Winner of Game 5 in 2-2 NBA Series

How important is Game 5 in a 2-2 series? Winner of the game advances 82 percent of the time in NBA: http://t.co/NASoXmgSbI

— RealGM (@RealGM) May 13, 2015

The above stat has been brought up a lot in this year's NBA Playoffs, since 3 of the 4 current series were tied 2-2. Per the above Real GM article (and every ESPN infographic on Sportscenter for the past week), "In NBA history, the winner of Game 5 is 147-33 to go on to win the entire series."



But how likely is it that the winner of Game 5 wins the series 82% of the time? The team up 3-2 has to lose 2 in a row to lose the series. As before, I'm using the Log5's from this year's three 2-2 series to approximate the average chance the trailing team has to come back and win 4-3:

3	2	Game 6	Game 7	Prob Win Series
GS	MEM	38.77%	21.61%	8.38%
ATL	WSH	40.92%	23.28%	9.53%
CLE	CHI	51.12%	31.85%	16.28%
				11.40%

So in fact (based on this year's predictions), it's actually even more unlikely that a team comes back in win 4-3. The most "balanced" series, CLE-CHI, is still short of the 18% threshold that we've seen in the 180 instances that have occurred so far (and by the time that I wrote this, CLE already finished the series and won 4-2). The z-score for the difference between the actual 18% comeback rate and the predicted 11% is 2.40, which results in a p-value of 0.992.

So why have teams historically won at a higher clip? Firstly, the three series from this year are likely not representative of all of the 7-game NBA series that were tied 2-2. Secondly, it raises a different question: has competitive balance decreased in recent years? I.E. Were NBA teams more balanced historically? That's a question for another time.