Thursday Night Football - Prediction and Intro to Prediction Systems

IT'S HERE! IT'S FINALLY HERE!

The NFL season kicks off tonight when the defending champion Baltimore Ravens battle the Denver Broncos on the airy heights of a mile-high stadium.

Baltimore comes into the season riding a wave of optimism despite few experts picking them to repeat as Super Bowl champions. Age and retirements have slowed them down, but Baltimore is still a force to be reckoned with in the NFL.

Denver, on the other hand, seems to be a favorite of many, with Manning having proved that he can still play after his neck injury. The defense remains a bit of a question mark, but the Broncos took steps to improve that side of the ball during the offseason. The Broncos look like a team ready to prove the end of 2012 – an embarrassing, last-second loss to the Ravens - was just a fluke. Since the opening game of the 2013 season pits these two NFL elites against one another, the first 60 minutes of football could have a huge impact on where the AFC championship game is played in late January.

Now, on with the predictions! For every NFL game this season, I will post predictions based on Creative Misfortune's Points-Per-Yard formulas. Two variations of this formula will be used. One based on overall team performance and another based on how well each team does in their home and away games. But what fun would it be to just pick games based on these formulas in a vaccum without having other prediction systems to compare against?

It is very difficult, traditionally, for any prediction system to hit 70% accuracy or more just picking the winners of NFL games - one of the reasons the NFL is such a popular sport. However, Vegas has a pretty good rate picking winners of games (though never quite hits the 70% mark) just based on who their favorites are and even blindly picking the home team results in a better than average success rate of around 57%, but that approach, again, will never produce success over 70% of the time.

To that end, all games will be predicted by the following systems, with the goal of finding a somewhat simple formula that can hit the magical 70% accuracy mark:

First, the Creative Misfortune formula systems -

Creative Misfortune's Overall Points-Per-Yard Pick – These picks are based on a Points-Per-Yard formula I developed that proved to be quite successful in 2009, but fell short of my expectations in 2010. 2011 had modest results, matching the Vegas success rate. One more season of predictions should help determine how accurate (or not) Creative Misfortune's Points-Per-Yard formula is.

Creative Misfortune's Home & Away Points-Per-Yard Pick – A new variation of the Overall Points-Per-Yard formula, the Home & Away Points-Per-Yard formula will take into consideration where the game is being played and how well the two teams involved play "at home" or "on the road". Whereas the Overall PPY formula measures the overall strength of teams, the Home & Away PPY formula will add a second variable - where the game is played - into its calculations to try and increase the accuracy of the Overall PPY formula. This variation of the formula has never been tested over an entire season, so we'll see how it stands up to the pressure of picking games.

Vegas Lines Pick – These picks will be determined by looking at the favorite and the over/under for the game. Basically, if Team A is a 5.5 point favorite and the Over/Under for the game is 46, the Vegas Lines prediction will be based on Team A winning the game by 6 (beating or matching the spread). The Over/Under value is also included with each Vegas Lines pick to give you an idea of how high (or low) scoring Vegas thinks the game will be.


Second, the "blind" control systems -

The blind control predictions are exactly that - blind to the teams involved in the game. They only care where the game is played and the record of the teams involved. No fancy stats. No gut feelings. Just blind picks, even if it means picking Kansas City to beat San Francisco.

Best Record Wins, Else Home Team Wins – This is the so-called "Isaacson-Tarbell Postulate", proposed by ESPN's Tuesday Morning Quarterback (TMQ) readers Eric Isaacson of Indiana University and Catey Tarbell of Kirkland, Washington. You can view a discussion of this system in TMQ's archived column for February 9, 2010 under the heading "Weekly Game Predictions". This system basically predicts that the team with the best record will win the game. In case both teams have identical records, the Home Team will be predicted to win. Because all the predictions on The Art of Creative Misfortune will have point spreads associated with them, the NFL’s default score of 20-17 will be used for any Best Record Wins predictions. Teams picked to win by the Best Record Wins system will be expected to win by 3 points.

Home Team Wins - ***The True Control Value*** - Let’s face it. Any prediction system that loses out to blindly predicting that the Home Team will win, regardless of the matchup, is not a good prediction system. As such, this will be used as a control value to blindly pick all games. Typically, NFL teams win at home about 57% of the time. All the other predictions systems featured on The Art of Creative Misfortune should be able to beat the Home Team Wins system. As with the Best Record Wins system, all Home Team Wins predictions will predict based on the NFL's default score of 20-17, i.e. Home Team wins by 3.

Now, let's pick tonight's matchup...

Baltimore Ravens at Denver Broncos
     Overall Points Per Yard Pick: Denver Broncos by 9
     Home/Away Points Per Yard Pick: Denver Broncos by 18
     Vegas Lines Pick: Denver Broncos by 9. Broncos by 9, Over/Under 48
     Best Record Wins Pick: Denver Broncos by 3. Both teams at 0-0, home team wins
     Home Team Wins Pick: Denver Broncos by 3

Tomorrow, predictions for the Sunday games slated for Week 1 NFL action...