Might as well finish off the series

For those who haven't been following along, on Thursday I jumped the gun on a blog post, and found that an assumption about what I would find if I changed a study completed by Steven Levitt was absolutely incorrect. On Friday, I finished my analysis and figured out (to my satisfaction) why the results came through like they did. Today, I'm going to take a look at the NBA and see what happens if you take a look at it and compare that to the randomly paired poker pro results .

I have used the playoffs from 2007-2010 as a base for the basketball population test. Here's how the percentages break down:

Champion (1 team): 71.95% win percentage vs. the rest of the league, 81.60% vs non playoff teams
Made the finals (2 teams): 72.19% win percentage vs. teams that did not make the finals, 79.32% vs. non playoff teams, overall win percentage 71.65%
Made the Conference finals (4 teams): 67.28% win percentage vs. teams that did not make the finals, 73.79% vs. non playoff teams, overall win percentage 65.63%
Made the second round (8 teams): 64.91% vs. teasm that did not make the second round of the playoffs, 70.16% vs. non playoff teams, overall win percentage 61.43%
Made the playoffs (16 teams): 65.84% vs. non playoff teams, overall win percentage 57.70%

It's worth noting that there are 30 teams in the NBA, which means that 14 teams do not make the playoffs.

If anything, this destroys Levitt's main point, which is the random pairings he made of "elite players" vs. non-elite players show about the same win percentage of baseball teams when doing a playoff vs. non playoff cut. Basketball actually has much larger spread of win percentages, with teams having individual win percentage ranging from 10% to 75%+, which indicates a greater spread in talent between the teams. This would also match better with World Series of Poker fields, which have a large spread in talent.

What would be ideal would be to know more abou what Levitt's population did. How often do elite players cash vs. non-elite players? If you randomly pair an elite and non-elite player, how often do both cash, how often does only once cash, and how often do neither cash? Just throwing out the random pairing percentage doesn't give us a proper idea of it's significance.

All of this isn't to say that poker is not a game of skill; it is. The question is how much of it is skill, and how much is also random occurance of the cards.