I recently read an interesting analysis of the NFL that basically established  NFL games were determined 50% of the time by skill and 50% of the time by luck.   The way that statistician came to the conclusion was somewhat just comparing graphs and while that works in showing how females hemlines go up and down with the economic well being of a country… it also may just be two graphs that look alike…  kind of like how female hemlines go up and down with the economic well being of a country.  Still,  the author is the statistician and I’m just a guy who struggled in stats, so I’m the wrong one to question the data and the analysis, but most of it sounds right even if I’m little skeptical of the methodology.  Which I have no right to be (as we’ve established).

Here’s some highlights…  the best any picker of games should be is a little over 75% accurate.  If you can do better you should be sports betting as a career.   The fact that the best computer models are near that number again, implies, there may be some truth behind it his analysis or some coincidence.   How doe 75% prove 50%?  Well if luck is a factor half the time the favored team would benefit from it just as the underdog would.  Thus, in the 50% of the games not determined by skill  half of the 50% the favorite benefits from it and half the time the dog does.   ( He hasn’t done the study to see if fortune favors the bold yet).  Therefore you get the 75% number, 50% of the time skill prevails, 25% of the time the better skilled team gets the bounces and the other 25% team the less skilled team does…  50%+25%= 75/25.  For his far better descriptions go here.

What is interesting is if luck determined the outcomes of games solely, there would be very few 0-16 teams and very few 16-0 teams.  Parity reigning supreme could easily confuse people into believing a team is good rather than lucky, because gaudy records and long winning streaks fit into the data even if there is little separation between the teams.

This is interesting from a poker perspective in evaluating how good a tournament player you are.  It’s possible to run good for a career, with some hiccups obviously, and simply be an outlier.  It’s also possible to run bad for a career, with some modest successes.  There are better graphs that prove just that as related to poker, however, this NFL analysis is useful because it demonstrates how easily it is to misperceive skill and luck.

Even a favored team needs luck, happenstance, strange bounces and twists of fate to win half the time.  Hell, they probably need it to counter the other team’s healthy dose of it another 25% of the time, so that they just break even… and then skill can win out.  As applied to poker this elucidates the maxim you are never as good you think you are when you are running good, and you are never as bad as you think you are when you are running bad.  More to come next post…