The NFC and AFC championship weekend has arrived and with it a whole new dynamic for interesting stats to compare. If we go just by season averages on points scored and points given up, Los Angeles has about a 2 point advantage over New Orleans and Kansas City is favored by about 1 point over New England. The predicted scores given in the below table are the mean of the season averages of the opposing teams' offense and defense.
If you've been paying attention to this blog, you know that location can play a huge role in the final score. This week the team side with the most significant home/away effect is the KC defense ((p=0.00004), where they hold opposing teams on average to less than -5.2 points what they should normally score when playing at Arrowhead Stadium. Conversely, the offense with the most significant home/away effect this weekend is New England (p=0.007), which scores about 1.2 points less than what you would expect based on their season averages. I would say this is KC's game to lose, but year after year I've seen Belichick and Brady defy statistics, so I wouldn't be surprised if the Patriots make this a close game.
The game that interests me the most is the New Orleans-Los Angeles contest. Both of these teams are definitely offensively driven, with LA and NO having the second and fourth highest season scoring average respectively in the NFL. The NO defense is slightly better than LA's defense, but only by less than a point based on season averages. NO is also playing at home where their offense scores 5.9 points more than you would expect based on their season average (p=0.08), but their defense actually allows more points at home than they do at away games [this may be a byproduct of a more efficient offense at home that gives the other team more opportunities with the ball]. As I've mentioned in the past, LA is a weird team at home and on the road, with their offense simply out-performing expectations about equally both home and away. [Last year their offense played significantly WORSE at home]. The LA defense also tends to allow fewer points on the road than at home (p=0.15), which could benefit them today playing in New Orleans.
Given above are my predictions for today's games based on season averages and location effects. I also include turnover and sack information into my final model, which gives LA as 2.4 point favorites over NO and KC as 6.9 point favorites over NE. As the great statistician George Box once said, "All models are wrong... but some are useful", so anything can happen today. Bill Belichick and Sean Payton are great coaches with amazing, experienced QBs so we'll see how well this models real life.
As always, I'm struggling to get this out before the first game so please forgive the grammatical errors.
-Michael Edwards, Bioinfo Solutions LLC
P.S. I was accurate on my picks last week even though I didn't have time to post on it. (See below, from Can You Beat FiveThirtyEight’s NFL Forecasts?). My previous weekend was a disaster! The Bears KILLED me!
* Microsoft Excel for Mac 2011 and JMP Pro ver.14 were used to collect, analyze and display NFL data.