an NBA playoff series (it has never been done in a best-of-seven series in league history).

According to ESPN’s BPI on Monday morning, the Celtics have a 31 percent chance to do what no team in the NBA has ever done.

Perhaps one day an NBA franchise will pull off a stunning turnaround like the Red Sox did to the Yankees in the 2004 ALCS, but the odds are a lot longer than about three in ten.

The gambling odds cited in ESPN’s graphic would give the Celtics about a 12 percent chance to come back, and it’s probably even longer than that given more bettors are likely to take a flier at a big seven-to-one multiple on their bet than lay $12 to win $1.