How Much 2-3-2 Home Court Affects Win Probability

Up until 2014, the NBA Finals had been a 2-3-2 format, where the lower seed hosted Games 3-5 (as opposed to the rest of the series which went 2-2-1-1-1, where the lower seed hosted Games 3, 4, and 6). In theory, if the higher seed lost Game 1 or 2, they could lose the series without being able to utilize the rest of their home court advantage.

How much does this really penalize the higher seed? Home court isn't flipping, just the ordering of home court.

I previously simulated this year's playoffs here. Here's how the new odds would look if every series changed to 2-3-2:

Team	Second Round	Conf Finals	NBA Finals	Champion
Western Conf
Golden State	87.95%	54.43%	38.70%	16.28%
LA Clippers	12.05%	2.18%	0.65%	0.07%
Houston	48.78%	20.38%	11.81%	3.35%
Utah	51.22%	23.01%	14.02%	4.71%
Portland	59.32%	34.80%	13.27%	3.49%
Oklahoma City	40.68%	19.13%	5.99%	1.17%
Denver	69.86%	36.00%	13.46%	3.63%
San Antonio	30.14%	10.07%	2.10%	0.31%
Eastern Conf
Milwaukee	96.20%	80.19%	62.56%	48.62%
Detroit	3.80%	0.64%	0.10%	0.02%
Boston	60.45%	12.79%	6.19%	2.50%
Indiana	39.55%	6.38%	2.48%	0.82%
Philadelphia	71.64%	22.76%	3.91%	0.95%
Brooklyn	28.36%	4.11%	0.30%	0.02%
Toronto	86.19%	67.50%	23.92%	13.98%
Orlando	13.81%	5.63%	0.54%	0.08%

The differences are fairly marginal, improving most underdogs in the first round by less than 1%:

Western Conf	Pyth	Actual	Alt HFA	Dif
Golden State	0.719	87.42%	87.95%	0.53%
Denver	0.647	70.08%	69.86%	-0.22%
Portland	0.649	57.81%	59.32%	1.51%
Houston	0.670	48.09%	48.78%	0.69%
Utah	0.687	51.91%	51.22%	-0.69%
Oklahoma City	0.622	42.19%	40.68%	-1.51%
San Antonio	0.563	29.92%	30.14%	0.22%
LA Clippers	0.529	12.58%	12.05%	-0.53%
Eastern Conf
Milwaukee	0.782	96.44%	96.20%	-0.24%
Toronto	0.709	86.72%	86.19%	-0.53%
Philadelphia	0.597	72.47%	71.64%	-0.83%
Boston	0.659	60.56%	60.45%	-0.11%
Indiana	0.625	39.44%	39.55%	0.11%
Brooklyn	0.500	27.53%	28.36%	0.83%
Orlando	0.527	13.28%	13.81%	0.53%
Detroit	0.488	3.56%	3.80%	0.24%

Golden State and Portland actually go up slightly, and I reran the sims a couple of times and got that result every time. So it doesn't appear to be randomness in the sim results.

So let's look at one series, Philadelphia/Brooklyn (the series in which the underdog stands to gain the most), and see where the changes occur:

	Team Wins	in4Rd1	in5Rd1	in6Rd1	in7Rd1
Actual	Philadelphia	12.24%	22.53%	18.44%	19.26%
	Brooklyn	2.43%	5.17%	10.43%	9.50%
Alt HFA	Philadelphia	12.53%	17.80%	22.95%	18.36%
	Brooklyn	2.45%	7.57%	8.99%	9.35%
Dif	Philadelphia	0.29%	-4.73%	4.51%	-0.90%
	Brooklyn	0.02%	2.40%	-1.44%	-0.15%
	Cumul	0.31%	-2.33%	3.07%	-1.05%

Clearly the differences in either team sweeping are due to randomness: in both formats, each team gets their first 2 games at home.

The obvious changes show in Games 5-6: both teams are more likely to clinch on their home floor when that home floor is flipped. But the real change that helps Brooklyn, the underdog, is in the decreased probability of the series going 7: in a shorter run scenario there's higher variance, which increases the underdog's probability.

What about the projected most competitive series, where the home team is actually the underdog?

	Team Wins	in4Rd1	in5Rd1	in6Rd1	in7Rd1
Actual	Houston	5.49%	12.89%	12.70%	17.01%
	Utah	6.73%	11.98%	19.04%	14.16%
Alt HFA	Houston	5.15%	9.22%	16.09%	18.32%
	Utah	7.56%	15.44%	14.33%	13.89%
Dif	Houston	-0.34%	-3.67%	3.39%	1.31%
	Utah	0.83%	3.46%	-4.71%	-0.27%
	Cumul	0.49%	-0.21%	-1.32%	1.04%

This time, the change that helps Houston, the home team, is the INCREASED probability of the series going 7: since they have home court, they then stand a better chance of winning that last game and clinching the series.

So the underdog does appear to be aided by changing the series order, but by less than 1%. The better team still would win more often than not.