Notoriously, the brilliant political forecaster Nate Silver blundered in picking the Seahawks and the Patriots as the teams to face off in Super Bowl XLVII.
That drew some headlines, coming close on the heels of Silver's 100 percent record calling states in the Presidential election. In the welter of post-game analysis -- including chatter about the black-out and Beyoncé singing live -- it's easy to miss the fact that analytics-based predictions of the final victor struck out across the board.
Silver couldn't resist another bout of number crunching. On January 30, he analyzed data from the Pro Football Reference Simple Rating System, and came up with a win for the '49ers, saying that a Ravens victory would be an "upset."
Meanwhile, Prediction Machine brought the big guns to bear, running some 50 thousand simulations of the game in advance of kick-off. It found the 49ers winning about 67 percent of the time. The "Predictalator" -- the "most advanced sports forecasting software available" -- had more detailed predictions to offer:
The 49ers will rush for 167 yards thanks in part to running back Frank Gore's 89 yards on 19 carries and quarterback Colin Kaepernick's 38 yards on six carries. The Ravens will be held to only 98 yards rushing...
Not bad. Gore actually rushed for 110 yards on 19 carries, and Kaepernick 62 yards on seven. Which goes to show how little individual statistics can matter. Baltimore rushed only 93 yards in total -- but that doesn't include, for example, Jacoby's 108 yard kick-off return for a touchdown at the beginning of the second half.
The problem, of course, is that for all the bandying of the term "big-data," the information on which analysis of a single football game is based is really quite limited. Sports fans may seem to be drowning in statistics, but the data on a few dozen players, and how they've performed as a team over the last few months, does not rival in scale the data now available to the enterprise.
To take one, dramatic example, Walmart topped 7 billion transactions in 2008, surpassing the figure for the world's total population. Every such transaction generates data, as does web traffic -- topping 40 million unique visitors to Walmart websites per month in 2012. Not to mention the global conversation about Walmart products on social media platforms.
While it's true that most enterprises are dealing with much smaller data pools than a retail giant like Walmart, it's also the case that the data available is more plentiful, and richer, than the data available to even the geekiest sports fans.
This is good news all round: Enterprises need to be able to drive analytics against big-data to generate sound decision-making. But football fans don't really want to know which team is going to win a month before the game.
Unless they have money on it, of course.
-- Kim Davis , Community Editor, Internet Evolution