Bing Correctly Predicts Super Bowl Matchup, Still Favors Patriots to Win
In a wild NFL Conference Championship weekend (well, half of it was wild), Bing Predicts accurately picked both winners. Now Bing has a tough call to make in the Patriots-Seahawks showdown that will take place when Super Bowl XLIX kicks off on February 1st.
Bing Predicts Beats the Odds, Sort Of
Headed into the Conference games, the odds were against Bing—sort of. With predicted chances for both games under 60%, things looked a bit hazy. The search engine performed poorly all season under these conditions, going 46-45 in such games. Add in the fact that Bing changed its mind about the Packers-Seahawks game midway through the week, and you had a recipe for disaster (Bing was a horrendous 7-12 when switching its pick during the week).
However, Bing pulled it out against the odds; based on the regular season data, the search engine had just a 21% chance of getting both games correct. Looking at the Vegas odds, Bing's chances were actually pretty good, but that doesn't make for an exciting story.
The improbable for Bing was made possible by the even more improbable comeback by the Seahawks. Seattle proved that, as long as you keep it relatively close, you only have to outplay your opponent for 4 minutes in order to win.
In reality, Seattle's win was more about Green Bay not capitalizing on opportunities. When you only score 6 points off 5 turnovers, you aren't a championship team. We can cast all the blame we want on Brandon Bostick for not recovering the onside kick that would have clinched the game for the Packers, but what about all the other failures of the Packers? This is a game that could have easily been 21-0 at the end of the 1st quarter. The Packers could have put this out of reach early, rendering any number of onside kicks irrelevant.
On the other side of the league, things were pretty unexciting. The Patriots 45-7 win over the Colts made Bing's 56.6% prediction of a Patriots win look like a joke. Did the Colts have any chance to win this game? Even the news of a potential "Deflate-Gate" can't save this game from being the boring mismatch that it was.
Bing's 2-0 record this week puts the search engine at 8-2 so far for the playoffs. Not bad at all, but there hasn't been much in the way of the unexpected either. Bing's getting it right because it's picking the teams that are supposed to win. Picking against the odds just isn't in Bing's blood. The search engine makes its predictions based on the available data, so it can't really go with the underdog.
Still, the experiment has been rather successful, even if Bing hasn't been more impressive than any other predictive model. This is about more than saying which team is going to win. It's about saying how likely a team is to win. Looking at the data across the entire season, it's hard to argue that Bing's predicted chances are wrong. I'll get deeper into this after the Super Bowl when I wrap up the season for Bing.
Bing Predicts Says Patriots Will Win It All…Or Maybe the Seahawks Will
As of Tuesday morning, Bing has the Patriots winning with a 51.5% chance. This is identical to the percentage the search engine assigned a month ago when the playoffs started (and it thought the Patriots would be defeating the Packers in the big game). It's a bit strange that none of the events that have transpired in the past three weeks have caused Bing to rethink this percentage.
With the Super Bowl still two weeks away, it's possible that Bing will make an adjustment, but those hoping for a Seahawks victory won't likely see Bing swing in their favor. If anything, the percentage will grow more to the advantage of the Patriots. After all, New England looked dominant in the AFC Championship game while the Seahawks looked vulnerable and desperate. Still, anything could happen. Remember last year? Did anyone predict a blowout of that magnitude?
Championship football—American style, that is—hasn't treated Bing kindly. The search engine predicted Oregon to defeat Ohio State in the NCAA championship game. Contrary to Bing's pregame pick, the Buckeyes rolled over the Ducks like there was never a chance for anything else to happen. On the other side of the coin, Bing did accurately predict all 15 games of the World Cup knockout stage (does anyone really care about that third place game?).
Looking at Bing's results for the entire season, the chances of a Patriots win are about 50-50. Here's a quick look at the data:
- Bing was 46-45 in games with a predicted chance under 60%
- Bing was 11-3 when predicting the Patriots to win
- Bing was 1-2 when predicting the Seahawks to lose
- Bing was 3-8 when the predicted chance was exactly 51.5%
Of course, none of this data has anything to do with the game itself. On February 1st, the Seahawks and Patriots will play a game that's independent of any other data available. That's what makes predicting the outcome of sporting events based solely on data so difficult.
Personally, I think it's hard to pick against the Patriots, but I won't count the Seahawks out unless they're down by 20 points with 2 minutes left. Will Bing pick the Super Bowl winner and end the playoffs with an impressive 9-2 record? Although the odds aren't necessarily in Bing's favor, my gut says that's precisely what will happen.
If Bing is right, Super Bowl XLIX will be a thriller. On Friday morning, the search engine expanded its prediction model to include the predicted final score:
Sounds like one heck of a game. Let's just hope all the balls are properly inflated.