by A.J. Coltrane
Ken Pomeroy made some great points in February of 2012 (subscriber only):
…It’s the manufactured stories that attempt to explain the often-unexplainable variability in a team’s performance that I take issue with. Some team salvages its season by going on a late winning-streak and the origins of the streak are explained by a players-only meeting or the team captain stepping up and being a leader, or a renewed emphasis on defense, etc. When in reality, the causes of the change may have been more complicated that anyone could truly understand. (Naturally, this xkcd comic comes to mind.)
Murray State’s loss last week provided one of the clearest such examples of this method of analysis. The general assumption after the loss was that the Racers cracked under the pressure [(1), (2), (3)] of their unbeaten record. Even the coach said so! The thing is, Murray never reached a point during the season where they were better than a 50% proposition to go unbeaten in conference. You play enough games in which you are heavily favored, and you are going to lose eventually. Put more precisely, a team that plays ten games as a 90% favorite is expected to lose once during that span, and the Racers have played a lot of such games this season, including the game against Tennessee State…
and
…There’s lots of unexplained variance in a college basketball game. The Vegas line has long been proven to be the best predictor of outcomes, and while it has the reputation among some of being scarily accurate, the average error in the Vegas line is 8.4 points. And, with all due respect to other prognosticators out there, that’s the best we can do.
Keep in mind that 8.4 points of unpredictable variability is the combination of the variability of the two teams involved in a game. When Duke unexpectedly won at North Carolina last week, was it because Duke played better than usual and UNC played worse than usual, or was it because UNC played better than usual and Duke played much, much better than normal? Or was it because both teams played worse than usual, but UNC just really played badly? I think it’s nearly impossible to disentangle the two. If one team shoots poorer than expected, is it because their form was off or because the defense was better than usual? It is difficult to determine the answers to questions like this without some serious video breakdown…
I love the comic that Pomeroy linked, it’s exactly why I don’t listen much to the talking heads:
FlatbreadGarden 