Can social media predict NFL game outcomes? To answer that question, this site collected more than 200 random fan predictions from Twitter over the past few days. The aggregate of those crowdsourced numbers predicts that the Patriots will win Sunday’s Super Bowl 30-27.
This is almost exactly in line with Vegas, which posits that New England will win by 3 points, with an over/under of 58.5, 1.5 points higher than the Twitter mean.
61% of posters who included a specific score pick the Patriots to win, against 39% for the Falcons. (Coincidentally, 538.com likewise gives New England a 61% chance of winning, 39% for Atlanta, though this is mixing apples and oranges...)
However, that average (mean) score hardly tells the whole story. There are many other ways to break down the numbers.
The most commonly predicted combined score is Patriots 27, Falcons 24 (9 times), followed by Patriots 34-Falcons 31 (6x), Falcons 31-Patriots 28 (5x), and Falcons 30-Patriots 27 (5x).
The most commonly predicted Atlanta-only score is 24 (37 times) followed by 31 (21 times); while the most commonly predicted Patriots-only scores are 31 and 27 (30 times each) followed closely by 34 (27 times).
The most commonly-predicted margin between the two teams’ scores puts New England by up 3 (27x), followed by Atlanta by 3 (20x), Atlanta by 4 (15x), the Patriots by 10 (14x), and the Patriots by 4 (14x).
My own prediction goes against all of these numbers, anticipating a lower-scoring game than most are expecting—with the Patriots eventually winning 24-20.
Everyone agrees that both offenses are explosive. But given that both of these teams have succeeded mostly with long scoring drives, requiring multiple first downs rather than big, quick bombs, it doesn’t seem there will be enough time for each team to score more than 4-5 times, and not all of those will be touchdown. So if I were a betting man, I would take the under, even if I favored the Falcons.
The big difference-maker in my scenario would be the Patriots defense holding Atlanta to two touchdowns and three field goals, while the Patriots prevail (despite scoring fewer times) by scoring three touchdowns and one field goal. The Patriots’ top-ranked defense has been excellent all year in this regard, keeping opponents “in front of them” all the way downfield, then either getting a turnover, a stop, or forcing the other team to settle for a field goal
Hedges: My score could easily vary somewhat due to a two-point conversion attempt (say, by Atlanta trying to pull within one field goal on its final TD drive) or a missed PAT (perhaps by the Patriots). So 24-21, 24-19, 23-21, 23-20 and 23-19 seem equally plausible to me.
And of course, all it takes is one untimely turnover by either team, a badly-missed tackle in the open field, an unusual special teams play, or a catastrophic injury for either side to skew both the score and outcome significantly.
Which is why I never bet on NFL football.
(Note: A handful of absurd predictions, such as the guy who thinks over 100 points will be scored on Sunday night, or the person who though Atlanta would win by a margin of nearly 40 points, were discarded.)